diff --git a/1-first-project/Abgabe.ipynb b/1-first-project/Abgabe.ipynb
index 2fefb0c..bdb5960 100644
--- a/1-first-project/Abgabe.ipynb
+++ b/1-first-project/Abgabe.ipynb
@@ -3,7 +3,7 @@
   {
    "cell_type": "code",
    "execution_count": 1,
-   "id": "920f21b6",
+   "id": "f9b2f6c2",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -15,7 +15,7 @@
   {
    "cell_type": "code",
    "execution_count": 2,
-   "id": "9494dff5",
+   "id": "eb49db4b",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -44,7 +44,7 @@
   {
    "cell_type": "code",
    "execution_count": 3,
-   "id": "ac87093f",
+   "id": "daefd4a8",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -57,7 +57,7 @@
   {
    "cell_type": "code",
    "execution_count": 4,
-   "id": "ff9eeb53",
+   "id": "3b04c1ee",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -74,7 +74,7 @@
   {
    "cell_type": "code",
    "execution_count": 5,
-   "id": "ec8ff24a",
+   "id": "5cf901e4",
    "metadata": {},
    "outputs": [
     {
@@ -106,7 +106,7 @@
   {
    "cell_type": "code",
    "execution_count": 6,
-   "id": "18326042",
+   "id": "a68cb0bb",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -162,7 +162,7 @@
   {
    "cell_type": "code",
    "execution_count": 7,
-   "id": "8252cda5",
+   "id": "ae002a37",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -178,7 +178,7 @@
   {
    "cell_type": "code",
    "execution_count": 8,
-   "id": "330b79aa",
+   "id": "0d0b3544",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -208,7 +208,7 @@
   {
    "cell_type": "code",
    "execution_count": 9,
-   "id": "7989f97f",
+   "id": "d371d6e9",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -227,7 +227,7 @@
   {
    "cell_type": "code",
    "execution_count": 10,
-   "id": "06926b00",
+   "id": "1c14b2a1",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -245,7 +245,7 @@
   {
    "cell_type": "code",
    "execution_count": 11,
-   "id": "56a8c615",
+   "id": "189de319",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -275,9 +275,24 @@
   {
    "cell_type": "code",
    "execution_count": 12,
-   "id": "0a347c17",
+   "id": "a796b9b2",
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Preprocessing...\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "100%|██████████| 26179/26179 [01:30<00:00, 290.22it/s]\n"
+     ]
+    }
+   ],
    "source": [
     "def preproc(d):\n",
     "    flist = {}  \n",
@@ -317,9 +332,17 @@
   {
    "cell_type": "code",
    "execution_count": 13,
-   "id": "aadf64f9",
+   "id": "d3e56332",
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Truncating...\n"
+     ]
+    }
+   ],
    "source": [
     "def throw(pdata):\n",
     "    llist = pd.Series([len(x['data']) for x in pdata])\n",
@@ -339,9 +362,31 @@
   {
    "cell_type": "code",
    "execution_count": 14,
-   "id": "e7bfb918",
+   "id": "dabc3af0",
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      " 19%|█▉        | 3723/19640 [00:00<00:00, 18633.70it/s]"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Padding...\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "100%|██████████| 19640/19640 [00:01<00:00, 18655.32it/s]\n"
+     ]
+    }
+   ],
    "source": [
     "from tensorflow.keras.preprocessing.sequence import pad_sequences\n",
     "# ltpdata = []\n",
@@ -359,7 +404,7 @@
   {
    "cell_type": "code",
    "execution_count": 15,
-   "id": "7ea5ec4d",
+   "id": "17fece5a",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -372,19 +417,13 @@
     "    model = Sequential()\n",
     "    ncount = train_shape[0]*train_shape[1]\n",
     "    \n",
-    "#     model.add(Conv2D(64, (5, 5), input_shape=train_shape, activation='relu', padding='same'))\n",
-    "#     model.add(MaxPooling2D(pool_size=(2, 2)))\n",
-    "    \n",
-    "#     model.add(Conv2D(64, (5, 5), activation='relu', padding='same'))\n",
-    "#     model.add(MaxPooling2D(pool_size=(2, 2)))\n",
-    "    \n",
     "    model.add(Flatten(input_shape=train_shape))\n",
     "    \n",
     "    model.add(BatchNormalization())\n",
     "    \n",
     "    model.add(Dropout(0.1))\n",
     "    \n",
-    "    for i in range(2,5):\n",
+    "    for i in range(1,5):\n",
     "        model.add(Dense(int(ncount/i), activation='relu'))\n",
     "        model.add(Dropout(0.1))\n",
     "    \n",
@@ -401,8 +440,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 16,
-   "id": "b62d2f11",
+   "execution_count": 24,
+   "id": "1ef39498",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -421,7 +460,7 @@
     "        save_best_only=True\n",
     "    )\n",
     "    \n",
-    "    model.fit(X_train, y_train, \n",
+    "    history = model.fit(X_train, y_train, \n",
     "              epochs=30,\n",
     "              batch_size=256,\n",
     "              shuffle=True,\n",
@@ -432,13 +471,14 @@
     "    \n",
     "    print(\"Evaluate on test data\")\n",
     "    results = model.evaluate(X_test, y_test, batch_size=128, verbose=0)\n",
-    "    print(\"test loss, test acc:\", results)"
+    "    print(\"test loss, test acc:\", results)\n",
+    "    return model, history"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 17,
-   "id": "8c03d2a3",
+   "execution_count": 25,
+   "id": "160ec98a",
    "metadata": {
     "tags": []
    },
@@ -450,8 +490,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 23,
-   "id": "5c9f56eb",
+   "execution_count": 26,
+   "id": "a4799ab9",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -475,8 +515,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 24,
-   "id": "c86a5870",
+   "execution_count": 27,
+   "id": "e73dcbbb",
    "metadata": {
     "tags": []
    },
@@ -486,811 +526,169 @@
      "output_type": "stream",
      "text": [
       "Training...\n",
-      "Model: \"sequential\"\n",
+      "Model: \"sequential_1\"\n",
       "_________________________________________________________________\n",
       "Layer (type)                 Output Shape              Param #   \n",
       "=================================================================\n",
-      "conv2d (Conv2D)              (None, 14, 75, 64)        1664      \n",
+      "flatten_1 (Flatten)          (None, 1050)              0         \n",
       "_________________________________________________________________\n",
-      "max_pooling2d (MaxPooling2D) (None, 7, 37, 64)         0         \n",
+      "batch_normalization_1 (Batch (None, 1050)              4200      \n",
       "_________________________________________________________________\n",
-      "conv2d_1 (Conv2D)            (None, 7, 37, 64)         102464    \n",
+      "dropout_5 (Dropout)          (None, 1050)              0         \n",
       "_________________________________________________________________\n",
-      "max_pooling2d_1 (MaxPooling2 (None, 3, 18, 64)         0         \n",
+      "dense_5 (Dense)              (None, 1050)              1103550   \n",
       "_________________________________________________________________\n",
-      "flatten (Flatten)            (None, 3456)              0         \n",
+      "dropout_6 (Dropout)          (None, 1050)              0         \n",
       "_________________________________________________________________\n",
-      "batch_normalization (BatchNo (None, 3456)              13824     \n",
+      "dense_6 (Dense)              (None, 525)               551775    \n",
       "_________________________________________________________________\n",
-      "dropout (Dropout)            (None, 3456)              0         \n",
+      "dropout_7 (Dropout)          (None, 525)               0         \n",
       "_________________________________________________________________\n",
-      "dense (Dense)                (None, 525)               1814925   \n",
+      "dense_7 (Dense)              (None, 350)               184100    \n",
       "_________________________________________________________________\n",
-      "dropout_1 (Dropout)          (None, 525)               0         \n",
+      "dropout_8 (Dropout)          (None, 350)               0         \n",
       "_________________________________________________________________\n",
-      "dense_1 (Dense)              (None, 350)               184100    \n",
+      "dense_8 (Dense)              (None, 262)               91962     \n",
       "_________________________________________________________________\n",
-      "dropout_2 (Dropout)          (None, 350)               0         \n",
+      "dropout_9 (Dropout)          (None, 262)               0         \n",
       "_________________________________________________________________\n",
-      "dense_2 (Dense)              (None, 262)               91962     \n",
-      "_________________________________________________________________\n",
-      "dropout_3 (Dropout)          (None, 262)               0         \n",
-      "_________________________________________________________________\n",
-      "dense_3 (Dense)              (None, 52)                13676     \n",
+      "dense_9 (Dense)              (None, 52)                13676     \n",
       "=================================================================\n",
-      "Total params: 2,222,615\n",
-      "Trainable params: 2,215,703\n",
-      "Non-trainable params: 6,912\n",
+      "Total params: 1,949,263\n",
+      "Trainable params: 1,947,163\n",
+      "Non-trainable params: 2,100\n",
       "_________________________________________________________________\n",
       "Epoch 1/30\n",
-      "62/62 [==============================] - 4s 14ms/step - loss: 3.4102 - acc: 0.1025 - val_loss: 3.8896 - val_acc: 0.0438\n",
+      "62/62 [==============================] - 1s 6ms/step - loss: 3.3481 - acc: 0.1160 - val_loss: 3.5396 - val_acc: 0.0687\n",
       "Epoch 2/30\n",
-      "62/62 [==============================] - 1s 10ms/step - loss: 2.7475 - acc: 0.2131 - val_loss: 3.7641 - val_acc: 0.0461\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 2.4941 - acc: 0.2746 - val_loss: 3.1564 - val_acc: 0.1263\n",
       "Epoch 3/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 2.2471 - acc: 0.3301 - val_loss: 3.5046 - val_acc: 0.1347\n",
+      "62/62 [==============================] - 0s 5ms/step - loss: 1.9611 - acc: 0.3980 - val_loss: 3.0374 - val_acc: 0.1533\n",
       "Epoch 4/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 1.8689 - acc: 0.4238 - val_loss: 3.2023 - val_acc: 0.2352\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 1.6416 - acc: 0.4826 - val_loss: 2.7437 - val_acc: 0.2085\n",
       "Epoch 5/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 1.6068 - acc: 0.4923 - val_loss: 3.8644 - val_acc: 0.0558\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 1.4033 - acc: 0.5439 - val_loss: 2.4287 - val_acc: 0.2632\n",
       "Epoch 6/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 1.3984 - acc: 0.5521 - val_loss: 2.1733 - val_acc: 0.4010\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 1.2683 - acc: 0.5797 - val_loss: 2.1105 - val_acc: 0.3564\n",
       "Epoch 7/30\n",
-      "62/62 [==============================] - 1s 10ms/step - loss: 1.2403 - acc: 0.5896 - val_loss: 1.9064 - val_acc: 0.4376\n",
+      "62/62 [==============================] - 0s 5ms/step - loss: 1.1270 - acc: 0.6207 - val_loss: 1.8558 - val_acc: 0.4155\n",
       "Epoch 8/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 1.1112 - acc: 0.6230 - val_loss: 1.8146 - val_acc: 0.4743\n",
+      "62/62 [==============================] - 0s 5ms/step - loss: 1.0280 - acc: 0.6520 - val_loss: 1.6051 - val_acc: 0.4776\n",
       "Epoch 9/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 1.0028 - acc: 0.6547 - val_loss: 2.0000 - val_acc: 0.4236\n",
+      "62/62 [==============================] - 0s 5ms/step - loss: 0.9315 - acc: 0.6812 - val_loss: 1.3901 - val_acc: 0.5489\n",
       "Epoch 10/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 0.9017 - acc: 0.6884 - val_loss: 2.1602 - val_acc: 0.5038\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.8726 - acc: 0.6988 - val_loss: 1.2578 - val_acc: 0.5939\n",
       "Epoch 11/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 0.8505 - acc: 0.7011 - val_loss: 2.1521 - val_acc: 0.5624\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.7879 - acc: 0.7230 - val_loss: 1.1692 - val_acc: 0.6191\n",
       "Epoch 12/30\n",
-      "62/62 [==============================] - 1s 10ms/step - loss: 0.7718 - acc: 0.7303 - val_loss: 2.2699 - val_acc: 0.5736\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.7392 - acc: 0.7379 - val_loss: 1.1623 - val_acc: 0.6283\n",
       "Epoch 13/30\n",
-      "62/62 [==============================] - 1s 10ms/step - loss: 0.7100 - acc: 0.7485 - val_loss: 1.8627 - val_acc: 0.5550\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.6912 - acc: 0.7543 - val_loss: 1.1486 - val_acc: 0.6359\n",
       "Epoch 14/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 0.6737 - acc: 0.7576 - val_loss: 1.9876 - val_acc: 0.5636\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.6471 - acc: 0.7709 - val_loss: 1.1279 - val_acc: 0.6586\n",
       "Epoch 15/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 0.6239 - acc: 0.7723 - val_loss: 2.1203 - val_acc: 0.5540\n",
+      "62/62 [==============================] - 0s 5ms/step - loss: 0.5918 - acc: 0.7853 - val_loss: 1.1477 - val_acc: 0.6469\n",
       "Epoch 16/30\n",
-      "62/62 [==============================] - 1s 10ms/step - loss: 0.5808 - acc: 0.7914 - val_loss: 8.2953 - val_acc: 0.3977\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.5488 - acc: 0.8007 - val_loss: 1.2157 - val_acc: 0.6477\n",
       "Epoch 17/30\n",
-      "62/62 [==============================] - 1s 10ms/step - loss: 0.5578 - acc: 0.7981 - val_loss: 2.3404 - val_acc: 0.5339\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.5421 - acc: 0.8056 - val_loss: 1.1407 - val_acc: 0.6647\n",
       "Epoch 18/30\n",
-      "62/62 [==============================] - 1s 10ms/step - loss: 0.5090 - acc: 0.8137 - val_loss: 1.7944 - val_acc: 0.5937\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.5035 - acc: 0.8180 - val_loss: 1.1731 - val_acc: 0.6617\n",
       "Epoch 19/30\n",
-      "62/62 [==============================] - 1s 10ms/step - loss: 0.4675 - acc: 0.8289 - val_loss: 2.0554 - val_acc: 0.5866\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.4780 - acc: 0.8278 - val_loss: 1.2031 - val_acc: 0.6550\n",
       "Epoch 20/30\n",
-      "62/62 [==============================] - 1s 10ms/step - loss: 0.4484 - acc: 0.8343 - val_loss: 1.8284 - val_acc: 0.5832\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.4620 - acc: 0.8346 - val_loss: 1.1839 - val_acc: 0.6642\n",
       "Epoch 21/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 0.4210 - acc: 0.8490 - val_loss: 2.1521 - val_acc: 0.6219\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.4153 - acc: 0.8489 - val_loss: 1.2167 - val_acc: 0.6606\n",
       "Epoch 22/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 0.4036 - acc: 0.8523 - val_loss: 1.9749 - val_acc: 0.6477\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.4120 - acc: 0.8494 - val_loss: 1.1883 - val_acc: 0.6678\n",
       "Epoch 23/30\n",
-      "62/62 [==============================] - 1s 10ms/step - loss: 0.3790 - acc: 0.8581 - val_loss: 2.8492 - val_acc: 0.6143\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.3817 - acc: 0.8624 - val_loss: 1.2221 - val_acc: 0.6673\n",
       "Epoch 24/30\n",
-      "62/62 [==============================] - 1s 10ms/step - loss: 0.3494 - acc: 0.8700 - val_loss: 10.8548 - val_acc: 0.5950\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.3635 - acc: 0.8696 - val_loss: 1.2405 - val_acc: 0.6843\n",
       "Epoch 25/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 0.3566 - acc: 0.8714 - val_loss: 1.9813 - val_acc: 0.6278\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.3626 - acc: 0.8721 - val_loss: 1.2756 - val_acc: 0.6634\n",
       "Epoch 26/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 0.3400 - acc: 0.8782 - val_loss: 4.9607 - val_acc: 0.3447\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.3432 - acc: 0.8789 - val_loss: 1.2590 - val_acc: 0.6708\n",
       "Epoch 27/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 0.3165 - acc: 0.8824 - val_loss: 2.1550 - val_acc: 0.6049\n",
+      "62/62 [==============================] - 0s 5ms/step - loss: 0.3165 - acc: 0.8909 - val_loss: 1.3211 - val_acc: 0.6662\n",
       "Epoch 28/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 0.3147 - acc: 0.8845 - val_loss: 3.1088 - val_acc: 0.4463\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.2937 - acc: 0.8960 - val_loss: 1.3015 - val_acc: 0.6746\n",
       "Epoch 29/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 0.2944 - acc: 0.8921 - val_loss: 3.7178 - val_acc: 0.5479\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.3091 - acc: 0.8910 - val_loss: 1.3578 - val_acc: 0.6637\n",
       "Epoch 30/30\n",
-      "62/62 [==============================] - 1s 9ms/step - loss: 0.2819 - acc: 0.8980 - val_loss: 3.5398 - val_acc: 0.5876\n",
+      "62/62 [==============================] - 0s 4ms/step - loss: 0.3003 - acc: 0.8931 - val_loss: 1.3836 - val_acc: 0.6673\n",
       "Evaluate on test data\n",
-      "test loss, test acc: [3.5338056087493896, 0.5878309607505798]\n"
+      "test loss, test acc: [1.3836346864700317, 0.6675152778625488]\n"
      ]
     }
    ],
    "source": [
     "print(\"Training...\")\n",
-    "train(X_train, y_train, X_test, y_test)"
+    "model, history = train(X_train, y_train, X_test, y_test)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 26,
-   "id": "de7a2614",
+   "execution_count": 104,
+   "id": "ce37826d",
    "metadata": {},
    "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "(52,)\n"
+     ]
+    },
     {
      "data": {
       "text/plain": [
-       "((14, 75, 1), 52)"
+       "array(['Q'], dtype='<U1')"
       ]
      },
-     "execution_count": 26,
+     "execution_count": 104,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "train_shape, classes"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 27,
-   "id": "9f67f663",
-   "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "(14, 75, 1)"
-      ]
-     },
-     "execution_count": 27,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "X_train[0].shape"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 28,
-   "id": "376ed54c",
-   "metadata": {
-    "tags": []
-   },
-   "outputs": [],
-   "source": [
-    "d = data[0]['data'].reshape(1,data[0]['data'].shape[0], data[0]['data'].shape[1], X_train[0].shape[2])\n",
-    "pd.DataFrame(d.reshape(d.shape[1], d.shape[2]).T)\n",
-    "dd = d.reshape(d.shape[1], d.shape[2])\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 29,
-   "id": "8bb6a389",
-   "metadata": {
-    "tags": []
-   },
-   "outputs": [
-    {
-     "data": {
-      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjUAAANOCAYAAADph/0uAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/Z1A+gAAAACXBIWXMAAAsTAAALEwEAmpwYAAEAAElEQVR4nOy9d3hkZ3n3/7mna0a9rFbbu+11t+V1w5jgNZhm02xsCJgAr0MSUl7SSMhL8sLLLwYSIAUIpppeHIqJDcY2Nhjjsuvu7evtu2qrLo2mP78/TpkzoxlpJE3RjJ7Pde21U45mHo3OnHOf7/2971uUUmg0Go1Go9FUO65KL0Cj0Wg0Go2mGOigRqPRaDQaTU2ggxqNRqPRaDQ1gQ5qNBqNRqPR1AQ6qNFoNBqNRlMTeCq9gPnQ3t6u1q1bV+llaGqAp5566rRSqqPc76v3YU0xqcR+rPdhTTEp1j5clUHNunXr2LlzZ6WXoakBRORoJd5X78OaYlKJ/Vjvw5piUqx9WKefNBqNRqPR1AQ6qNFoNBqNRlMT6KBGo5kFEblORPaJyEER+VCO59eIyEMi8oyIPC8ir63EOjWafBSwD/tF5Pvm80+IyLoKLFOjWTA6qNFoZkBE3MDngNcAW4FbRGRr1mb/APxAKXUhcDPw+fKuUqPJT4H78HuBYaXUJuAzwCfKu0qNpjhUpVFYMz9u//leROBvrzuz0kupJrYBB5VShwBE5HvADcBuxzYKaDRvNwGnyrrCGuU9X9/BG87v4k0Xrqr0UqqdQvbhG4B/Mm/fBfyniIjSwwGrmmRK8Zp/+w3HhsKVXgoADQEvOz68vaTvoYOaJcSThwdxu6TSy6g2VgLHHfdPAJdmbfNPwC9F5E+BEJDzWysitwG3AaxZs6boC601fnvgNKtb6nRQs3AK2YftbZRSCREZBdqA086N9D5cXUxEE+zvm+DyDW2ct6qp0svB7yl9ckgHNUuIeFKRTOkLrxJwC/B1pdS/isjlwDdF5BylVMq5kVLqDuAOgO7ubv2HmIFUShFLpojEU7NvrCkbeh+uLmIJ4/vz2nOX887L11V2MWVCe2qWEPFkilhSH4fmyElgteP+KvMxJ+8FfgCglHoMCADtZVldjRJLGgfjSCJZ4ZXUBIXsw/Y2IuLBSKMOlmV1mpJhfY+87qVzql86v6mGWDJFPKmvfOfIDmCziKwXER+GEfjurG2OAdcAiMhZGEHNQFlXWWNEzSvMSFwHNUWgkH34buBW8/ZbgV9pP031Eze/R74ypH0WCzr9tISIJ1MI2lMzF0x/wQeA+wA38FWl1C4R+SiwUyl1N/CXwJdE5H9jmIbfrU8IC8OSza3gRjN/CtyHv4KRNj0IDGEEPpoqx1JqdFCjqUniCYWIPtfOFaXUvcC9WY99xHF7N3BluddVy0TNtJNWaopDAftwBLix3OvSlBbr4sC3hNJPOqhZQsSTKUQLNZoqIGann7RSo9HMl6hOP2lqmZj202iqBO2p0WgWTkwHNZpaRpuENdWCdTCOaU+NRjNvrGO+Tj9papK4LufWVAlaqdFoFs5SVGqWzm+6xEmmVMY/jWYxY3tqtFKj0cybpVj9VNLftIDJsB8Ukd3mZOMHRWRtKdezlHGmnnQaSrPY0dVPGs3CWYrVTyX7TQucDPsM0K2UOg9jiNonS7WepY4zkNG9PzSLnZgj/aRb/mg080Onn4qLPRlWKRUDrMmwNkqph5RS1vjQxzHad2tKgNNPo5UazWLHCrxTChI6XarRzIvoEjQKl/I3zTUZduUM278X+Hm+J0XkNhHZKSI7BwZ0B/q5otNPmmrCWfWkU1AazfxYimMSFsVvKiK/D3QDn8q3jVLqDqVUt1Kqu6Ojo3yLqxGcJwldJqtZ7EQdgyx1Az6NZn4sRaNwKUu6C5kMi4hsBz4MXK2UipZwPUsardRoqomoVmo0mgWjjcLFZdbJsCJyIfBF4HqlVH8J17LkcXpqYgntUdAsbpxBjTa2azTzI5ZI4RLw6KBm4SilEoA1GXYP8ANrMqyIXG9u9imgHvihiDwrInfneTnNAnGqM3pcgmaxoz01Gs3CiSVTeJdQQAMl7ihcwGTY7aV8f02amE4/aaqITKVGBzUazXyIJVJLyk8Di8QorCk9ccdJIq7lfM0iJ1Op0furRjMfYskUfh3UaGoRp6cmqpUazSLHqc5opUajmR+xRGpJmYRBBzVLhozqJ63UaBY5WqkpDiLSKiL3i8gB8/+WPNv9QkRGROR/yr1GTenQ6SdNzZLpqdHVT5rFTTSRQsS4rY3CC+JDwINKqc3Ag+b9XHwKeGfZVqUpCzqo0dQsmdVP+iShWdzEEika/EYdg1ZqFsQNwJ3m7TuBN+baSCn1IDBepjVpysRSrH5aWr/tEiYz/aSVGs3iJppI0ljnBbRSs0A6lVI95u1eoLOSi1mqPH5okKODk2V/33hSKzWaGsUZyGijsGaxE0umaDKDGt18b2a2b98OcLaIvJj1L3uAsAIWdEWjZ/DNj7/8wXN84eGXyv6+0SVoFC5pnxrN4iGmjcKaKiIaT9EY0EpNITzwwAOIyC6lVHf2cyLSJyJdSqkeEekCFtS5XSl1B3AHQHd3t5Z8CyQSTzIeSZT9fWOJFA2BpXWaX1oh3BLGWU2im+9pFjuxZIqA14XP4yKiS7oXwt3ArebtW4GfVnAtS5ZESjEZq0xQo/vUaGqSDKOwVmrmhIhcJyL7ROSgiOSsHhGRm0Rkt4jsEpHvlHuNtUY0bngB/B4XUW0UXgi3A9eKyAFgu3kfEekWkS9bG4nII8APgWtE5ISIvLoiq61REskU4Wj5g/OlaBReWrrUEkZP6Z4fIuIGPgdcC5wAdojI3Uqp3Y5tNgN/B1yplBoWkWWVWW3tYHRCdRPwunXzvQWglBoErsnx+E7gfY77V5VzXUuNREoRjpdfqdFGYU3NEjN70/jcLvu2piC2AQeVUoeUUjHgexhlsk7+F/A5pdQwgJ44v3Ci8SQ+j4uA16VLujVVTyKlKqPULEGj8NL6bZcw8aSxc/s8Lp1+mhsrgeOO+yfMx5xsAbaIyKMi8riIXJfrhXTlSOFYM2sCHvecjcJHByf57AP7SaZ08L6UUEpx8x2P8YsXe2bfuIwopUhW0FOjlRpNTRJPpPC6Ba9bdPqp+HiAzcArgFuAL4lIc/ZGSqk7lFLdSqnujo6Oki7oyOlJxiPxkr5HKbE9NV7XnIOae17o4bMPHOC3B0+XaHWaxUg8qXj80BDPnRit9FIySJjBdThWIaVGBzWaWiEcS/CLF3sBQ6nxegylRgc1c+IksNpxf5X5mJMTwN1KqbhS6jCwHyPIqRhv/a/f8bmHyt8Xo1hELU+Nxz3n9NPgRAyA7+84VoqlaRYpVpXcYmsBkEimgxqjVVD5iCZ1+klTQ9zzfA/v/9ZTnBqZIpZUeN0uvG6dfpojO4DNIrJeRHzAzRhlsk5+gqHSICLtGOmoQ2VcYwbRRJLTEzGODZW/g2kxUErZV5jzMQoPTkQBuH93H6fN25raJ2IqIYutWWMiZawnmVJlXZtSShuFNbXFRNTI4Y5HEmlPjduV0YhPMzNKqQTwAeA+YA/wA6XULhH5qIhcb252HzAoIruBh4C/NqtOKsJo2Eg79Y9V5wnd2j/98zQKD07GaK/3EU8qfvx0tqimqVWs/WSxKjUAU2VMQSVSCqVYckqNLumuYayrgnDMCGq8btHpp3mglLoXuDfrsY84bivgg+a/ijNsBjV945GSvL5SivFowu74W2ys/dbvceH3uOfcfG9wIsb5q5oZmYrz30+f4H+9fEMplqlZZFj7yeJTatJBzWQsQUvIV5b3tRR5rdRoagbrimUqljSDmnT6KZFM8dGf7aZndKrCq9QUm+Gw4SnpG4uWJIf/6/0DdH/sAU6OlGbfiTmDGu/cm+8NTcZoq/dx8doWDp+eLLuPQVMZLBUkutiUmlR6/y2nWVgHNZqaI63UJIklDE+NodQojg6F+eqjh3l4ny4trjVGzKAmlkgxNlX8MtJ9vePEkimeOjpc9NeG9H47H0+NUorBySitIT/LGwNEEylGwtVbBaYpHOsibtEpNY70U66gJpFMLajB5Fgkzr89cIBwVsm4lcZdah2Fl9Zvu8SwvuTheNKufvK6hVgyxYQ5XG2x5Z81C8d5Ei9FCqpn1HjNF06MFP21wanUzL36aTyaIJ5UtNf76GoKAOn1amqbqfgirX5ypJ/C0ekXGZ/65T5uuePxeb/+7w6e5jMP7OfLjxzOeFwrNZqawzoZTMUso7DY6SfLRKy7tdYew86gZqz4J/ReM0h4vkT9QKyr1vn0qbHKuVtDPjrNoKZ3TKdYlwLWsawSSk3/eIQrb/8VO44MTXsu4fAwTuZQao4Nhjk2FJ73e1vTv+/4zSGGJmP2407D/VJiaf22Swzr5DAZTXtq/KZR2PoiTC2yqxrNwrHST1CaCqgeM1DadWqMVAm69jo9NQGPm0RKZZwYZmJo0vh92+r9tlLTO1qdVWCauRGpoFLz8N4BTo5M8cCevmnPZSg1OboKh2PJBXltJs0L1Ilogi88fNB+3FZqdPpJUytYBsupeHJanxrri7DYTHWahTMSjtMYMAobS5F+6h2dIuB1MRFNcHiw+L1wMj01rozHZuO0qdS0hXx01PtxibFeTe1TSU/NI2b36meOjUx7bjZPTTiWYCo+/8Z8lur+hvNX8I3HjtoXADr9pKk5bE9NLGGOSTCCmnjSmX7SQU2tMRyO0dVUR0PAU3SlJp5M0T8e5eotxpiHF0qQgsrw1HjdQOH7qSW/t9X78LhdLGsIaE9NlfPIgQEO9k/Mul2llJpUSvGoGdS8cGJ0mqrorH6azOGpMToNz98KMB5N4PO4uGhNM9FEitEpI/1spZ90UFNEROQ6EdknIgdF5EM5nn+5iDwtIgkReWsp17IUcVY/GZ0lxa5+soIanX6qPUbCcZqDXjobA/QXWakZGI+iFLxscwcBr4sXThY/qMnw1JgH5EiBV99WN+FWsxfI8qYAvSXwFWnKx9/c9Twfv2f3rNtNVchTs7tnjKHJGK84o4OpeJJ9feMZzzvTT7ma71mP5UpNFcJkNEGD30NTndE3asy0FlgXB7r6qUiIiBv4HPAaYCtwi4hszdrsGPBu4DulWsdSJl+fmqg2Ctc0w+EYLUEfyxr89BVZqbFUj1UtdWztaiyxUuOas1IzOBmjwe/B7zF+rqtJKzXVzkQ0wc4jw7NOXa+UUmMNTv3A720Cpqeg4rMYhcN2UDO/dU9EEoQcQY1WakrHNuCgUuqQUioGfA+4wbmBUuqIUup5QJ9ZS4DVYdNQalSGUXhCG4VrlpGptFJT7Oonq/KpqynAOSub2N0zVtTXh9yemqePDvP6/3hk1llOgxMxWuvTHVuXNwXsNWuqk0g8yXg0we5TYxmPPXd8ZNp2YOw/5Wy4+NsDpzmjs4GL17bQXu+bFtQkM5SaXOmnhPn/zMfiRDLFbd/Yyf/5yYsZj09Ek9TnCmq0UbjorASOO+6fMB/TlAnLKByOJYnZSo1oT00No5RiJByjOehjWaOf/vHidhW2UjnLGwN0NdUxEU0UfZ5NxpgEU6n54m8O8eLJMR45MHOzyKHJGG2ONvRdTQEmognGI0uzAZ+ItIrI/SJywPy/Jcc2F4jIYyKyS0SeF5G3VWKtuYgnU8RNo+0Th9Pj1L7zxDHe/IXf2SdwSB/LlKJs8+3iyRQ7jgxx5aZ2RIQLVjfzzPHMppROo3AupWYqXlj66eP37uGXu/t4+ljm609E49QH8gc1uqR7kSIit4nIThHZOTBQ+S64D+/rX/QBQVqpmalPzeL+HTRzY9JU5VqCXpY1BIg5jIOzUch2VuVTU53XDh4GJ4ub4nIqNdYB2TKKPnl4eh8QJ6cnjG7CFp2NRll3Kfr1VAkfAh5USm0GHjTvZxMG3qWUOhu4DvisiDSXb4n5cSrJjx9K/+0Pn54kmVIZyp1z23L5avb1jhNNpLhwTTMAF65p4dDAJNd99je85+s7gJlLumOJdNA208XBvS/08LVHj+DzuBjLCtAnoomZlRod1BSNk8Bqx/1V5mPzQil1h1KqWynV3dHRseDFLYSe0Sne/bUd/Oy5UxVdx2w4lRqr+snncZFIKcbMHV97amoLq0eNkX4yTu6HT0/y+KH0Ve7+vvFpV3v7ese58KO/zJD4c9EzGqGrqQ4Rsc24zoZfxSBX9RNAZ6OfJ2YJaobMCd0WXU119rqXKDcAd5q37wTemL2BUmq/UuqAefsU0A9U9iBrEjFP9D63iycPD9qpnFPm3DHnvuc8lpXrYs0yyp+3qgmA685ZziXrWgjHkjy0rx+l0j2W/B7XtBSTM5CZKf30+KFBGvwebrx41bTRJ5Nm+qnRMgpneWq0Ubh47AA2i8h6EfEBNwN3l/D9yoYVCVs9MRYrmUZhZY5JMP7kVit9rdTUFtbftTnos1WKd3z5CW6+43F2mt1O//5HL/D+bz6VkZY6fHqClDICoJnoHY2w3Hxdy7syWOTvgVX9ZDXfA8OYfOsV6zg0MJnXV5NKKYYmY3awBehRCdCplOoxb/cCnTNtLCLbAB/wUp7ny6qYWyf6Sze0MhZJsLfXCLqtYarOfc95LJvrENT58vyJURoDHta0BgHY2FHPD99/BTdevAqlDD9N3AzEGuu8hKOZx9twPOG4nf9YHIknCfk9tIZ8jEfiGd/dcdMoHPC68XtcWqkp1QsrpRLAB4D7gD3AD5RSu0TkoyJyPYCIXCIiJ4AbgS+KyK5SraeYWF+0kalFHtRYJd3xhO2psUxjQ+YVvTYK1xbWhO6WoI81rUE8LmFVSx0iRpVGOJbg2eMj9I9H2d+X7v0xNGkcCIfCM+/ThlJjBArtZppnsERKjc/tIuQ3gpo3X7iSS9e3AbAjj1ozFomTSCna6tPpp2WmWlXLZuHt27cDnC0iL2b9yy7MUEBeg5WIdAHfBP5AKZUzKii3Ym4dn15xxjIAdh4xFMaTw9OVmsz0U7mUmhHOW9WMiGQ87jUDiXhSkTT71DQGPExmpZ+c6kwuE7FFJJ7C73XREPCQUpnenMloggaz2WZjnZdR88ImvkSrnzylfHGl1L3AvVmPfcRxewdGWqqqsKLt0UU8/VcpZZ8cLLnS5xZ7Bx82DwY6/VRbWHOfWszqp4f+6hUsbwrwps8/ymMvDXLx2hY7x//IgQHOWN4ApMcLDM8QoKRSir6xCMubMpWaoRJ4arxuweUS1rQG+eRbzuM15y4301Eunjg8xGvO7Zr2c1Zw5TQK+z1u2ut9Na3UPPDAA4jILqVUd/ZzItInIl1KqR4zaOnP9Roi0gjcA3xYKTX/6YpFxjrpb2gP0Rz0sq9vnLFInHHTE+jc96IZ6afSH9ci8SR7e8a57eUbpj1nKeIxh9G5qc47bWJ8oemnaCJJwOOmMZBOMdX7PSSSKabiSUI+j/0ellIT1dVPmkKxou3sHXQx4TTKWcYyq08NpM1rOv1UW4yaSktT0Dj4rW4N4nW7uHxDG88cG+GhvQN4XMLq1jp+c+C0/XO2UjNDUHN6MkoipWylJuRz4/O4ip5+iiVSdp8ZEeGmS1bTEPDi87i4cHVLzqGBYDQGBKObsJOupjo7XbEEuRu41bx9K/DT7A1Me8CPgW8ope4q49pmxTo+1fncbF5Wz4G+cVulgUyVcCqexOMyFJNyKDV7e8dJpJTtp3HicxvriCdTtg+osc47o1IzU1ATiacIeF1p34x5TJ80L7DrA+mgxnpOl3RrCmaqCtJP1sGgIeDBSr9aJd25ttPUBpZS01yXeWK/YmM7sWSK7+04xnmrmth+VidPHBq0//62UjND+umOXx8CYOuKRsAIONpCvqKnn6KJZF7J/JJ1LezuGctZKWJVSG3oqM94fH17iMOnZ2+zX6PcDlwrIgeA7eZ9RKRbRL5sbnMT8HLg3SLyrPnvgoqsNgvr7xz0udnc2cD+vomMoCbTKJy0K4Dm46k5PhTmD7+5k7+567mCtn/hxAgA565qnvacdfEYT6Zso3BjIIenxhHkzFT9FE0k8WcoNeYQS/PnG/zTlZpYMoXHZSieSwkd1MyDalJqWoLpk5vX48o4WTQHvSRSKqPjpaa6GQ7HqPd7pgUFl6xvxe0SwrEkl21o4+WbO4gmUrZHYcjcl4fz7NMP7O7jy789zK2Xr+Xita324231vpJUP+XrrbF1RRNKMa0VPcCenjEaAx5WmEqSxcaOek4MTy3JAF4pNaiUukYptVkptV0pNWQ+vlMp9T7z9reUUl6l1AWOf89WdOEmlnm2zmsoNaNTcZ4zg4mVzXWZQU0iaSuUkTkqNU8eHuLaz/ya+3b18d9Pn8w5oymb50+M0hbyTdvfwBHUJJSdfmqs8xDOGlxZaPrJ8tQ01hnBi1XhZDVRDeUKahKpJeenAR3UzAsr2l7MQY11AG9x+At8bsmQIjtMQ+VSPNjXKqPm3Kds6v0eWya/fGMbl25oxesWHjloVLDM5KkZi8T567ue4+wVjfzda8/KeK415C+BUpP/YHxWl+EB2pOjk/He3nHO7GqcZtrcuCyEKqCyS7P4sEq663xutnQaf/uH9w3g87jY0lmfkfqciqVonqdS86u9/SSSik++5TySKcVTR4dn/ZkXT41xzsqmafsbpI3CMWf6KeAlmVIZ1gCn4XcqPpNROEupMVNMVr8xZ/rJCmqMeX9L7xS/9H7jIlAN1U+WUa7VcYKz+tRYtNtBjVZqaoXhcCxnUAPwii3LqPd7uHhtC0Gfh03LGjhgVkANz+Cp+dpvjzAcjvOJt5yX0TcGDFPu4CyjC+bKTErN6pYgIZ+bvVlBTSql2Nc7zlmm8dnJRjMd9dLAkk1BVQUvnhzldf/+iN1rCdIVTZZSA0ZvmJXNdbTV+zPSpdF4kmZTmZ6rUjNstgJ43XlduF2S0b04F/FkioP945zV1ZjzeaenJm5VP5kBV66Kp5DPPaNSE0tkeWqmsoIas0qwsc7LeCRBMmUUiiw1Pw3ooGZeWHnQSDy1aFUOyyjnVGqcRmFIGyoX6++gmTsnR6ZY1jBdDgf4o1ds5IEPXk3QrJRY2VxnNzEbzOOpGZ2K85XfHuLarZ2cs3K6IbI1VPz000xKjcslnNnVyJ6ezPTTieEpJqKJnCeZ9e0hROClfq3ULGa+8dgRdp0ay2g1EHYoNR0Nftszs7K5zvZzWemcqXhy3krNoBnUhPwezl3ZxBOHcpvRLQ4NTBJPKls5zMbpqUk6qp+M38nRm8b8/dob/LOkn5IEvG67dNuaxG2ln+r93oz3GI/EdfpJUzhOB3uhLejLTVqpyR/UdDTo9FMhiMh1IrJPRA6KSK4289Z2bxERJSLTSmsL5aWBiXnPahqPxDnQP5GzGgOMfhXLHfn/lc0BTo5MMRVLEomnaPB7CMeSGfvD1x49zFgkwZ9fsznna7bV+6b9zEJxVj/l4szlDezpHcv4nPaYTdnOzBHUBLxuVrXUaaVmEROJJ/n5i71AuooN0kpNwONGRGy1ZkVzgNaQj1giZY4GSZFIqbSnZo7743A4Zl/kXbqhledOjGT4XQ70jWfsb1YTwDOX51ZqnEGN1XzPCkhyVTy1BH2zGIUN9dLrdlHndduzzCzvj9XPyTkqIarTT5pCce6Ui9VXk0up8XkkZ/pJN+DLj4i4gc8BrwG2AreIyNYc2zUAfw48Md/3GhiPcs2//ppLPv4gf/bdZ/i3Bw7ww53HeXBPH8eHwrP+/PMnRlHKmD9TCCua6xiPJDgyaCgYG80ThqXWKKX43pPHeeWZy3KqNIBj/lPx1JrJWGLGIXxndTUyHklklGnv7RlHBLZ01uf8mY0d9TqoWcQ8vK+fcVN1yJjnFEsQ8LrsCp7Npq9mZXMwPaZjImYHMVbV31xnPw1NxuyiisvWtxFPKp4xR4k8e3yEaz/zG361N93iZ0/POF63sKEjlPP17D41CaP5nscldi8Zpwl5Kp4k4HVR7/fMONDSUmrAMBxb1U9Wv56GLKVmdCqu00+awnGW5Y3M0oG1UlhKTUuWUpPbKDw/T82J4TAnhmc/2VY524CDSqlDSqkY8D2MeTrZfAz4BDDvLm91Pjefeut5XLGxjR1HhvjMA/v567ue57137uTqTz3EB7//7Iyft3UQviBHiWkuVrYYc5FeNOfXWN4TK52069QYvWMRXnPO8ryvYQ2PLJavJpZIsadnzDaF5sKS/Pc6UlB7esZY1xayU2vZbOyo59DAJKnU/FQwTXGYiiV54tAgO44MZQQvP3nmFO31flwyXampc/i4rKB1ZUudrawMTkbtY1iTWR0012Oac7p797oWXAKPm52rnzW/Vw86gpq9vWNsWtaQd66Sz5P21CSSCo9bqPO57c/AIhxLEPR5qJvFUxN1+MwaA+leNOnqJzPgsdJTU4klm34qaUfhWmUylsDnMaZdjzjST0opEim1KAaIWVcuraFMo7DXk3bqtzcszFPz5997lhdOjPKBV27i/VdvrNUv0ErguOP+CeBS5wYichGwWil1j4j89XzfqN7v4cbu1dzYbcyBjcST9I1FGJqM8fMXe/nGY0fYcXSIn33gZbYh0skzx0bY2BGyJfjZWNGcFdQsM646LdPwQ3v7EUm3qM9Fa5GVmhdPjRKJp7h0fWvebc4wJf89PWNs32qMMtrbO2b3z8nFho4QU/EkPWMRVpq/t6a8DE5EeceXn2BvrxGM+twu3nThSup8bn61t593XLaG/3m+J0upSWUEqlZqdfOyenvegzHE1Aiug34PXrfMqflePGlMsrdU7YaAl60rGu1xHLtNU/pv9g+glEJE2NszzuUb2/K+Zkb6KanwuFxppSYr/VTndRP0ufMq5gkztZZWatJBzaSpZHnM97O++6NTcaP6aRGci8qNDmrmwVQsyYqmAEcGwxmjEv7vz3azt3eM7912eQVXZ5CzT02WUrPQ9FPvaAS/x8Wn799PwOvitpdvXMCKqxMRcQGfBt5dwLa3AbcBrFmzZtbXDnjdrG0LsbYtxIVrWnjNOct52xcf58++9yxfe/cluB1NtZRSPHN8hFeemT8AycY6uVuThjdZSo2pPj64t5/zVzXb3qtcWBOxh4rUVfhJ80RyyQxBTb3fGCBonRz7xyIcHQrz5ovyT1yxVKjP3L+fX+7qJZlSrGyp44fvv8KW7DXFI5VS7O4Z41d7+9l5dJiuxgDPHB/m2FCYf73xfNob/PxyVy93PXUCt0u4dEMr77tqA48fGspSaoyTtsXFa1v59V+/grVtITslOziZTj/Ved34Pe45KTVWutU5XuPiNS388KkTJJIpdveMIWKY0Y8MhmkJeukdi3Bmjko7iwyjcCqFxy0ETTUlwygcTRL0GUHNZDT3cdia4Wd9Do0Bjz1MeTySsE3CkCP9VJsXmjOig5p5MBlLsqK5jiOD4YxqkeNDYXYcGWYqlrSlxkqRVmqcfWrSRmHj6sCTse1cGQ7HuGXbGn7+Qg/7emvWr3ASWO24v8p8zKIBOAd42OxXsRy4W0SuV0rtdL6QUuoO4A6A7u7uOedBLlzTwj9dfzZ//+MX+Nj/7OYf37DV7pFxbCjM0GSMC9c0F/x6HfV+vG6xr0QtT81IOMbAeJTnTozwv7dvmfE1bF9DkZSaJw8PsbEjZAfc+bhwTTMP7unn5MgUn3voIG4Rrj9/Rd7traDmrqdOcPmGNlY01/HfT5/g8UODvPrs/Ok1zdxQSvHBHzzHw/v6GQ7HDZ/TsgZePDlKIpniq7dewhWb2gG4eksH//C6rXjdYisNHQ3+LKUmOS2luLbNUBSd+54VxAS8xnywuSg1ljJppVIBLlrbwp2PHeVFsxrr1VuX84tdvfxmf3peWi5TukV69pMxpdvjchE0zwnOOWThuBHU1Hk9eQdaRuPW1Pq0UnPI7Lk0GU3Y5dyQFdQkU3b/mqXE0vuNi8BULMGWznq8bslIP0UTRqOl3T1jXLy2MLNmqcjVfM/rETsvWx/w2JH/fIKaaCJJOGaUUC5vCtA7VrOzdXYAm0VkPUYwczPwdutJpdQo0G7dF5GHgb/KDmiKxS3bVnOwf4KvPnqYhoCHP33lZnwel61wXLi68P3O5RK6muo4NhTGbQ6PBOMk8fC+fpRiVuWn3u/B53ZxughDLZMpxY7DQ7x+huDE4q9edQYP7O7j/d98it09Y7zzsrWsa89t2gRDUbr5ktVsWlbPe65cTzyV4mfPn2LnkSEd1BQREUEpxe+duYwrNrbz8i3tLGsIoJQipchQF4FpF3/t9T4OOrpFZ3tqnAR9bvweF0OTsYx+NnNVaqx2Bi2OVL11/P7hzuPEEimuO2c5u3vG+M3+AbsDe66eSBY+u6OwUdLtcQnt9X42Lavn9p/v5aX+Cf75zecyZXpqgj633W04u5nfdKXGm9Gnxhm41HndeN1iKzWLwQpRbnRQMw8mzauHpjpfRvWTNUDs+RMjFQ9qrPRTvd+D2yUkTa+PtZM3+D32wWI+RmEr7dYc8tHVVGeX1NYaSqmEiHwAuA9wA19VSu0SkY8CO5VSd5dzPSLCP7zuLCaicf7jVwf5z4cOEvJ5jIOb35O3+icfK5oDHBsK0xL04XW7aKrzMjwZ40DfBJ2Nfs6ewadirac15Jtz+ikcS1DndWccwPf0jDEeTXDZhvypJ4vVrUH+/nVn8eEfv0i938OfvnLTrOu8/S3n2ff9Ljfnr2pix5HZO8dq5sZnb75w2mMigruAEUSGUhOzT+5TsSRNOfxj1msazR/T6aeA14V/nkpNm0OpWdlcR2ejnx8/Y4iyW1c08vIt7Xz7iWM8uLefZQ3+GdOyXk9m8z2PW/C6Xdz9gSv5xM/3cudjR7n+ghWEY0mWN3qp87lRyjhuZze4nK7UeBiLJFBKMRExvvfOz8TqKhyJ55+hVsvooGYehKMJQj43zUEvo46uwtYX6YUTo5Vamk00nkQE/B4XQa+b8WjCSD9lKDWmG38eSo01I6glaCg1D+3rz3mVUQsope4F7s167CN5tn1Fqdfjcgn//ObzeNnmDl7qn2B0Ks7K5jq2rW+1ZfxCWdkcBIZsP0FL0MvARJTfHjjNdecsL+jvOZf5T1a66Ic7j3PV5g4+fdP5NNV5OT40xbefOAbAJetmD2oA3r5tDS/1T3L+6ibaZklX5aJ7XStf+s2hRZEu1hh01PuJJVOMTSVoCnqZiifpyqPUALTW+xgOp5WawDyUGmtEiDNVLyJcvLaFe1/oxedxsaE9xC3b1nByeIpL1rfy+nNXzPjdyBxoqezp4UGfh9uu3sidjx3lxPCUve9ZqalwLDktqEmn1swLUnPcwlQ8yUQ0wYrmzGabjXVeHjkwwInhKW66ZDVLDR3UzBGllJ0Hba7zZig1ljpiDVyrJBGzBFDEKCUcjyYyjMIhXzqomU/6yfISNdf56GoKEI4lGYsktOmyTLhdM3tICmWleUC0pPeWkI9HDw4yFknw8i0dBb1Ge72fvvHZK9kPn57kzZ9/lMlokmvO7OTBvX1s//RvUErZ1VNXbW63q7JmQ0T4yBumtQwqmEvWtfCFhxXPHh+ZsZJFUz4s9WNgImoHNTMFnNbssYgjqJmrp8ba97LHi1y0xghqzuhswON2cfaKJr72B9sKek2npyaZUhkXG50Nftwu4eTwFOFY2igMhoLpDK4gPfLBb1U/OSZ1Wwqtk6Y6L88cG2FNa5D3XLm+oPXWEjqomSOReAqljNLB5qCXUyPpg3nMzLUeOj3JeCROQ6ByJ/ioOQANzAmu41G8bsFrasD1ASMt5XO75pV+soK5ZlOpAaMaSgc11YUVQFjSe2vQxzNTI7gEXrapfaYftVnfHmLnkaEZlbrTE1He/bUnAbj3z69i07J6nj0+wr89sJ+OBj/nrmrmio1tbJjBF1NsLl7TigjsODKkg5pFgmUQHxiPsmlZ/awqWnvIx0v9E/ZYhIDXhd/jmtOYhOHJGE113mn+E8tCsHUGQ3A+fBkl3SlbqQHwuF0sbzS6eaf71Bin4lxdha3fxe5TY03qjsSZjCbsCd0W1jH4n67fOk31WQrooGaOTDoGkDXV+TJm0ETjKdrrfZyeiLHr1BiXbajcgTIST9lypeWd8ZrKjdctNJhfhIDXNS+lxmo62BLy2VdFPaNTdmWApjqwghqnUgNw/urmnL1wcrGhI8RkLEn/eJTOxulzp/rHI9z61R30jkb47m2XscmssrpgdXPBV76loCno5YzOBnYcSc/5SaWU3b1WU34spcaqgJqK5TcKA6xpC/LjZ0/aw4XrvG4CXnfOafP5sOY+ZXP2iibOX91s90KaC9bFY9wsHvFkGYpWNteZQY0RtIUc6adsLKUmME2piTOeZRQGeO05XWzqqOeVZ8593bXA0nMRLRArkq7zeWgJejM6CseSKTu6f77CKahoIp2btaRN6+rB53bZ0X3A6541qJmKJfn0/fszriIsT41R/WScGHtH591MV1MhrK7CVjmrdXC/usDUE8CG9vxTsI+cnuTG/3qMI6cn+dK7urmowBEO5eKyDW387qVB/uW+fXz90cNs+/8e4A++9uScToqa4uFUaqxU/0xBzcaOepTCvrg0PDVzU5+Hw7mDGp/HxU//5EqunUdQ43YJIunZTx5X5ql2ZUsdRwcnSaQUQa/bVqNyBTXTlRojqDk9ESOWSFGfVfJ+0yWr+YfXzz8tW+3ooGaOOJWa5qCXyVjSrnqKJVJ0NdWxoikwbYpwuYnE0221rS+MJa/+w+u3cvM2w0AW8ObvZGnxy929/PuDB/jtwdP2YyNTMXxuo/fCsgY/Ipn9FzTVwaqWOjYvq+eC1UanVqtZY6F+Gkh3In5pIHMK9s9f6OEN//FbRsJxvvW+S+f0muXiL7Zv5obzV/CfDx3kn362m1UtQR49OMjr/+O39gRzTflorvPicQmnJ6JEE0aqf6b0k6X6WV2xA6ZSMydPzUTuoGYhGIq4i1hSkchKP4Gh1PSNGWqUYRQ200/x6b1qotOUGmPbnlFj/1yKvWhmQn8ac8Tq+ljnc9ulhiNTMZY1BIgmjBK6ruY6+sYqe4KPZCk1Lkn3iLhlW7qbbV0BSs3jhwYBI41gMTIZpznotdNZHfV+rdRUIX6Pm/s/eLV9/7XnLmcqluD8AudHASxvDBD0uTnkUGp+d/A0f/Ttpzl/dTP/ecuFrDZ74Cw2moM+Pv22C3jn5WuJJVJsW9/K44eGuOVLj/Pg3n7eednaSi9xQYhIK/B9YB1wBLhJKTWctc1a4McYF7le4D+UUv9V3pUauFxCW72PgfFoRpfgfKxvDyFiqIQ+twu3S+as1AxNxua0vxeKz+2yp4dPSz+1pM3wVp8ayJN+imcFNaZS899PnwBgdcvi/G5VCq3UzBErBRPye2g2d66RcBylFDGz4mhZg5/+8Zmbkb14cpQv/vol/vWX+0oyPTgaTxGwjMI+T94mTIanZuYDwBOHDM+BdWUBhmTrrBboagrQU+FATrNw1raF+OCrzpjWJG0mRIT17SEOOZSabz1xlJagl+/fdtmiDWicXLimhUs3tCEinL3SMIZG59lpe5HxIeBBpdRm4EHzfjY9wOVKqQsw5pp9SEQWXlo3T6yuwuFY+gIyHwGvm1UtdaQU+E0P4VyUGqWUkX6qL65SA4avxijpTk1LPzkr/EJ+tx245Uw/JTLTTw2mMvPiyTFetqmda84qfDTKUkAHNQ5+99LpWXPpVvop6HPbpXST0QSJlNEx0+d20dHgz5hfkou/+9EL/PPP9/IfvzrIrV99klFHZ+JiEEkk7S95nc89Q1Azc/qpfyxit+QecCo1U/EMI+nypgC9o1quX6ps6Kjn0GkjOB+ajHH/7j7edOGqqqy+sC4G5js+ZJFxA3CneftO4I3ZGyilYkop64Dlp8LnhY56PwMTUfu4FJylh5A1s8wKDCyl5pljw7z23x6xVXOlpk8mGY8miCdVxtynYuE1lZp8RmELa6Al5K5+ylZq/B6jbL3e7+H2t5xbk73BFoIOakzCsQTv/MqTfOmRQ7NuB4Zk6LfHDKRsX43PVGqsjo756Bmd4uZLVvOjP76C3tEIf3PXc9O+dMOTsVmDIydKKT7xi728eHLU9NQYX4I3X7SKv9i+OefPBLzuGa9IHzfb7we8rgylZiQcoyVDqamjZ0QrNUuVDe0hTgxPEYkn+ckzJ4knFW+r0sZfXrdh8rSukKucTqVUj3m7F8jpehWR1SLyPMZE+k8opU7l2e42EdkpIjsHBgZKsuD2ej+nx2P2CX62wNia62VtZyk1v94/wO6eMf7twQMkU4o/+PoO/vf3n834WesitqXASr+54HW7iCWUPaXbiTOoMdJPxgVyIUZhgFsvX8e/3Hgeq3TqaRraU2NyYniKZEpxoH/mVJC104V87nTzukTaLOz3uOy0zMB4NKf0Hk+mOD0Ro6upjovWtPC3153Jx+/dw327ernunC57u/fcuYOjg2Hu+bOX0dU0e0OysakEX3j4JaZiSbP6yfgSXLy2Je/Yhjqvm54ZgponDg1S7/dw0dqWDJ/QcDjOhXWZSs14NFHx/jyayrBxmVGFcmRwkh/sPM75q5qqtrxfRAh4ZveaLRa2b98OcLaIvJj11Iedd5RSSkRyDlJVSh0HzjPTTj8RkbuUUn05tlvQUNZCsNJPE9G0Kj4T1iBWp1KTUrDrlDG65fs7jgPw8L6BaYNSrcZ7pUg/+TwOpSbHzKu2kI/ByRh1PkN5ESHnUMtIIonbJRlq+9+99qyir7dW0EqNybFBY4z9oVn8LWHTKBz0e2yZOhpP2ld1Po+bZQ1Gr46Bidwqi6W+dDYaX7D3vGw9HQ1+fvJM+uJoT88YzxwbYWgyxge+84w9RG0mrK6uB/rHiTqUmpmYzVPz+KFBute1sLI5YCs1SilGw3GaQ5meGqDiBmlNZbCa5v39j15gb+94hhm9GinEa7ZYeOCBBwB2KaXOyfr3U6BPRLoAzP/7Z3otU6F5Ebiq1OvOx7r2EImU4oA52HImozA4lZq0pwaMcTUXrmnG53bxnSeO0RDwcHoimmExsGaWtZZEqZGM2U/ZWGbhoM+YgVbndecxCqcILMEZTvOl5J+UiFwnIvtE5KCITDOpiYhfRL5vPv+EiKwr9ZpycXzYCGqODYVJJFP0jkb45C/2kkxlXoxYnpo6sx03TE8/WQ2k+sdyBzW95onfalTmdgmvO7eLh/b121cnP9h5HJ/bxcduOJunjg7z0Z/tzpkTdmIFFAf6JojE00rNTMzkqZmMJnhpYJKL17TQ0RBgcDJKIpkiHEsSS6YyJNvl5u9yYlj7apYiGzqMoObpYyPcsm111aaeLPyeuZUFL2LuBm41b98K/DR7AxFZJSJ15u0W4GXAvrKtMAurTPsFR5l2IdsHHEoNGMfZKze285ev2sLZKxr5+JvOBWC/Ywq4NaG7fYbhlPPF8tQ4Zz85WdGUDmqs/8M5jsXRRNIekaCZnZIGNSLiBj4HvAbYCtwiItldgd4LDCulNgGfAT5RyjXl4/iQcTKOJxXHh6f4/o7jfP7hlzh8OrP3xlTMCBbcLsmYnWQdAK3qJ8iv1PSbwceyxvQX6fXndRFNpHhgdx/RRJIfP3OSa8/u5J2Xr+O2l2/gm48f5auPHpnxd7BKqvvHo4xMxQtUavLL7GMRw7zc3uCns9GPUkbDp/Tcp7RSc8byBhr8Hj59//6qke01xSPo83DJuhbe1r2a//fG6jcvVpNSMwu3A9eKyAFgu3kfEekWkS+b25wFPCEizwG/Bv5FKfVCRVZLWnl53hwMPFv6qTXkoyXozTDSWmzurOd9V23gnj+7im4zBb/fYTE4bSo1pTIK55r9ZGEpNVZ1V53PnccorJWauVBqT8024KBS6hCAiHwPw42/27HNDcA/mbfvAv5TRETNJktk8cyxYb78yGE+deN5tulqLhwbCuN2CcmU4tDABE8fM1o5ZFclTcYShHzpbrxgBTVppaat3o9LYCBPKsZK4zhbyl+0poXljQF++uxJTo1OMRKO87Zu42r3Q9edybHBMP/vnt2cvaIx7/gFZxl5MqUKVmryzUmZiJiNBv0egubv2jcWsct9ndVPzUEf/3rT+dz2zaf4+x+9wDsuW0u938NiObd11PvtEQCa0vDD919R6SUUjUI6bVcDSqlB4Jocj+8E3mfevh84r8xLy0tTnZeOBr/tbyxkgvrbLlljp/P9juOe09fV1RSgwe+x01pgWAEa/J6SVOn53C7iiemznywuWN1Ma8hnz2pa1hDggT19fO3Rw7zzsrV2IBRNpLRSMwdKHdSsxHDTW5zA6IOQcxulVEJERoE24DRzIBxLcs8LPbzxwpXzamt9YjjMRWua2XFkmIP96aDGOQYBDE+N9SWz00+JlD3M0u8xVJzWkD+vUtM3FsHrlow8rsslvO68Lr7y28M8tG+Ai9e2cKU5UNDlEj7ztgu48GO/5P7dfXmDmr6xCCJghYOFfFHrvG5ippktuzeJlQpr8Htsg13fWMQOGluyptq+6uzlfOD3NvGfDx3kR8+cnPW9y8k/vO4s3nfVhkovQ1Ml+D2uWql+qko2ddTz2LjR9DPonf009aHXnGnftpQaj0vsER5gGMA3ddZnpZ9iJUk9AXg9QjSeu6QbDHX+9ed12armp956Hh/56S7+7892c3oiyl+/2vidIvFkRuWTZmaqpvpJRG4DbgNYs2a6CfGSda00+D38am/fnIMapRTHh8Lc2L2alwYm+eXuPsZNlcKaRm0RjiXTSo2jn4WldvjMnW9Zgz+vp6ZvLMqyhsC0wXnvvmIdA+NRXn9eF9vP6sx4vs7n5pwVTTx3fCTv79E3FmFjRz0nh6eYKvCLkPYFJadNe7WCmpDfY18F9Y1Haa4zftdcAw//8lVbeM25yxkYj9rdlxcDZ3VVZyWOpjL4a0SpqVY2LgvxmNnJPOCb2wndOqataw/Zx2OLLcsaeGBPuqjr9Hi0JKknMNJPE5GEqdRM/x2yU7QbOur55nu38YHvPMM3fneUP7x6I40BL9FEqir7PVWKUgc1JwGnY3CV+ViubU6IiAdoAgazX2i2UkKfx8XLt3Tw4J5+UinFHY8c4uTwFB95w9a8jecshiZjTMaSrGkNsqE9xM6j6S7iIznST0G/sYO5XILPbeTenUoNGGWJMyk1Tj+NxerWIP9+y4V513n+6ma+/cRR4slUzt+pbyxKV1OAOq+bF06OFqbUWE2fcgQ1k2ZQU+/35EypZSs1YHxRz17RNOv7ajSLmYDXXfSGmJrCsRrquSQ9iLdQLKXmjM7pFzKbO+v5/s7jDE5Eaas3SsctD0+xsWc/5SjpzoeI8P6rN3LPCz1894lj/OHVG7VSM0dK/UntADaLyHoR8QE3Y7jxnTjd+W8FfjVXP43FK89cRv94lLufO8Wn7tvHNx8/yl98/1n6xyPsOjVKIk9Z9HGzYmd1a9Cu4mgOenEJjGaln6ZiyQzjmt/rMpUayyhsPDezUhOhsyGQ87mZOH91M5F4KkM+zX7dZQ0BNi/LLHGciZm6p1pqVUPAg9sltNf76RuLMmKWRDblCGo0mlrA73HVypiEqmTTMiMgCfo8czadW8e9LTmCGuux/X2GX2dwMkZbCXrUQPbsp8JPteeuauLKTW189dHDRBPGuUUrNYVT0qBGKZUAPgDcB+wBfqCU2iUiHxWR683NvgK0ichB4IPknk1SEL935jJE4G/++3nqvG7+9JWbuOf5HrZ9/EFe9++/5d9/dTDnzx0fMsq5V7fW2VH7RWtaaKrzMhzOVmqSGUZkq3ulpdT4HErN6YkoQ5Mx/u5HzzPoUG36xiJ2OmcuXGAOXXvu+Oi051IpRf94lOVNfjZ1Gr9DIdVPfkf6KZtJR/oJDGNz33iE4XCcoM9d0OtrNNWI8b3WnppKkV2mPRc6GvyIwIVrmqc9ZwU1B/rHSSRTDIdj0xryFYvM2U9zC8ze97IN9I1F+e2B02b6SSs1hVJyT41S6l7g3qzHPuK4HQFuLMZ7tYZ8XLSmhaeODvPn12zmT35vE+eubOLkyBQP7Rvgy48c4p2XrbX7yFgcs4KaliAbzKDm4rUtHD49OS39FI4lCDmUGqv00+5T4057ahIpxb8/eIDvPnmcLZ0N/MGV65mKJRmLJOhsmrtSs7q1jpagl+eOj/D2SzN9RYOTMZIpRWdjwO4+XMgXwWpsNRWbfgBPe2qMbTob/Rw6PcnRwTBr20JzXr9GUy0EPC7tqakgnY1+6v0e6ubopwFY1RLkib+7hmWN04+xnY1+GgIe9veNMxSOoVRpetSAkX6KxlOkFDmNwjNx7iojhX98KGymn/QFZKFUjVG4UN52yWriyRTvuXI9YFTkAFy9pYNrP/MbPvfQQd51+VqePDzE4dOTNAd9HOyfoC3kI+T3cNGaZs5b1cSrz+7k/t1906ufYknqnEqN2U7dnqTqtZQa4wv1nSeOAUaL7j+4cj39Ztff+aSfRITzVzfz3ImRac9ZjfeWNQTYtq6VV5zRwXmmsjMTlgozmaM993g0gc/tSqfUGgM8sMdoSPqt92YXsWk0tYOVVtZUBhFhY0do3r2CcgU01utuWlbPS/2TnB43ju3tpTIKe1x2Y9PZfJ3ZtIV8+NwuesYiWqmZIzUX1NzUvZqbuqd3M93QUc9N3av4+u+O8PXfHQEMVcVKG52/uhmAtno/d3/gZYDhqxmccLTUnowxOBG1u+dCup/FNKXGTC/Fkik2doR4/NAgkXjSbpDXmedLNxvnr2rmN/sPMBlNZBh7++wuxX6agl6+/gfbCno96zXCOYKayWiC+kD6Paymgm++aCUv29w+r/VrNNVAwKPTT5XmPS9bz1gJzNqbOup5eP8ApydK100YjHOBFdRkt8uYDRFheVOA3tGIVmrmSM0FNTPxv6/dQjSR4ryVTVx9xjLWtAZ54vAgH79nDy/PcZJuCfp4yTEL6tf7+0kpeMUZHfZj2eknq0lSh5mnXdEU4MOvO4v3fH0njx8aZMw0387HUwNGw6aUMrptXr4x3a8mV0O/QrBSabnKryciCTv1BHDZhjYuXDPAP7wuuym0RlNbWBcrSqmq745crdxwwcqSvO7GZfX88KkTdrf40pV0i31emKunBowhwT0jEaOjsFZqCmZJBTXLGgJ8+qYLMh67YmM79/xZ7tltTXXejD41v9prTHk9d2W6ZDngdTMRTdhjEiylZrnZvfIPrlzPFRvb8XtcPLxvgFVma+x88uhsXLS2BRHYcWQoK6gxGu9l+4Vmw04/RacrNRPRJPX+dIXTZRva+PEfXzmvdVczInId8G+AG/iyUur2rOc/iNGdNQEMAO9RSh0t+0I1RcOa9BxPKnweHdTUElYxyJOHh4DSKTXOiqf5BDVdTQGeOjpMNKGrn+bCkgpq5kpz0Mt4JGGXgv96Xz+vPnt5RlM8v8fN6YmYHZF7TUNYwOvm0b97JQ1+oyTx8o1t3LerlxXNdQS8LhoD8/vom+q8nLm80f5CWvSPR2gL+eecu7UaCU7mmDkyEY3T4F/au4hjftm1GB2xd4jI3Uop56iPZ4BupVRYRP4I+CTwtvKvVlMsrJNINJGc1sBNU91YlVVPHB7E53GV7BjnPBbPpaTboqupjp7RHlIK3admDuhPagasgY2jU3GeOjrMWCTBK89clrFNwGv0s4gmUvg9rgypujHgte+/+uzl9IxG2NMzxpsuXLkgSXvbOqPCK+7ou9M3Fp1XSstqJBjOodRMRpMZ6aclij2/TCkVA6z5ZTZKqYeUUmHz7uMYTSY1VUy607b21dQaq1vq8LldnJ6I0R7ylSy96HNUPM1XqUmmjJZtWqkpnKV9GT4L1gDEkak4v9rXj9ct0wyyVu49mkjNeEX3tu7VXL2lg87GwJxNY9lsW9/GnY8dZdepMS4wDc49oxG65lEm7nW78HlcTOQwCk9EE6xrX/Kl24XML3PyXuDnJV2RpuT4Z2hKqaluPG4X69qD7O+bKFnqCRau1Cx3HM+1UlM4+pOaAWt66kg4zuMvDXLRmhYaAplddANelz3QciaHusslrGiuW3BAA3DJ+hYAnjxsTJOwJouvn2cAEvK5CecyCkcT1GulpmBE5PeBbuBTeZ6/TUR2isjOgYGB8i5OMyes1gy6Aqo2sXw1pTIJQ2ZQ451jnxqAFWa/MUBP6Z4DOqiZAWtg49BkjH1945yzcvpMI7tPTTxVtmh6WUOA9e0h21dzZHCSaCLFmcvnN7Qx6PPk7FMzEUlQv8Q9NRQ2vwwR2Q58GLheKZVzPoZS6g6lVLdSqrujoyPXJppFgiX3a6WmNrGCmlJ1EwajT43FfC5mnUqNTj8Vjg5qZsDy1Dx/YoRIPMUZOYKGdPqpvIbCbeta2XFkmFRKsa/XmAV1VlfjvF6r3u+ZVv2UTKmcQy6XILPOLxORC4EvYgQ0/RVYo6bIWBcoVlWjprawzMKlTD9lemrmfm5oC/lshUennwpHf1Iz0GIqNY8fMtI8Zy2fHjQEvEbpZzhW3kmql25oZXQqzu6eMfb2jOGS9Bd1rgT9bsJZ1U8TjgndS5kC55d9CqgHfigiz4pI9tBWTZVhVz9po3BNUu7003yMwi6X2H3HtFJTOEv7jDULDQEPIsYASZcYY+uzsXa2sal4WZWaKzYahuXfvXSavb3jrG8PzXvHD/nSSk0qpVCkg5qGeZae1xIFzC/bXvZFaUqKnX6qcqVGRFqB7wPrgCPATUqp4TzbNgK7gZ8opT5QrjVWgjOWN/C27tXTqlmLSaZReH5eyhVNdZwYntJKzRzQn9QMuFxCU52XWDLFujxBg2XgGp2K2433ysHypgAbO0I8enCQvb3jnJlDRSqUkN9tdxT+x7t38e6vPTltQrdGs5SwTiI1UNL9IeBBpdRm4EHzfj4+BvymLKuqMD6Pi0+89Tx7gHEpyDQKz+/cYPlqtFJTODqomQXLV5PPhBswD35jkbhdMVEurtzUzhOHBzk2FJ63SRhMpcY0Cu/vG+eZYyOMR3T6SbN0cTbfq3JuAO40b98JvDHXRiJyMdAJ/LI8y6p9nJ2o51v12mUHNfpUXSj6k5oFqwLqjM7cSkigQkoNGCko60oyl4m5UJyemuFwjIloguNDRi85HdRoliI11HyvUynVY97uxQhcMhARF/CvwF/N9mK6LUHhLLSkG9JBjR5oWTj6jDULzUFTqenKo9TYpZ8z96kpBZdvaMMlkFLzr3wCI8VkeWiGzVlXu06NAmRM6dZolgqBKmq+t337doCzReTFrKc+7LyjlFIionK8xB8D9yqlTszWXVcpdQdwB0B3d3eu19KYOIMa9zyqnwCuO6eLE8NTrGkNFmtZNY8+Y83CrOknhyxY7hkxTUEv565s4qWBSVY2183+A3kI+TzEEiniyRQj4RgAu06N2c9pNEuNamq+98ADDyAiu5RS3dnPiUifiHQppXpEpAvI1XLgcuAqEfljjCo+n4hMKKVm8t9oZmGh1U9geGr+4fVbi7WkJYE+Y81CV3MdzUEvq1tyR8pOA1clBt/92TWbOT4UzhiyOVcsM/DAeJR40rj4soIaXf2kWYpUk1IzC3cDtwK3m///NHsDpdQ7rNsi8m6M4aw6oFkgviJUP2nmjj5jzcIfv2Ijt1yyJm/QEHCknCpRdnfNWdNS5HMm5DN+h5MjU/Zjo1NGGkpXP2mWIi6X4HO7asFTczvwAxF5L3AUuAlARLqB9yul3lfJxdUyXs/Cmu9p5oc+Y81CQ8A7bd6Tk0qmn4pF0AxcTgwb5mARUOa4+/mWImo01Y7f46p6pUYpNQhck+PxncC0gEYp9XXg6yVf2BKgGEZhzdzRZ6wFUun0UzGwhlaeHDaUGqvbpk49aZYyfq+7Kjw1msWJL8MorIOaclGdZ+FFhLM3TbWW3QVNM7CVfrpgdTOgU0+apU3A6yJa5UqNpnIUo/meZu7oT3qBOJWaam1lbVU4nTCVmvPNoEb3qNEsZfweV9WPSdBUDmfKSSs15aM6z8KLiEobhYtByEw/2UHNqibzcR3UaJYuAa9bD7TUzBuv43zg1UbhsqE/6QXidQtWEF6tnhoreDk5PEVDwGPPQ2nQQY1mCRPwurVSo5k3uqS7MlTnWXgRISJ2CqrcYxKKRdAs6Y4lU7QEfdT7PXQ0+Gmqy1/1pdHUOkb1k1ZqNPPDq43CFaFkZ2ERaRWR+0XkgPl/S57tfiEiIyLyP6VaS6mxgppyD7QsFkFH1+CWkDHr6nNvv4g/vWZzpZak0VScgNddCwMtNRXC7Uqr+NooXD5K+UkXOvL+U8A7S7iOkmNN6va5q7P6ye0S6szArMWcdbVtfSvr20OVXJZGU1ECXq3UaBaGFcxooaZ8lDKoKWjkvVLqQWC8hOsoObZSU6WeGkj7alrMqeQazVLH73FXffM9TWXxuV143cJsg0I1xaOUZ+FZR97PhcU88t5veWqqOqgxfgdrKrlGs9QJeF26+Z5mQXg9Lu2nKTMLKm8RkQeA5TmeKnTkfcEs5pH31qiEag5qLF+NVmo0GgOt1GgWitctupy7zCwoqFFKbc/3XIEj72sCq1dNNaefrFEJLVqp0WgAw/iv+9RoFoLX7cLj1vtQOSnlWdgaeQ95Rt7XCrWk1DRrpUajAYyLlVgyRSq1qIRhTRXhc7twa6WmrJTy074duFZEDgDbzfuISLeIfNnaSEQeAX4IXCMiJ0Tk1SVcU0moDaOwpdTooEajgfT3WvtqNPPFaxqFNeWjZC1jCx15r5S6qlRrKBfpoKY6S7ohPf9JG4U1GgPrIiUST1Lnq97vtqZyeD2ijcJlpnqlhUVELaSf7JLukFZqshGR60Rkn4gcFJFp/ZZExC8i3zeff0JE1lVgmZoio5UazUIxlJrqPS9UI/rTLgL+GjAKW6MStFE4ExFxA58DXgNsBW4Rka1Zm70XGFZKbQI+A3yivKvUlALrYkVXQGnmi9ftwqOVmrKiJxYWgUAN9Kl5/XkrCHjddmdhjc024KBS6hCAiHwPo7Hkbsc2NwD/ZN6+C/hPERGllHaYVjHWxcrB/gmSi+BPGfJ5WN4UmNPPiEgr8H1gHXAEuEkpNZxjuyTwgnn3mFLq+gUtVgNYRmEd1JQTHdQUgdaQl4DXVbUDLQG2rmhk64rGSi9jMbISOO64fwK4NN82SqmEiIwCbcDpsqxQUxKsga7v+8bOCq/E4NqtnXzpXd1z/TFrXM3tZur0Q8Df5thuSil1wQKXqMmiqc5LLKnTl+VEBzVF4B2XruWqzR14qjio0ZQeEbkNuA1gzZo1FV6NZjYu29DKl97VTTiWqPRSAFjeODeVxuQG4BXm7TuBh8kd1GhKwEfesJW4DmrKig5qikDI7+GsLq1y1CgngdWO+6vMx3Jtc0JEPEATMJj9Qou5K7ZmOh63i2u3Lmi6y2Kg0HE1ARHZCSSA25VSP8m1kQ7M50bn/AJRzQLQ0oJGMzM7gM0isl5EfMDNGI0lnTgbTb4V+JX202jKxfbt2wHOFpEXs/7d4NzO3Cfz7ZdrlVLdwNuBz4rIxlwbKaXuUEp1K6W6Ozo6ivlraDRFQSs1Gs0MmB6ZDwD3AW7gq0qpXSLyUWCnUupu4CvAN0XkIDCEEfhoNGXhgQceQER2mUFJBoWOq1FKnTT/PyQiDwMXAi+Vct0aTSnQQY1GMwtKqXuBe7Me+4jjdgS4sdzr0mgKwFIRbyfPuBoRaQHCSqmoiLQDVwKfLOsqNZoiodNPGo1GU7sUMq7mLGCniDwHPIThqdmd89U0mkWOVGPqX0QGgKM5nmqnOstoq3Hd1bhmmL7utUqpspsDZtiHoTo/W73m8pFr3WXfj/U+vGioxnWXbB+uyqAmHyKyM1deebFTjeuuxjVDday7GtaYjV5z+aiGdVfDGrOpxjVDda67lGvW6SeNRqPRaDQ1gQ5qNBqNRqPR1AS1FtTcUekFzJNqXHc1rhmqY93VsMZs9JrLRzWsuxrWmE01rhmqc90lW3NNeWo0Go1Go9EsXWpNqdFoNBqNRrNE0UGNRqPRaDSamqBmghoRuU5E9onIQRH5UKXXkwsRWS0iD4nIbhHZJSJ/bj7+TyJyUkSeNf+9ttJrzUZEjojIC+b6dpqPtYrI/SJywPy/pdLrtBCRMxyf57MiMiYif7GYP2u9D5cWvQ+XnmrYh6F692O9DxfwnrXgqRERN7AfuBY4gTGE8JbF1hXTnL3SpZR6WkQagKeANwI3ARNKqX+p5PpmQkSOAN1KqdOOxz4JDCmlbjcPYC1Kqb+t1BrzYe4fJ4FLgT9gEX7Weh8uPXofLi3Vsg9D9e7Heh+enVpRarYBB5VSh5RSMeB7wA2z/EzZUUr1KKWeNm+PA3uAlZVd1YK4AbjTvH0nxkFhMXIN8JJSKl/308WA3ocrg96Hi0dV7MNQc/ux3ocd1EpQsxI47rh/gkW+g4rIOoxJuE+YD31ARJ4Xka8uJvnQgQJ+KSJPicht5mOdSqke83Yv0FmZpc3KzcB3HfcX42et9+HSo/fh0lJ1+zBU3X6s9+FZqJWgpqoQkXrgv4G/UEqNAV8ANgIXAD3Av1ZudXl5mVLqIuA1wJ+IyMudTyojj7nocpki4gOuB35oPlQNn/WiR+/D5UPvw6WjCvdjvQ/PQq0ENSeB1Y77q8zHFh0i4sX4En1bKfUjAKVUn1IqqZRKAV/CkHEXFUqpk+b//cCPMdbYZ+amrRx1f+VWmJfXAE8rpfpgUX/Weh8uMXofLjlVsw9Dde7Heh+enVoJanYAm0VkvRkR3gzcXeE1TUNEBPgKsEcp9WnH412Ozd4EvFjutc2EiIRMMx0iEgJehbHGu4Fbzc1uBX5amRXOyC04JM9F/FnrfbiE6H24LFTFPgzVuR/rfbgwaqL6CcAsCfss4Aa+qpT6eGVXNB0ReRnwCPACkDIf/nuMP/gFGLLhEeAPHTnSiiMiGzCuCgA8wHeUUh8XkTbgB8Aa4Chwk1JqqELLnIb5xT8GbFBKjZqPfZNF+lnrfbh06H24PFTDPgzVuR/rfbjA96uVoEaj0Wg0Gs3SplbSTxqNRqPRaJY4OqjRaDQajUZTE+igRqPRaDQaTU2ggxqNRqPRaDQ1gQ5qNBqNRqPR1AQ6qNFoNBqNRlMT6KBGo9FoNBpNTaCDGo1Go9FoNDWBDmo0Go1Go9HUBDqo0Wg0Go1GUxPooEaj0Wg0Gk1N4Kn0AuZDe3u7WrduXaWXoakBnnrqqdNKqY5yv6/ehzXFpBL7sd6HNcWkWPtwVQY169atY+fOnZVehqYGEJGjlXhfvQ9rikkl9mO9D2uKSbH2YZ1+0mg0Go1GUxPooEajmQURuU5E9onIQRH5UI7n/SLyffP5J0RkXQWWqdFoNEseHdRoNDMgIm7gc8BrgK3ALSKyNWuz9wLDSqlNwGeAT5R3lRqNRqOBKvXUaObHSDiGIDQFvZVeSjWxDTiolDoEICLfA24Adju2uQH4J/P2XcB/iogopVQ5F6opLqmU4tkTI0RiyUovBYCWkI+zuhorvQzNIiWeTHFqZIq1baFKL6Wi6KBmCfFXP3wel8Ad7+qu9FKqiZXAccf9E8Cl+bZRSiVEZBRoA047NxKR24DbANasWVOq9WqKxJNHhrj5jscrvQyba7d28iX93dXkQCnF//7+szywp49nP/IqAl53pZdUMXRQs4Q4PRHF7ZJKL2PJopS6A7gDoLu7W6s4i5yhyRgAn3rreaxpDVZ4NYZSo9Hk4ifPnuR/nu8BoGc0wvr2pavWlDSoEZHrgH8D3MCXlVK3Zz3/QeB9QAIYAN6jlKpIie1SIJZI4XHroGaOnARWO+6vMh/Ltc0JEfEATcBgeZanKRWRuJF26l7XuqRPEprFTd9YhI/8ZBft9T5OT8Q4OTy1pPfXkhmFCzRYPgN0K6XOw/AifLJU69EYOddYIlXpZVQbO4DNIrJeRHzAzcDdWdvcDdxq3n4r8Cvtp6l+InHjuxLw6noKzeLl6aPDjEcTfOQNZwNwciRc4RVVllJ+W22DpVIqBlgGSxul1ENKKesv8DjGVbCmRMSTKeJJHdTMBaVUAvgAcB+wB/iBUmqXiHxURK43N/sK0CYiB4EPAtPKvjXVh6XUBDxL15+gWfyMRxIAnLeyCRE4OTxV4RVVllKmnwoxWDp5L/DzfE9qk+XCMdJP+qpzriil7gXuzXrsI47bEeDGcq9LU1oiCTOoqQHT5WxWAMd2b8FQzS9RSul2wVXAeNQIalpCPjobApwciVR4RZVlUZzhROT3gW7gU/m2UUrdoZTqVkp1d3SUfVRPTRBLKq3UaDQFYqWf/J5FcZicNwVaARCRBuDPgSfKu0LNQhiPxAGo93tY2VKn008lfO1CDJaIyHbgw8D1SqloCdezIIYnYwyb1RDVivbUaDSFE4kn8XtcuKq/YnBWK4DJxzAaR1bkUj+VUjx9bLgSbz0jyZTiv586QTSxOPoVZTMRSRD0uXG7hJXNdZwcMdJPz58Y4fjQ0gtwShnUzGqwFJELgS9iBDT9JVzLgvnru57nL3/4XKWXsSBiiRQxrdRoNAURiSdrIvVEbivASucGInIRsFopdc9MLyQit4nIThHZOTAwUNRF/uiZk7z587/jYP9EUV93oTz20iB/+cPn+Nbjx6Y9p5TiU/ft5ab/eozX/Nsj7OkZy/ka337iKE8eHpr2s5ZvayGMRxI0BAwnycqWOnpGIiSSKd79tR185oH9C379aqNkQU2BBstPAfXAD0XkWRHJripZNPSPR6regKWNwhpN4RhBTXWnngpBRFzAp4G/nG3bUtoAfrW3D4BTI4vrOLu31whUvvX4UVKpzKLGbz1xjM899BLRZIo9PWP8am/ua/N/f/AA39uRGRTd/dwpLv3/HlxwYDMRTVDvN4Oa5joSKcUjB04zNBlj0vTbLCVK2qemAIPl9lK+fzEJx5J27rIaSaUUiZRC1xlrNIURiadqRamZzQrQAJwDPCwiAMuBu0Xk+nKZhRPJFI8cMBpwn55YXC6EA32GcnT49CS/PXial28xgrljg2H++d49XLW5nW+8ZxtXffIhdudRapIpRTSeeUF5bDDM6FSc8UhiQfvZeDRBQ8AYfbOypQ6Au54+AbAk7Qa1fxlSJKZiSUbC1RvUWGmnZEqRTOnQRqOZjUg8WSvl3DNaAZRSo0qpdqXUOqXUOoz2GmULaACePjZilyZnBzU/2HGcHUeGcv1YWdjfP85Fa5ppC/n4xmPp3rD/92e7cIvwibech4iwtasxb/opkVKEY5mqyZSp0CxUqRmPxO3006pmI6i5f5ehei1Fu4EOagpkKp4kmkgVJQdaCZxpJ52C0mhmJ5JI1UT6qUArQEV5eF8/bpfgc7s4PZFZkPHJ+/ZlBBPlRCnFwb4Jzl7RxM3bVvPg3j6OD4XpH4/w0L5+br1iHSvMQOKsrkaOnJ5kKscA1GRS2UGMhVVdt+D0U5anBtLBTLY6tBTQs58KxIqyR8JxljdV39VbPKkct2tGVtdoSkYknsRfI9+T2awAWY+/ohxrcvLQvgEuXtvCyeEpTo9nKjXhWIKRcGUqT3tGI4xHE2zprOeVZ3XyhYdf4rtPHmNZg5+UgusvWGFve1ZXIykF+/rGuWB1c8brJFKKqawAI63ULCzwGI+kPTVBn4eWoJfhcBwRrdRo8pBKKXvHG5mqzrJuZ27VGeBoNJrcRGun+mnRopTie08eY0/PGK84o4P2Bj8DjvRTKqUIx5IMVyio2d83DsDmzgZWNtdxzVmdfH/HcX78zEnOXN7Als4Ge9utXY0A7D41PQWVTCkisWylxrifreDMlQmHpwbSas15K5uWpFKjg5oCiDj6E1Srr6YY6adTI1P8169fmlYBoNHUIpF4ikCVN95bzMSTKd7/raf40I9e4LINrbx92xo6zKGMFtYJf3iyMsddyyRsBS+/f9laBidjPHdilDecvyJj21UtddT7PTl9NYlUalrwYqWpFpJ+SqZURvUTwJZlDZzR2cCatpBWajS5CcfKE9QkkikeP1Sa4c7OnXu+jvifv9jL7T/fy4FF1kdCoykFkYRWakqFUop/vHsX9+3q4+9ecybfed9lNAd9tNf7M4zCk2bav5JKTXu9j9aQD4CrNrWzti0IwPVZQY3LJZy5vGFaUJNKKVJquiIzVQSlxvp8LE8NwP+94Wy+878uxe9x6eonTW6cxq/REqaf7nmhh5vveJxDA8UPGpw793yj9ynzC/TciZFiLEmjWdQslT41leDbTxzjO08c449esZE/vHqj3bW5vd7P0GTMVoPDUePYG44lK1Kksb9/gs3L0ikml0v48GvP4o9fsZHVrcFp25/V1cje3vEMNTupjNvZ6adiVD9NRKYHNQ0BL231fnwe16LtglxK9De2AJyRdCmVmkMDkwB2m+tiUoz0k6VYPXd8pBhL0mgWNTXUp2bR8cXfvMS2da389avOyHi8rd5HMqVsZWbSUQZd7tT/WCTOwb5xtnTWZzz+qrOX8zfXnZnzZ7auaGQimuDI4KT9mNVCY3r108KDGqsMvt7vnfacz+0iqpUaTS4y0k9TpftiWXM6+saK33wqI6hJzM8TY30pnz8xWpQ1aTSLmRoak7CoODUyxfGhKa47Z/m0uVrt9X4A21fjPPaWIwU1PBnjdy+d5vFDg7zpc48STaR49dnLC/75qza3A0aq3sIKahIplaGYWxmAXCXg2UxGExw+PTnt8YmocT5yKjUWfq9OP2nyMFUmT80xM6jpHy/+PLloUdJPxuewp2esavv1aDSFoJQimtBG4VLwxGHDN3jphtZpz6WDGuPCztnmvxwDhf/hpy/y9i89wc13PM7gZIxvvvdSrtjUXvDPr2oJcvHaFn723Cn7sYQjFeVUa+z0UwGBx+ceOsj2T/+aH+w8nvH4mKXU5ApqTKVGqaVV2KG/sQUwFU9/sUrpqbGDmpIoNZl9auaD9SVMpFTezpkaTS1gXQTUSp+axcQTh4ZoDHg4c3njtOc6GgxDrhXUZCo1pU8/PXd8hCs3tXHHOy/mvr94OZdvbJvza9xwwQr29o7bM6OcHdydF4N2SXcBSs3gRIxkSvE3dz3Pnb87Yj9ueWoacwQ1PjMgX2otPHRQUwDWF6sh4CmZUjMVS9JvNp0qhVITTxTHU2NVAWhfjaaWsU44Ov1UfJ48PMS29a24s1JPkFZqBsZzKDVFSj/96OkTOXvJjE7FOTE8xRUb23nV2cvpbAzM6/Vfe24Xbpdw97OGWpNITU85gaOjcAFm3slYgjWtQc5Z2ciPnkmP7ZqI5vfU+M0RH0vNLKyDmgKwdsQVTXUlC2pODIft26VQamJFMApH4knWtQVZ1uDXvhpNTWOdcHT1U3HpH4tw6PQk29ZPTz0BNNV58bolt6emCOmnSDzJ39z1PN98/Mi05/aa6vPWFdMVpLnQXu/nyk3t/Ox5I6hJ5kg/KZUem5BdFZWLyWiCpjov69pCjDt8ndaQ5VzpJ0upWWq+Gv2NLQBr5+tqDjBaIqOwlXpa2xakrxRKTUafmvnJkeFYkqDPw3mrmnVZt6amsZWa2hhouWh44rAxmPLS9bnTOiJCWyjdq8aqfvK6pSjpp909YyRSirGpRM7nAM7uWlhQA3D+qiZODhtVrInk9KAmnkwPFi5kTMJkLEnQ56axzptxDpqIJBCBkG/6fuq3gpol1oBPBzUFYF0tdDXVlWwGiRXUdK9tpX8sWnRzV3H61BjVIF1NgbLktzWaSmGlBHT6qXj8Zv8An7pvHw0BD2fPoIa0N/jSnppoEpfAsoZAUY69L5gK81hk+vFr96kx2kI+Ohr8C34fj8tFShmN9zI8NVbFUw7D8ExMml2Dm+q8jEXi9vlhzJz7JDI9lWcpNUttVIIOagrASj8tbwwwGUuWRM47NhQm5HNzVlcD0UTKdrUXi4z00zzXPxU3rhbqfG57wKdGU4vo9FPxGI/E+eAPnuVdX30St0v40ru68bjzf67t9X4GJ9J9akI+D60hH0NFCGoshTnX8XVP7xhbVzTmDBDmisdtvEYipXJWP+UyDM9EOJYk6PfQGPASd0z8nogmaAxM99OAI/20xJQaPaW7AKbMzqKtIWPnGZ2KFyWad3J8KMzq1iDLTHNa/1iEprrcO+t8KIZReCqWpM7rJuB1E4mnSKXUtD4TGk0toI3CC+fBPX38Zv8A9+/uo3cswp++chMfeOUm28Caj/Z6P/t6jUGS4WiSoN9Nszl5eqFYXsDxLKUmnkyxv3eCP7hy3YLfA8DjsoKaVIZSE87Rm6ZwpcZtnxPGphIEfR7GI/GMuU9OrM95qXlqdFBTAOGYsQM1BY3Kn9GpWNGDmmNDYda1hVhmvm7/eJTNjgmwC6UYJd3hWII6n5ugz3LVp6jLkcvVaKqddFCjlZq5kkim+Oj/7OYbjx0l6HNz4Zpm/uPtF3Hx2paCft6a/6SUMpQav6HUWCn6+TIRTfCSOYIm21NzaGCSWDLFWUXw0wB2ZVciK/2Ua95TIUrNZNQ4BzXWGafs0ak4y5sC5oTu3KdxO/2kq5802UzFUtR53TSbUXKxK6CUUhwbCrOmNWiXEfaNFdcsnDHQcp59CyJxI4ipM69eaz0FJSKtInK/iBww/592VBaRC0TkMRHZJSLPi8jbKrFWTXGx0k+zqQqaTGKJFO+9cyffeOwot718A8//46v49vsuKzigAWgL+YgnFWORBOFYkpDPQ0vQx9As1U+plOJv7nou71DgF0+OohSc0dkwzVOzu8dQcBZa+WRhKzVJlbNPjRXUuF0yq1E4lVKE40lCPodSY65/IpLIWfkExpgEYMmNStBBTQFMxQ2Fojlo7FDD4TgP7+svqGlSIQxMRInEU6wxy6UBu2dNsYgtMP2USKaIJVMEvemgZiHTZauEDwEPKqU2Aw+a97MJA+9SSp0NXAd8VkSay7dETSmIaqPwvPB5XGzsqOf2N5/L37/2rBm9M/mwemENTcaYiCYImsfe8UiCxAzHrqeODfODnSd4aF9/zuctk/AVm9qIJVIZCsnRQUMFWt8emvN6c2H93olUKmefGssw3FznnfU4OhVPohSETE8NwKh5YT0eSdCQx1Pj9+qSbk0ewmY5XXOd8WX79hNHeffXdnDX0yeK8vrDk8YO2hbyE/J7CPncRe9VE0+msPxv8zEKW1+8OtMoDIV1wqxybgDuNG/fCbwxewOl1H6l1AHz9imgH+go1wI1pUGnn+bPR96wlZu3rZn3z7fWW0FNlLAj/QQzz96zmt05G/Y5ef7kKCub69hgBi7jDrPwRMQInrzzCMJyYSk1yTzpJ6u6riXkmzX9ZJW1B83qJ0grNeNmVVQutFKjyYtVytxkKjUP7xsA0s2aFkr6qtD4c3Q2BoreqyaWMFJoIvNTaqwAxpl+WgJKTadSqse83Qt0zrSxiGwDfMBLeZ6/TUR2isjOgYGB4q5UU1TS1U9aqSk3bWYAMzgRM4zCPjfNpp8xXwO+RDLFvS8YX9VwNPdx6fDpCTZ31tOYFRiA4bfJFxzMB1upSeaufpqKGftXS9A7a1Bj/T4hs08NYPeqGY/E83pqAlqpKT4icp2I7BORgyIyTboXkZeLyNMikhCRt5ZyLQvBKmVu8HsQARHobPRzoG+iKK9vz5kx8/cdDX4GSqDU+DwuvG7XvDw1tlLjTRuFa0Sp2SIiL+b4d4NzI2U0hsj7wYlIF/BN4A+UUjmPIkqpO5RS3Uqp7o4OLeYsZiKO/V1TXpzpJ6uku8WR+s/Foy8NMmgGPBN5lJqekQgrmuvsIMCp1IxH83tT5oMnj1E4u09NS9A3q6fG+n2M9JOxxrGpBPFkikg8RUNepSZd0LGUKFn1k4i4gc8B1wIngB0icrdSardjs2PAu4G/KtU6ikE4lmRVixuXS9i8rJ4rNrYTS6a45/kelFIL7mtgNUeycqDLGgM8X+SOvbGkwut24XOreUXuVili0OcmYAY14dpQavYrpbpzPSEifSLSpZTqMYOWnMl6EWkE7gE+rJR6vIRr1ZSJKV3SXTHaQoavcHAyZpd0t1hKjdmr5qmjQ3z2gQMEfW5CPg+7e8ZoDHhY1x6y0zVOIvEkg5MxVjQFbF/KWFZn3nzBwXyw+9QkUxlKTTgrqGkN+UzPTP7ziPUzIZ8Hj9tFyOdmdCpuD7PMaxTWYxKKzjbgoFLqkFIqBnwPw6Ngo5Q6opR6HljUn7rRn8XYce79s6v4xzds5YzOBkan4kUx9FrpJ6ut9bIGf9E9NbFECp/bhdct80s/OQ7y1tVrITNLqpy7gVvN27cCP83eQER8wI+Bbyil7irj2jQlJBJP4XVLzqGLmtJitY3IUGpM9aZ/LEI8meJD//0Cz58Y5cjpME8eGWIkHOc9L1tPc9DHRI70U8+okc7vaqrLn34qmVLjMApnzXtqDvpIptSMk7QnbaXGOO5aXYWtAM8qYMnGHpOwxEq6S9mnZiVw3HH/BHDpfF9MRG4DbgNYs2b+JrT5MBVPUuczdhArV7rF7CGzr3d83tNcLaxI2oqsm0xHfCKZmlf1QC6s9FM86VqQpybo86TTT7Wh1MzE7cAPROS9wFHgJgAR6Qber5R6n/nYy4E2EXm3+XPvVko9W/7laopFJJ7Uc58qSGvIx6mRKVIKgn43nQ1+1reH+MwDBzgyGOZA/wRfelc3127NtLn98bef4tTI1LTX6zEf62oO5Ew/TUQStNcHi7Z+j8vhqUnmL+m20mqRRBKfx0UqpXjrf/2O157bxfuu2gCkjcIhU0my5j+dNH+nFU11OdeQ7lOzqDWDolM1RuFK+hGs5ntOtnTWA7C/b3zBr5/tqbF23sk8hrf5EEsYV56Gp2YBRmGvs09NbQc1SqlBpdQ1SqnNSqntSqkh8/GdZkCDUupbSimvUuoCx79nK7pwzYKJJpL4ayj1VIC/8YMistvstfSgiKytxDot2kI+jg8bZdZW2uWr774EgK/89jBXbW5n+1nLpv1cyOfJWf10ylRqVjTV5U4/RRPU+4vXwd3tnt5RuN7vyWi+53WLfay3lJvesQhPHxvh/92zh58+exJIKzXWxWRjnZexqbg9MHNly8xBjU4/FY+TwGrH/VXmY1VFKqWMpnNZB7i2en9GO++FkJ1+snK749HiNfmzlBpDrZm7UTjsKOm2PDWFdMLUaKqRSDxVM+XcDn/ja4CtwC0isjVrs2eAbqXUecBdwCfLu8pMWkM+jg8ZJ23rZL6+PcRX330Jl29o45+uPzunByXk9+Q0CltKzfKmAEGfG7dLMtJPM1URzQdn+inhDGocYxIyUvmmr/LI6UnAKET56x8+z8H+Cfvi1qrOagyklRq3S1ieJ1PgcQkuWXqzn0r5rd0BbBaR9abv4GYMj0JVYfUTyDUO4Izl9UVWaow/h5Xbzefid9I/HuFzDx3kLV/4HV9/9HDe7WLJFF7LUzOPyD2So6S71pUazdIlEk/Wkkm4EH/jQ0opaw7B4xgXoRWjNeS3y5ZDDgPvBaub+e5tl7Gxoz7nz9X7DaXGmmJtcWp0ivZ6HwGvGxGhIeCx00/GOIZkcUu6HemnlLmW+oCHKTN4icTTc/QgnY46ZAY1/3rjBcSSKZ49PmJ3breyBU11RiPCk8NTLG8M5LUoiAg+j0unn4qFUioBfAC4D9gD/EAptUtEPioi1wOIyCUicgK4EfiiiOwq1Xrmi7PqJ5stnQ3s75sglZrf2AGLdPVTdvpp9qDmY/+zh0/dt4/nT4zw4N7cnTTBSj9ZSs18qp/ML5bXbQdHS8BTo1miRMwhtjVCLn/jyhm2fy/w81xPlKvXUpvZgA9yH3vzEfJ7SCmmlUmfGonQ5fCeNAa8dvopEjdSREU1CjvST5anpt7vsdXtSDxpNjJ12ffBUGoCXhcXrmkGYGA8ykTUSFVZ6aTGOg+jU3FOjEyxsjl36snC73EvufRTSQdaKqXuBe7Neuwjjts7qPAVwWw4vSTZnNHZwFQ8yYnhKda0zd9kZsmDtlLjn25ky8fuU6Ncu7WTeDLF4ET+2SjxZIqQ30M8OU9PjXmQsBSrgNddK31qNJppROKpJWkUFpHfB7qBq3M9r5S6A7gDoLu7e2FXczNg9aqBTKVmNurNCqGJaCJDXe8ZnWJdW3oEQkPAw5h5fLXS/HN5n9nI1aemIeCh1/T2TOVRag6fnmRdW4iQ30O930P/eIRkSmWsranOy0Q0wfGhMJdtaJtxHYZSs7SO0zVzKVIqnOMBslnVYgQyPaPT3fZzIRpPIpL+ItQXaBSOxJMcGQxz5vIGWmcZ+BY3+9R43a55Re5TsQQi6cAr6NNBjSaTx14a5PgCJykvFiKJmko/FeRvFJHtwIeB65VSxe0pMUecQc1clRqYrnJbjfcsGgNexh1DIYHi9qkx00/JZKanxlK8p+IpAo6gxlJqDg9O2vOnOhr8plJjlLU71w5GmfpsSo3PrdNPmixmSj+1hGbuclko0UQKv8dlG9/SnpqZX/elgQmSKcUZyxtoDvoYCecPaqzqJ597niXd5pWFtcY6r1unnzQZ/Ml3nuY9X99RE1eGtWQUpgB/o4hcCHwRI6DJn8cuE+2O9FPIV3iwYflOnH7EsUic8WiCrqa0obaxzsPYVCJj2+KOSXBWP6Xs17fSYpFYMqOSNGK28Dg2GJ4W1ISjSbtHDWDPf4L8lU8Wfu/8LmKrmZr51paKdPpp+g5vXU0MzxBMFIIR1KR3WuvLlauJlBPLpHxGZwMtQS+TsWTeHdiofnKbzffmUf1kDvW0qPN5dFCjsYklUgxNxjjQP8HnH8o5+qqqiMZrp6S7EH8j8CmgHvihiDwrIhUt6mg1uwqD0aemUOodSk0knuSRAwP0jJiN9xyqRkPAa1c/zdaZdz7krH4KeIglUySSKbP3WWb66eTIFImUYl1WUDOZ1VKk0RnUaKVmGiX11NQCU3Fjh8+VfrJad8+U9imEaCJpp3XAGFwG6S9bPvb2juNzu1jXHqLZmmIbjrEsR4mfUf1k9KmZr1LjlOPrvC6dftLYWIF9U52Xzz98kNed12U3qKxGaq35XgH+xu1lX9QMtIXmp9RYisZkLMG9L/TwwR88x6vMBn0rnEpNwGt7FsdLotSkq59sT43VkyaRspVvZ0m3VflkTRHvqPfzm/EozUHv/JUaj1ZqNFnMlH6y+gzMlPYphGgiZc99AuMLUed155xh4mR/7zgbOkJ43S5a7dkouVNW9pgEz/yb7znN0nU+nX7SpLFM6n973Zm4XcKdvztS2QUtkEiiptJPVYfTUzOXoaJOldsajfDL3X0AmZ6aOqOfTSKZSntqyqDUgHEstfrUWPvYVCxp96ixlJpljX7GowkGJ2OZnpq69G1d/TQd/a2dhZmqn8D48g1NLtxT48vqNRDye2atftrXO86Zy42rYavddj7VyG6+twClJiP95PXoPjUam8FJw1e6ubOeV21dzv8831PVB9Ma61NTdQR9bvweF0GfMUi4UJxG4dMTUfweF60hHy4xZupZNJhm24loorSemqSzo7A5EiGeNEu6XWmjcCLJ4dOTNAQ8tkrVUW+s98Tw1LTqJ8DuuzMTuvpJMw1LjcjnwG8OehfuqYlnemrAuGqYqfneWCTOqdEIW8ygpjmYTj/lIm5P6XYRT8zPUxPIUmp0R2GNhaXUtIZ8vOmilYxOxXloX8X9pvNCKVVrfWqqDhGhLeSbNp5mNjKDmhjLmwJ88Z0X86HXnJnRpK7RMf/JDmqKqNS4nUpNMkupiSft9JNRIGIYhw+fNiqfrGIMy0ZglHSnj71W9dNsKg2Y6SfdUVjjJBzLX9INllJTBE+NN1upcc/YfG9/b9okDLNXYlnN97ye+U3pjkxTarSnRpPm9ISh1LSH/Fy1qZ22kI+fPFN1U1EA4wIgpagpT0010lrvyziZF4LtR4wmGJyI0l7v55J1rdz28o0Z21lm29GpOOORBD63a9qF5UJIdxROkTQ7CltrC8eSGdWkAY+bSCLF/r7xjE7JllJj/Gw64Ar63HhcMqufBgylppoV0/mgg5pZsE7c+Q5wLbOUUheCVdLtpN7vmdEovM+qfLLTT/krsZRSxJIpfAsYaBmOJTMCu6Av3XNBoxmajOFxCY11xvDBN5y/ggf39NuThBcz/WMR/uqHz/GJX+wF0qNRdPqpsixvDNBcN7chkx63i4DXZaefnKXhTiz/zFgkzkQ0XlSVxliHs/leCo9L7OPn6FQcpdId5Ot8bo6cnqRvLMq5K5vs1+hwpMucipWIcOmGVi6fpfEesCTHJOjqp1mwIup8ed2WoHfBSk0skcpwtIORfz01wwnhYP8EdV63PXbeMi0P51iLZVTzeVz4EvNtvpfMKGsPeN3TWpFrli6DEzFaQz5bOn/HpWv4wc7jvP7fH+Ffbjyfa87qrPAKc/O7g6f5w28+ZVfAvGxTu30y0emnyvKR1589Lz9Ivd/DRDTJ6YkY3etac25jpXDGIwkmIomi+mlgulHY7RLbl2kdo637AY+LnUeHAThvVTqoaQ35cLtkWvoJ4Nvvu6ygdejqJ800JqKJGSXQlpCPsYjhop8v0UTKnuthUe93z+ipOTQwyYaOUEaw1RL05kw/WTu11VF43s33fOk11nndds8FjWZwMkqbQy7f3NnA//zpy+hqquN/fWMnjx8arODq8vP13x2hzufmF39xFWvbgvzDT17k/d98ioaAh5dv6aj08pY0a9qCbJ5HW4CQ38NYJM5wOEa7Y590Yl1EjobjTERLEdSYHYVTimRSZSg11kWwPXLG52ZoMoZLYOuKRvs13C6xTcPzHeGwFJUaHdTMwngkYTvlc2GVHo5Mzb8CKrtPDRimspk8NS8NTLAha1Jtvq7CVhBjBTUphe3IL5SpWDJDArX8NbqsWwNweiI2Terf0FHPD95/OWvbQvz5955ZsKJZbJRS7Dw6zFWbOzhzeSMfveEcDp+e5OTIFF999yWsdcwK0lQPIZ+HE0NhlIKOPOmnZY1+3C7h6NAk45FE8dNPllKTzKHUhLOVGuP/zcsaphmjLdVwLqMinPjcuqRbk8V4JD5j/wKr6ihX2qdQclU/hfweWxLPJmJ2n9zYkXnQbQ35cnpqLA+Nz2MYhYE5qTWplJrWfC+ggxqNg6HJWEbDNIt6v4f/fPuFDE/G+bPvPjOtYk4pxa5To9z93CnbbFwuXhqYYGgyxrb1LQBcvaWDj73xHL7xnm1ckidtoVn81Ps9HDVnkLXlUWr8Hjfr20Ps75tgIpoo6twnAJdLcIk1JkHZvcfA2O8g7dmyFJtzHH4aC6sMfb5K0lIck6A9NbMwNjVzUNNahK7C2c33wOg+GUukjKZ5WSrOkcFJlCKHUpPbh2Pt1D6zpBuMQKdQI6QlXzqvFoJWe29dAaUBBieiGa3tnZy9oomPv+kc/ua/n+edX3mCz7ztAprqvHx/x3G+9ugR20zsEnjbJWv45zefW5Y1P3nY8DE4A5h3Xra2LO+tKR0hv5sRMw2fL/0EsKWznt2nxlAUt5zbwuNyZXhq2uv9XLy2hXtf6AXSwYwV7Dj9NBZppWae6SezMCSVUnPq91PN6KBmFsYjCTpzjB2wSJdSzz+oieVIPzn7Lfg8mVfAL/VnttO21xL0MZQz/WSkmrweo/oJID6H6N2qcsruKAxaqVksKLNs1DLqlpOpWJLJWJK2PFI/wI3dq6nzufnf33+Wl33iIfvxyza08ufbN7NpWT1ffuQQ333yGH/1qi15r7CLyY4jQ7TX++wBgprawOk/yVf9BEa65+cv9hLyeYruqQGjAspovpfCLYLLJXzrvZfyV3c9xz3P99gNUy1D+rk5gpplDca5ZyFKDZgXsa6lUc2ng5pZMDw1+T+mllnGExRCbqNwetpsS5asf8iULzdkpZ9aQj5Gp+IkzSsDCyvV5HO78XlS5mMze2p2nxqjNeRjeVPADlzqMvrUaKWm0vzupdPs6x1nf984D+8boL3ez8/+9GVlX4fVTXimEwjA689bwcaOenYeHWZ4MsYVG9syqlPec+V67n2hlx1HhrnunOUlXTPAk4eH2La+tSKBoKZ0OAOA9ob8wfEZyxtQyjjGlkKpcbskQ6kB4xj6HzdfyJ+9cjNbOg2lPeB143YJW7sap72GrdTMsV+PxXyU+WpHBzWzYHhq8huFFzrUUik1bUo3pPso5KqAOnR6khVNgWmSZEvQi1JGyswZCKWrnxxKzQyemn2947zp849y/fkr+NSN5+ccFWErNTqoqQj3vtDDH3/7acBIVbaEfLxwcpR4MmX/jcuFte+35Uk/OTmrq5Gzchy8wbhS9Xtc7DgyVPSgRinFrV/bwcs3t/O+qzZwamSKkyNTvO+q9UV9H03lsZQan9s1o1fGCiqAontqwCjMsAZaWn1rwPDbWP3FAK7c1E6935Mz6LjmrGXs7xtnbWtwXmuwMgDReAryJxxqCh3UzEAypZiMJWdUaup8+fvDFIJl4s2XfsoV1OSqfILMBnwZQY1V/eRx4TW/XPka8E3Fkvzpd58mmkjZ6lOuURG2UqPTT2UnmVL86y/3sXlZPd+77TJaQz6+8+QxPvzjFxk0W8OXE3tEwixKzWz4PW4uXNPMk4eHirGsDPrGovxm/wC/2T9A72iEF0+NAnDp+tkbmGmqC+vY2V7vm1GFW9sWwusW4kk175Lpmcil1OTilm1ruGXbmpzPrWoJ8vE3zd9jZl0sL6VRCbr6aQbS01tn7mqZrz9MIVgm3FwdhWF6UKOUsnvUZNMczByVEEukeO74iO2f8TuNwnk8Nf/24AH2903QEvQyETVeJ9eoiDpHy29NefnJMyd5aWCSD15reE9ExG6p3j8eKft6nCMSFsq2da3sOjU6Y4+m+XCw30jZnrOykS//9jB7e8f52BvPyegLoqkN6s1UzUypJzCUFGssQUk8NS7DU5NKKbvEu9z4bKVm6RyndVAzA2MR46Q+20j6ljyl1IVgBRf+LOnRDmqyRiUMjEeZiCYyZoRYWD1zLNXonhdO8cbPP8phc6S9odTMnH7accTwGVywupnJqPFFsKaFN/jTwd1SUGpEpFVE7heRA+b/LTNs2ygiJ0TkP0u5ppFwjM88sJ9zVjZmpGis4Xf9Y+UtiwYYtNJPC1RqALatbyOl4Cmzw2qxsMpov/Subu5458X8+q9+T1c61ShppWb2INtq7jfbMX4+eNxGN2BDqanMqda6WNZKTQ2QSilSc2wwl411Mm+cZYdfyFDLqENFcWIZ17Ib8FlXnLmUmuz5T31jUZSC3T1jgNl8zzNzUHN8KMza1iD1Aa99tWwpNs4vvqXU1Pik7g8BDyqlNgMPmvfz8THgN6VczHgkzq1ffZL+sSj/+IazM6R1q59F/3j5g5qhyRh+j2veDcKcXLimGbdLePJwcTsQH+yfoMHvYXljgFedvZym4NxmCmmqB+uCMFffpGy2LLOUmuLvDx6Xi3jK9NRUWKlZSr1qas5Tc3oiyhd//RI/fuYUZyyvL3hGRi7GbaVm5h2+OejjmNnsaa5YsmB2n5p86adnjo8AcM6K6eV/VvrJ6tEwanY53mdO9Pa5HZ6axPSALxJP0j8eZVVLkN6xKfu9reDOWSFgKTU1nn66AXiFeftO4GHgb7M3EpGLgU7gF0B3Md54eDLGfz99gkcPniYSTzERTXByZIqxqTj/9fsXT2sO117h9FO7mQZbKCG/hzOXN/DCybEirCzNwf4JNi6r15VOSwBrovVs6SdIl1Evayx+CwGPS0imUrN6akqJnX7SQU3xEJHrgH8D3MCXlVK3Zz3vB74BXAwMAm9TSh2Zz3ulUoo/+tZTPHNshA0dIR49OMje3jHOXD6/vLmddplNqQl6520Uzuepsb6Y41npp2eOjbChPTStzBuMQMjrFlupsYKa/eZEb59HbE9NLqXGaoK2urWOiWjcTn3ZQY1/elBT49VPnUqpHvN2L0bgkoGIuIB/BX4f2D7Ti4nIbcBtAGvWTDcGTkQT/Oy5U9y/u4/fHjxNLJFiS2c9zXU+2ut9bOls4PoLVnB1jplEPo+L1pCvIkqNNcyyWDTVeZkq8gT4gwMTOT83Te0xl/TT1Vs6+PmfX8WWecyYmg23S8zqp1TFlBrbKKyDmuIgIm7gc8C1wAlgh4jcrZTa7djsvcCwUmqTiNwMfAJ423ze75uPH2XHkWE++dbz2H5WJ5f+fw/w/R3H+cc3nD2v9Y9HC1dqrKGWnjmW06aDmkzp3uUSQj53RvpJKcUzx4Z5xRnLcr6WiNBU57ONwlZQY933ul125J4rqDkxbAQ1q1qCHBsKMxVPkkwpJqIJfG5XRsmhyyX4Pa5p6addp0b54Pef4yNv2MqVm9oL/yAqxxYReTHH4x923lFKKRHJlc/8Y+BepdSJ2VQApdQdwB0A3d3d015rKpbk73/8Aiub6/j9S9dyY/eqvOXPuVjW4GegAkHNwHi0qBVXPo9rmkJ5sH+cD/33C3zpXd05A/qZGJ2KMzAeZdOy6T40Te1hXYR2FKDUiMicvmNzwes2OwonK6/U6KCmeGwDDiqlDgGIyPcwJH1nUHMD8E/m7buA/xQRUVaL1AI5PhTmE7/Yy8u3dHDjxasQEV519nJ+/MxJ/va6M+fVeKhQpcZ6PhxP0jjXoMZKP3mm/1zI78k4uB8bCjM4GeOitc15X68l6LWHWo5lDdn0zWIUPm6m0Fa31lF/Ip3+Go/EczanqvO5p6WfHjlwmn1947z7a0/yF9u30FXm8uJ8nLuyKd/E3/1KqZwpIxHpE5EupVSPiHQB/Tk2uxy4SkT+GKgHfCIyoZSayX+Tk44GPw/95StY2xacV5qko8E/b6UmmVLc+0IPrzln+ZwD84GJaM4W7/PF554+r+ae53vZeXSYJ48M8eqz59bDxvKhbcphrtfUHlu7GvmnN2xl+1m5L/7KhVXSnUypac1Vy4WlzEcTNa2oZ1DqoGYlcNxx/wRwab5tlFIJERkF2oDTzo1mk+739o5T7/fwz28+1z4h3HLJGu55voefPnuSt12yhqHJGA/u6aN7XWtBrdELDWqsJnjhaJLGWVSdbJzDJrOpD2QGNU8fMypCLlydtwiHlqBvWvrJwprSbbzv9JjxxPAUXrewrCFgp5omowkmIomcJY9Br3ta9dPRwUma6rycs7KRT923L+86y80/vO6sfEHNTNwN3Arcbv7/0+wNlFLvsG6LyLuB7vkENBbrFtCyv6PBz0vmCXyuPPbSIH/63Wf49E3n8+aLVpnqRoRNy2b+zJIpxeBEtKCr4kLxeaYHNTuPGr1rdp0am3NQY30mWqlZGrhcwruvrHxTRa85JiGRUtRVKv3k1Z6aRcts0v21Wzu5anN7hiJzxcY2zlzewN/+9wvct6uPHUeG7EDlyk1tfOXWS2ZUcMYicXwe17TUUDYhsy/C5Dx8ANF47vQTGB4WZ1DzzLERQj53RjfKbJqDXtu0PDoVzzhBeGfpU3N8OMzK5jrcLrGVmYlowmgjniOoCfjc0zw1h09PsrEjxDffcyknR6ZIzU1wKxnWNPU5cjvwAxF5L3AUuAlARLqB9yul3le8FS6cZQ0BBiaiKKXmrPQcGTTK/n/+Yi9vvmgVf3PXc9y3q4/utS2saw+xt3cMtwjLmwL84xvOZkVzHWCMSEipdPVVMfB5XBkH4UQyxdNmiffuU3M3EB8cmMDndrGqpa5oa6w2yult1Bg4lZqKVT/N0pesFil1UHMSWO24v8p8LNc2J0TEAzRhfKnmTHaA4nIJd/3RFXzh4YN8/dEjdK9r5U9+bxO/e+k0n33gAD/ceZx3Xr4u7+uNRxKzlnNDplIzV2xPjTeHUuP3ZBiFnz42zPmrm2fMz7YEfTx3YgQwgpqzVzTyzDHjvt/jwusxfjafp2ZVi9GO29nReCzP/Kugzz0tkDs6GObyDW24XMLqebb2XiwopQaBa3I8vhOYFtAopb4OfL3kC8vDsgY/8aRiOBy3jbvO6byplGIqnszZPdVKPf5m/wAH+8e5f3cfV25q4+TwFEcGJzmrq5FkSnHfrj6u3rKMt19qqKWWh6eYSo3f48roq7G3d5zJWJKQz82enrkHNS/1T7C+PTTntFqtUG5vo8bA43IRjiUq26dGKzVFZwewWUTWYwQvNwNvz9rGkvgfA94K/GqufpqZqPd7+OtXn8lfv/pM+7FL1rXw2wOn+fzDL3HTJavzKjHGMMvZ00kh3wKUmkR+T83atiD/83wPqZQikkiyp2ecP7p644yv12x2N06lFGNTcS5Y3WwHNc70U86gZijMq842CnwaHM3/JiIJVjRP98a0BH0ZVV+ReJKe0Qhr2/TU40pglaX2j0doDfkYCcd4+Scf4l9vuoBrt3by+YcP8tVHj/D4310zLd15fDiMxyVEEyk+8J1nUMAn3nKeHeSCkWo68//8nKNDk/Zj/SUJatwZV5ZWI763XLyKbzx2lOHJ2JzMwseGwkt9EnfZvI2aNFbzvYpWP7mNc9PvXjq9KFRzr9uVdyREsShpUGN6ZD4A3Iche35VKbVLRD4K7FRK3Q18BfimiBwEhjACn5IiIvz59s288ytPctdTJ3jHpUZn0Z+/0MNFa1voNLuzGsMsC1BqzAAgPK+gJn/6qXttK9998jj7+8fpHY2QTCm2rW+dtp2T5qCPWCLFwISRFljZXEdHg5/BiShuV3qgZbYcGY4lGJyMTVNqJmdIP3XU+zk0kD7BWWmvde3VrdBUK8sa0l2Fz1wORwbDjEUSfPmRQ/zeGR188/GjDE3G2Ns7xnmrmjN+9vjQFJdvbGNPzxh7e8e55sxlGQENGHL66pYgxwbTPZkspcZ672JgpJ/SqueOI0N0NQW4dmsn33jsKHt6xrhiDpV1g5MxLpnle1PjlM3bqEnjcRlzpZIphdtdmaAm6HfTGvJx7wu93PtCb0XW4KQh4KnuoAZAKXUvcG/WYx9x3I4AN5Z6Hdm8bFM7F65p5rMPHODaszp54vAQf/rdZ/izazbzwWu3AEb1UCFBja3UzCP9ZAUXuYzCVoO1HUeGOXp6Ep/HNWtQ02I24DtijkZorPOyvi1kNxJM96nJjNrT5dyG78AKYsbN6qdcilV7gz/Dw2GNY1inlZqKYKklVqDRP2Y04nvi8BBfe/QIfeYIheeOj0wLao4NhTl/dRerWoJ898ljvOOy3AeeNW1BjuYIagrpCVIozuonpRQ7jwzTva6FrWbp7a5ThQc1yZRiOByjvYh9dJYys3kbNWk8Lpep1FTOU+N1u/jdh145rTN9pShH88uqMQoXGxHhn998Lm/+/O947507OWoaJQcn0iWx45FEQVegC1Nq8qefVrfWsazBz84jQ+zrHad7bcuspemWIdY68TTVedm4LMRBc/ZNvj41J4atcm7j6tw5e2oimshZ0t1eb6hC49EEjQGv/RnqoKYyZI9KGHDsy7f/Yi/t9X6UUjx7fJR3Xp7+ubFInNGpOGtag7xq63I6GvxcvSV3Oeza1iA7jwzbgezAeJQGvydj2OlC8XlcpJRhEB6cjNE7FuHitS201ftZ3hiwx35kMxFN8P0dx7mpe5UdhA+HYyhFUZsDViFl9TZqDNxuIV7hjsJgeE3n09KkWlmazjmTM5c38q83ns8LJ0dJpBRtWYMpx/MYZLNZiFKTrn6a/qcQES5Z38qv9w+wt3ecl22e/erUVmrMAKOpzsufXbOZz7/jIsBIIbhkelBzfChTqbHST0OTMeJJlTP9ZF2dnzZPokcGw7QEvXquToUI+T2EfG57VIKlomw/q5NkSvHWi1dxwepmnj8xglKKP/72U3z5kUPp/kQtQda1h/jgtVvyHoTXtIWYiCbsWWcD41E6itxi3ucYwme1JbBUqLNXNOatgPrPXx3kY/+zm0/fv99+bMgetln8NvhVhO1tFBEfRor/7qxtLG8jlMDbuBTxuqTiSs1SZEkHNQCvObeLf7/lQr78rm42dIQYnkz3dhmPxGmsm/0EbVc/zdNT43ZJ3sqMS9a22LOcrto0e5t3y0DpVGq6muq4bEObvY3X7Zo2tfX4UBi/x0WHefD3eYzuwz2jxgkyVxWYHdRMGCeOI6cntUm4wixrDNhKTf94lNaQj/dfvYHORj9v37aG81c3c3BggkcPDnLvC718+4ljjqaLs3uh1prbHDV/ZmA8au8zxcJZhhqxm1MaFw7nrWrmQP84hwYy+/H0jE7xtUcPE/S5+cZjRzlgjgY5bapVhQw3rFWUUgnA8jbuAX5geRtF5Hpzs68Abaa38YPMPLxVUwBul4tEsrJTupci+pMGrj9/BVdsaqfZ0bgumVJMxpIFKTVGp16Z13DHaCKZU6Wx6DZ9Nc1BL1tXzN7OuzlLqWnOoZr43C5bIbI4PhxmdWtmJ9sGv4feMUPByZ1+soIa48RxdDDMujZtEq4kyxr89I6mlZplDX6617XyxN9vZ01bkPNXN6MU/J+fGpMhDp+e5LcHDS9oQUGN+fe1zML94xGWNRa3a7RVhhpLpGwjfcB87O2XrqHO6+ZffpnZ2PGz9x9AKfjebZcR9Ln52D17AK3UWCil7lVKbVFKbVRKfdx87CNmsQZKqYhS6kal1Cal1DarUkozf7xuIZFKGUZhfaYtG/qjdtDqCGom7G7ChaVSgj7PvIKaWCI1Ywvts7oaaQh4uHJTe0F52ea66UrNtLX63dNUpeNDU6zOak4W8nvsE2S9P5dR2Hiv0xNRIvEkp0anFtQRV7Nw1rQGbeVlYHx6p9/zzXEGh09PcpWZzvzx0ydpDHhy7ivZWIGPtX+VUqmJ5lBqOhr8/K+Xb+DeF3p55li6Id8PnzrO71+2lvNWNXPbVRv4zf4B+scidlCzxD01mgrgNtNPiWQKj1Zqyob+pB20hHwMT8ZRSjEWsYZZFualzh4+WSjRRGpGpcbtEr7zvsv4yOu3FvR6Po/L7kTsdYs9TTtjrX7PNP+PpdQ4qc8IaqZ/Dq1BHyKGp+b4UBiltEm40qxpDdI/HmUqlswZcDQHfbba8vevPYu1bUEmY0nWFKiwBbxuljcGODo0yWQ0wWQsWdQeNZD21EQTKVtRDDiaU77vqg201/v4h5+8yOhUnH+8+0Wagz7+/JrNAJy90lA0jw9P2anRFu3z0pQZr9uVLunWnpqyoYMaBy1BL7FkislY0u7kW0hHYTAqoCyl5sE9fbx4crSgnzOCmpmd6eeuarJ75xSCdcXdVOfNWULXkDV+YTQcZzySYHXL9KBm0vydcgV3HreL1qCPgYmYPTRwox4aWFGs4OTo0GReE+/rzu3iNecs56yuRl6xxfBpZf/tZ3uPY4NhR4+a4gY1fsdk4UgiU6kBY7/8xFvOY3/fOK/+zG/YcWSYv73uDNugbvXXOTkyxdBklJagd8l2E9ZUDlup0UbhsqK/6Q4sk+3wZMzu61Jo+inkGBnwf37yIl/8TWEp6dk8NfOhJWSsOZ/J2VBq0kHNcbucOzP95PTR5FOs2uv9nJ6IpoOaZVqpqSSWUfvFk2PEkqmcqaG/ue5MvvD7FwPwijOM0u25jLRY2xrk2FDYLhkvlVITS+ZWagCuOauTL7zjYoYmY1ywupkbL05XLK8051KdGA4zOBHTqSdNRTCa76W0UlNmlmyfmly0mD1ehsOxgid0WwR9Hnv201A4xkQkPstPGETjqZxznxaC9Xvk80iE/B6GJtMN1CwPRnYHWeeMoFzpJzB8Nacnohzon2Blc51dCaapDGvM4OQpc6r1bAHHZRvaOHdlE1dsbJtxOydr24wUl+WrKXpQY7Z2z6fUWGzf2skv/uIq2kJ+e74VGPttS9DLieEpBidjS94krKkM9pgEpZWacqLPQA5aQ1bDrjjj0TkqNX43PaMRpmJJIvFUwT1rYsmUbYwsFs2zBDXZ07+PD+cu6XUGMrmqn8BQap4+Nkw0nmJzp049VZqWoJcGv4cdRwwT7WzNI+t8bn72py+b03uca3Yj/n/37Dbfo8jpJ2f1Ux6lxmJDnnTnqpYgJ4enGJyIsqUz/1R7jaZUuF0uEill39aUB/1JO7CVmskYp0YMg2xngY3FrOonq3qq0OGW0fjsnpq5Ypki8ys1mabm40NTOatf6v3GunweV941ttf7GRiP8tLABJu0n6biiAhr2oJ2OrDYKgrA1Vs6+ORbziMcS+Jzu+zvTbFIVz8lZ5yNNhOrWuo4MRxmaDJGW71OP2nKj9ehzngqNPtpKaKVGgfWwXloMsbRwUmWNfgLTqdYgYJVQlpoJVQ0kZzTxOFCmF2p8WYoSbkqn6ztID2xOxft9X4i5tW0VmoWB2vbguwyu+6WIqgBuOmS1Zy3uomekUhG6qcY+JxG4Xj+MSIzsaqljl/t7SeWTNEa0uknTflxDrHUnpryoYMaB411XlwCI+EYRwbDdulrIVhKjdX9d6LA9NNsJd3zYTalpt7vJpZM2T1yjg+F2bxsukQfMpWamXxF7Y6r4E3LdFCzGLACVL/HVXD13nw4c3kjZy6fvSHkXMkwCieM9OxcA6dVLUFb5WnXSo2mAngdKSftqSkfOv3kwO0Smuq8DIVjHBsMs6a18Eoeq/ppyEw/FToyoZCS7rlSiFEYDDUplVIcH56aVvkE6WAmn58GjEndFps6tHdhMbDW3G87GvxlmYpbbLKb783HSG9VQIFuvKepDE51xlWF38NqRSs1WbSEfPSORugdi8yp5X/Q70Ep6BkxxgqEY0lSKTXrFeZsHYXnQ/OsnhpzAnc0YSs2a3Kkn6zt8lU+AXbJcEeDXw+yXCRYCmOpUk+lxtmnZr5B/6pWHdRoKovXrT01lUAHNVm0Bn08e9xonFdol1VIT+o+aQY1YJiFZ6ueKkWfmq4m44Cer2FfvSOomRwzFKVVOT01VlCT/3ewTpzaJLx4sALUYo8vKBdWEGNUPyXzVj7NhFOpaa/Sz0FT3TgrnrSnpnzo9FMWzUGfPaBxLi3/LUPxyeF0UFPILKhSVD+dsbyBn/zJlfZsn2zqHeknawr3iqbp6Sdru5l8Ga0hHy7RJuHFRFdTAK9b5tSFejGRMSZhnp6zhoDXViy1UqOpBE51RntqyodWarKwetUAczIKW6baE46gZiKaoHOGn1FKMTXPK9HZuGB1c97nnOknqwQ9V9lrfQGeGq/bxWdvvpALVuV/P0158bhdfO7tF3HG8ur0OGVXPwVyzC8rhFUtdYxOxYtecq7RFIIzkNF9asqHDmqysMqrGwMeuzS6EGylxpl+mqWsezgcJ5FSZfc+pJWaJIPmwL/mHP6bkG92Tw3A9eevKPIKNQvlVWcvr/QS5o3bJbhdQiyZXFB14KrmIKdGIlr611QE536nlZryocPHLKyrunXtc5thZCk1E9GErbzM1lXYmoBd7jRBeq1xhiZjNOcZ+NcYmHmGVK0jIq0icr+IHDD/b8mz3RoR+aWI7BGR3SKyrsxLrTl8bteClZr3vGw9H7ruzCKvTKMpDK9be2oqgQ5qsmg1g5pc1UAz4WzSZ5kUZ1Nq+sYrE9SkjcJJhsIx+3fOpino5TNvO5+3XLSqnMtbTHwIeFAptRl40Lyfi28An1JKnQVsA/rLtL6axe91Oaqf5neY2ra+lZsuWT37hhpNCdBKTWXQQU0WlrlwLiZhSKdqID0YcrZRCX2jcxvFUCycfWqGZpli/KYLV1VtaXARuAG407x9J/DG7A1EZCvgUUrdD6CUmlBKhbO308wNn9tFLLkwpUajqSRe3VG4IpQsqJmDdP8LERkRkf8p1VrmgnWCn0s5N0DQnz7wrmoxlJqJ2ZSaMaPKarahg8XG63bh97jssQ66OiQvnUqpHvN2L+T0fW8BRkTkRyLyjIh8SkT0WXiB+DwuovGFKTUaTSVxmoN1n5ryUcqjRaHS/aeAd5ZwHXPinJVN3LJtDa88c9mcfi6XUhOezVMzFqEt5Ct6871CsCZ1D4WXfFCzRURezPHvBudGSikFqBw/7wGuAv4KuATYALw71xuJyG0islNEdg4MDBT3t6gxfB4XUa3UaKoYr+4oXBFKeTadVboHUEo9CIyXcB1zIuB1889vPnfODbsCXhfWfruyQKWmfyxSsV4iITOoGdZKzX6l1Dk5/v0U6BORLgDz/1xemRPAs0qpQ0qpBPAT4KJcb6SUukMp1a2U6u7o6CjRr1MbWEZhrdRoqpVMT43eh8tFKT/pQqT7glnsV7kiYqs17SEfdV73rEbh3rFI2f00FiG/h56RCImUWupBzUzcDdxq3r4V+GmObXYAzSJiRSmvBHaXYW01jd+z8OonjaaSeHT1U0VYUFAjIg8sULovmGq4yg2aoxKagz5Cfg+Ts3QU7huLsrypMkpNvd/NsSHDz6qDmrzcDlwrIgeA7eZ9RKRbRL4MoJRKYqSeHhSRFwABvlSh9dYMfo+baGJhfWo0mkrirHjSnprysaDme0qp7fmeE5E+EelSSvXMIN3XFCG/B8ajtIS8hPwzKzXxZIrByWjZTcIW9X6PXVKug5rcKKUGgWtyPL4TeJ/j/v3AeWVcWs3j87gYmTIaQ/q1UjNvRKQV+D6wDjgC3KSUGs7a5gLgC0AjkAQ+rpT6flkXWoO4Xbr6qRKU8hKoEOm+prCUmpagj5DPM2NQMzAeRSkqptSEzKnioIMazeLD53ExNmV8f7RSsyAKKdgIA+9SSp0NXAd8VkSay7fE2sTZfE/3qSkfpTxazCrdm/cfAX4IXCMiJ0Tk1SVcU0kJ+TzUed0EvG7q/Z4Z+9T0jlWmR42Fc/SBDmo0iw2f28V4JA6gPTULY9aCDaXUfqXUAfP2KQxVfXHm+KsIrdRUhpLNfpqDdH9VqdZQboJ+Ny1m876g323PVcpF/1hluglbhBxBTVtoyTbX0yxSfB4XYxGt1BSBORVsiMg2wAe8lOf524DbANasWVPEZdYeXreufqoEeqBlEblkXSudpkcm5PdwbDB/Y9lKzX2ysIKagNdFnU9fCWsWFz6Pi2TKyI9qpWZmtm/fDnC2iLyY9dSHnXeUUkpE8hZsmN7HbwK3KqVSubZRSt0B3AHQ3d29oOKPWkcrNZVBBzVF5E9+b5N9u943c/qpbzyK1y155y6VmgYzqNEqjWYx4lRntFIzMw888AAisksp1Z39XKEFGyLSCNwDfFgp9XiJl7wk0J6ayqCPFiUi5PfMOKW7bzTCsoYArgrt7JZS0xJamhO4NYsbZ5dtrdQsiFkLNkTEB/wY+IZS6q4yrq2m0UpNZdBBTYkI+d1MxhIolVuh7RuvXOM9MNYH0KqVGs0ixKeVmmJRSMHGTcDLgXeLyLPmvwsqstoawuvSzfcqgU4/lQirZHoqniTom/4x94xGOKOzoQIrM6i300+68kmz+PC7tVJTDAop2FBKfQv4VpmXVvO4M4zCOqgpF/oSqERY6Z1c85+UUvSORuhqqiv3smzs9FOFPD0azUxkKDVefZjSVB8enX6qCPpoUSJCZkVRLl/NWCRBOJakq0KN98Ch1NTroEaz+Mjw1Hi0UqOpPjx6oGVF0J90ibCUkFxdha1y7kp1EwZoqjMMwh1znEau0ZQDvyOQ0UqNphrJMArr2U9lQ3tqSkT9DEFNz+gUQEWVmtWtQb7wjot4xRnLKrYGjSYfWqnRVDsigsclJFJKe2rKiA5qSoQ1BypXr5rFoNQAvObcroq+v0aTD59be2o01Y/bDGq0p6Z86KNFiUgrNdM9NT2jEUQq101Yo1nsZJZ0a6VGU51YDfjcooOacqGDmhIxm6emo96f0XFSo9GksYIar1v0Va6manG7BBEq1mR1KaLPqiXCCmoO9E9Me65nLFJRP41Gs9ixghrtp9FUM163aD9NmdFBTYloqvPy+vO6+MpvD/PNx45kPNc7OlVxP41Gs5ixughrP42mmnG7tNJYbvQRo4R8+qYL2H5WJ//np7t44cSo/XhPhRvvaTSLHTuo0UqNporxuFy6R02Z0Z92CfF5XPx/bz4HgCePDAFGh+HxSEIrNRrNDPjcRjCjlRpNNePRnrCyo48YJWZZQ4COBj+7T40BRuoJKtujRqNZ7GhPjaYWcLu0p6bc6KCmDGztamR3jxHU9Fg9anQ5t0aTF5/21GhqAK/LpZWaMqOPGGVg64pGDvSNE00k7aBGe2o0mvxopUZTC2ilpvzooKYMnL2ikURKcaBvwu4m3NmkZy5pNPnQ1U+aWsDrFj33qczoI0YZ2NrVCMDunjH29IzR0eDXVR1VgIi0isj9InLA/L8lz3afFJFdIrJHRP5dRLcPXShaqdHUAoZSo0+z5UR/2mVgXVuIoM/NPc/3cN+uXt584cpKL0lTGB8CHlRKbQYeNO9nICJXAFcC5wHnAJcAV5dzkbWINftJKzWaasbjdqGzT+WlZEeMQq5yReQCEXnMvMp9XkTeVqr1VBKXSzirq5Ff7x/A53Hxvqs2VHpJmsK4AbjTvH0n8MYc2yggAPgAP+AF+sqxuFrGCmq0UqOpZjxaqSk7pfy0Z73KBcLAu5RSZwPXAZ8VkeYSrqliWCmot29bS0eD9tNUCZ1KqR7zdi/Qmb2BUuox4CGgx/x3n1JqT64XE5HbRGSniOwcGBgo1ZprApdL8LpFKzWaqsbj1tVP5aaUR4xZr3KVUvuVUgfM26eAfqCjhGuqGC/b3E57vY8/vFqrNIuMLSLyYo5/Nzg3UkopDFUmAxHZBJwFrAJWAq+U/5+98w5z66zy/+dIGk3vzWN73FtsJ3FLHBJSwHEKkDi0EKphgcBCqLs/MMsuZIGwhl1g6WwgARNIQgghMZDqFAikuSdxi7s99ng8vTdJ7+8P3SvLM5qudqXzeZ55Rrr31dXRzKt7v/e8p4hcGumNjDG3G2NWGGNWlJen5DSPKgurCphXmZ9oMxRl3MyryGP+JJ3D8cQTw2OPeJcbjohcSNCFf3CI/TcDNwNMmzYtimbGh6sXTeKqhZVoDGnS8ZoxZkWkHSJSJyJVxphaEakiKLoH8lbgBWNMh/WaR4DXAc/GzOI04aFbXp9oExRlQvz7WxYm2oS0Y0KeGhHZNJG73LDjVAF3AR8yxgQijUmFu1wVNI5jI7DWerwWeCjCmGPA5SLiEZEMgkHCEZefFEVRlNgyIU+NMebKofaN8i4XESkA/gJ82RjzwkTsUZQosx64T0Q+DBwFbgQQkRXAx40xHwHuB94IvEJQuD9qjPlTguxVlLMQkRLgd8AM4AhwozGmeYixBcBu4EFjzC3xslFRokksY2pGvMsVES/wR+DXxpj7Y2iLoowZY0yjMWaVMWauMeZKY0yTtX2LJWgwxviNMR8zxpxjjFlojPl8Yq1WlLMYTcKGzdeBv8XFKkWJEbEUNeuB1SKyH7jSeo6IrBCRX1hjbgQuAz4oIjusnyUxtElRFCWdGE1ZAkRkOcG4x8fjY5aixIaYBQobYxqBVRG2bwHsu9zfAL+JlQ2KoihpzogJGyLiAr4DvI/gDeiQOD1hQ0l9JBjD6yxEpJ5gjMNAyoCGOJsTDZxotxNthsF2TzfGxD3yfJg5DM7826rN8WOg3fOAHGDvgHFfBjYYY4rsDSLSbIw5qxCqiNwC5Bhjvi0iHwRWjCamRudw0uBEuyPZHJVzcSxTumPGUB9cRLYMlZ6bzDjRbifaDMlj93Bf3mSxcSyozfFjLHaPMmHjdcClIvIJIA/wikiHMWa4+Budw0mCE+2Opc2OFDWKoijKqLATNtYzRMKGMea99uMwT82wgkZRkhWtQa4oipK6jCZhQ1FShlTz1NyeaAPGiRPtdqLN4Ay7nWDjQNTm+DFqu0eTsDFg+6+AX03ANhsn/m2daDM40+6Y2ezIQGFFURRFUZSB6PKToiiKoigpQcqIGhG5RkT2icgBEUnKIDcRqRaRp0Vkt4jsEpHPWNtvFZETYQUI35RoWwciIkdE5BXLvi3WthIReUJE9lu/i0c6TrwQkflhf88dItImIp9N5r+1zuHYonM49jhhDoNz57HO4VG8ZyosP4mIG3gNWA3UAJuBdxtjdifUsAFYKZVVxphtIpIPbCVY4fNGoMMY8z+JtG84ROQIwayIhrBt3waajDHrrRNYsTHmi4mycSis+XECWAl8iCT8W+scjj06h2OLU+YwOHce6xwemVTx1FwIHDDGHDLG9AH3EiwPnlQYY2qNMdusx+0EuzlPSaxVE2JUJdiTgFXAQWPMUIXCkgGdw4lB53D0cMQchpSbxzqHw0gVUTMFOB72vIYkn6AiMgNYCrxobbpFRF4WkTuTyX0YhgEeF5GtEiyVDqMowZ4k3ATcE/Y8Gf/WOodjj87h2OK4OQyOm8c6h0cgVUSNoxCRPOAPwGeNMW3AT4HZwBKglmAflmTj9caYZcC1wCdF5LLwnSa4jpl0a5kS7AR/PfB7a5MT/tZJj87h+KFzOHY4cB7rHB6BVBE1J4DqsOdTrW1Jh4hkEPwS/dYY8wCAMabOGOM3xgSAnxN04yYVxpgT1u/TwB8J2lhnrU3ba9SRSrAnmmuBbcaYOkjqv7XO4RijczjmOGYOgzPnsc7hkUkVUbMZmCsiMy1FeBPB8uBJhYgIcAewxxjz3bDtVWHD3gq8Gm/bhkNEcq1gOkQkF7iKoI12CXYYogR7EvBuwlyeSfy31jkcQ3QOxwVHzGFw5jzWOTw6UiL7CcBKCftfwA3caYy5LbEWDUZEXg88C7wCBKzN/0bwH76EoNvwCPCxsDXShCMiswjeFUCwCvXdxpjbRKQUuA+YRrBb743GmKYEmTkI64t/DJhljGm1tt1Fkv6tdQ7HDp3D8cEJcxicOY91Do/y/VJF1CiKoiiKkt6kyvKToiiKoihpjooaRVEURVFSAhU1iqIoiqKkBCpqFEVRFEVJCVTUKIqiKIqSEqioURRFURQlJVBRoyiKoihKSqCiRlEURVGUlEBFjaIoiqIoKYGKGkVRFEVRUgIVNYqiKIqipAQqahRFURRFSQk8iTZgPJSVlZkZM2Yk2gwlBdi6dWuDMaY83u+rc1iJJomYxzqHlWgSrTnsSFEzY8YMtmzZkmgzlBRARI4m4n11DivRJBHzWOewEk2iNYd1+UlRFEVRlJRARY2iKIqiKCmBihpFURRFUVICFTXKmNl9so2/vFybaDOUJMQYw6/+cZi6tp6I+w+cbudUa+R9ipJIjDE8+uopntl3OtGmKBNARY0yJIGA4ZsP7+G1uvaztv/4mQN86p5tHGvsSpBlSrKyu7aNW/+0m39/8NVB+7r7/LzjZ8/z9p8+R3tPfwKsU5TInGzp5m0/fY6P/2YrX/7j4LkbDx7acYIn99Ql5L1TCRU1ypCcaOnm9r8d4g9ba87afrSxk4CB//vbwQRZpiQrmw83AfDE7jqeO9hw1r4/7TxJS1c/J1q6+dqfdifCPEWJyN0vHmPn8RYunFnCqbYefP5A3G34/qb9/Mvvd6rgnyBRETUico2I7BORAyKyLsL+TBH5nbX/RRGZEbbvPBF5XkR2icgrIpIVDZuUiXOooROAfWGeGmMMRxu7cAn8fmsNp9t1KSFdCAQMd/z9MK3dQ590Nx9tZlJBFlOKsvnGn/fgDxggOG9+/cIR5lXm8YkrZvP7rTU8/IouYSrJwcmWbqoKs3nr0in4A4a69t6429Dc1UdLVz93/v1I3N87lZiwqBERN/Bj4FpgIfBuEVk4YNiHgWZjzBzge8C3rNd6gN8AHzfGLAKuAFSmJgmH6zsA2HfqjKhp7uqnvcfHe1dOx+cP8Mt/HEmQdUq82V3bxtf/vJuf/+1QxP3GGLYcaWLlrBK+eO0Cdte28cirQeGy/XgLr55o4/2vm8Fnr5zHkuoi/uW+nbx6ojWeH0FRInKytZuqwqAYBzjR3B3X9w8EDK3d/YjAL549REtXX1zfP5WIhqfmQuCAMeaQMaYPuBdYM2DMGmCD9fh+YJWICHAV8LIxZieAMabRGOOPgk1KFDhixczUtvaE7s6PNga9N5fPK2dJdRE7j7ckyjwlzjR0BO9e7918nD7fYPf88aZu6tp6WTGjhDefW8XMslx+/rdDGGP4ydMHycv08NalU/B6XNz+geWU5Hr58IbNoeMqSqKobe2hqiibKcWWqGmJb7xge4+PgIEbl1fT3uvj91tqRn6REpFoiJopwPGw5zXWtohjjDE+oBUoBeYBRkQeE5FtIvKFKNijRAl7+Qlgv7UEddQSOtNLcyjJzaS5Sx1r6UJjR/DusaGjlyd2Dw5o3HwkGE9zwYxi3C7hn14/k501rfznn3azaU8dn3zDHPIyg0XMK/Kz+L/3L6eurZf7thwfdCxFiRfGGGpbe5hcmMXkwsR4alq6g9+tC2eWUJDloaZZkzDGS6IDhT3A64H3Wr/fKiKrIg0UkZtFZIuIbKmvr4+njWnL4YYOlk8vBmDvqTOiRgSqS3IoysmgVd2kaUNjZ9CjUp6fyW9eGFzRfMvRJgqyPMyryAfgHcumUpyTwa+eO8L51UV89NKZZ41fPKWQZdOK2LjjZOyNV5QhaOzso88XYFJhFtleN6W5Xk60xDdW0L45LM7NoCwvk8ZOPa+Ol2iImhNAddjzqda2iGOsOJpCoJGgV+dvxpgGY0wX8DCwLNKbGGNuN8asMMasKC+Pe//BmLH1aBNNSTiBe31+TjR3c8nsUvIyPaG07qONnVQVZJGV4aYoO4OWYYJGldSisbMPr8fFBy+ewfOHGjnedPbd5OYjzayYUYLLJQBke9189LJZ5HjdfOed5+FxDz7dXH/+ZPaeah9UNkBR4kWtJWCqLC/NlOJsTrTE2VNj3RwWZnspyfWGvKLK2ImGqNkMzBWRmSLiBW4CNg4YsxFYaz1+B/CUMcYAjwHnikiOJXYuB9Im17Oj18dNt7/AD57cn2hTBnG8qYuAgVnlecyrzAsFCx9t6mJaaQ4Axbleuvr89Po0DCodaOzooyzXy6pzKoAzy00A7T39HDjdwflTi856zT9fPpuXvnwlcyzvzUDefN5kXIJ6a5SEUdsaFDCTi4KJt5MLszkR5+WfFttTk5NBaZ43KW90ncKERY0VI3MLQYGyB7jPGLNLRL4mItdbw+4ASkXkAPB5YJ312mbguwSF0Q5gmzHmLxO1ySlsOdJEv9/w0uGmkQfHmUP1wXiaGWW5zJ+Uz766diudu5MZpbkAFGZnANCqcTVpQWNHLyV5XuZW5JOf6WHL0ebQvj21QdF77tSCs14jIqE4mkiU52dyyZwyNu48SfA+R1HiS21rZE9NPOej7akpyvFSkpsZWupVxk5UYmqMMQ8bY+YZY2YbY26ztn3FGLPRetxjjHmnMWaOMeZCY8yhsNf+xhizyBiz2BiTVoHCLxwKipm9p9ro6PUl2JqzOWwFCc8szWV+ZT4tXf0cbuikoaMv5KkpygmKmlRYghpvrSURWS0iW60aS1tF5I1xNz5ONHX2UZqbidslLJ1ezNYjZ0SNnZq9eHLhmI/7lvOqONbUFYrbUpR4crK1G6/bRWmuF4ApRdn09AfimgRhv1dhdgZllqcmEFCRPx4SHSic1rx4uJFMj4uAgR3HWhJtzlkcaeykJNdLYU4Gi6cEL1Rf3bgLIOSpKc4JngSaHe4qnUitJaABuM4Ycy7BJda74mN1/Gno6Aud+JdPK+a10+2hVP9XT7ZSnp9JRcHYa2deMqcMgOcPNkbPWEUZJbUtPUwqzArFgoXSuuOYAdXa3U9Blge3SyjJ9RIwqXGzmAhU1CSIrj4fr9S0cuOKakRga5grPxk4VN/JzLKgeFk+vZiPXjqTZ/cHy95PKwl6auzlpxT48o271pIxZrsxxg4I2QVki0hmXKyOM42dvZTmBUXNihnFGAPbjwXn7a4TbSyeXDDcy4dkanEO00pyeP6Qihol/tRahfdsQgX44lirprmrj2LrhqE0L3j6aNIlqHGhoiZBbD3ajC9guHJhJfMr89lyNLniak60dFNt3bGICP/2pnP49Kq5VBVmMas8KHbs5acUiKmZSK2lcN5OMC4s4tnIyWUJuvp89PQHQifcJdVFuAS2HW2mu8/P/tPtIY/eeHjdrFJePNQYaqugKPHiZEsPky0hA2dETU0cPTUtXf0UWTeJtje0QTOgxoWKmgTxwqFG3C5h+fRilk8vZsexlqQ6oXf2+sjPygg9FxE+v3oez617IzneYOBnaPlJa9UgIosILkl9bKgxTi5LYKeYllgn3NxMD+dUFbDlaDN7T7URMLBoHPE0NhfPKaWtx8fuk21RsVdRRoM/YKhrCy4/2RTlZJDjdXMyjrVqWrr6KMqxPTXB35oBNT5U1CSIrUebWTy5gLxMD8unF9Pe60uqWh2dfX5yMt2Dtge7WwTJ8brJcEsqLD9NpNYSIjIV+CPwAWNMSrYut4uBlVknXAhWP918pImfPxuM+188ZXzLTxD01AA8f6hhhJGKEj0aOnrxBQyTw0SNiFBVmMXJONaqae7qD3m+7RuHRm0fMi5U1CSI+vZephYHY1Psqr3JElfT7w/Q5wuQ5x06FReCX/7CbG8qNF8bd60lESkC/gKsM8b8I14Gxxv7BFuaeyZc6FNvnMs5VQU8/MopinIyQm778VBREFzW1GBhJZ7YwsVO57apyM+Ka0+ylq6+kOe7xPqtVYXHh4qaBNHe4yM/KygappXkUJLrZUeSNIfs6gsW08sZpr6ITXFORqhwlFOZSK0l63VzgK+IyA7rpyLOHyHmDFx+sh/f/dGLuHbxJG5YMuUsL954uHh2KS8ebqKnX4s5KvHhlF2jpujsrL3y/Ezq4yRqfP4AbT2+UOKFx+2iKCdDqwqPk5GvWkpMCBc1IsLS6qIkEjXBmjm53sHLTwMpSgFRA8FaSwTbdIRv+0rY4x7gnRFe9w3gGzE3MMHYd42lYctPAHmZHn76vuVReY9VCyr5zQvHeOFQI1fMTzldqCQhJy1RM3mQpyaT0229GGMmLNZHoq0neL4tzjkTw1iaq1WFx4t6ahJAvz9Ad7+fgrBA3KXTijhwuiNU9yORdPaO3lNTmO3VQOE0oLGjlxyvOxQkHgteN7uUHK+bTXsGdwBXlFhwqrWbTI8rFM9iU56fSXe/n86+2HsNm8OqCduU5mbGdfkrlVBRkwDaLWVue2oAllQH42permlJhEln0dk7ek9NcU5GUggxJbY0dvadtfQUC7Iy3Fw2t5xNu09rywQlLpxsDaZzD/TGlOcHY8fq22MvLGxPd7iw0v5P40dFTQJo7wlO4vCU6fOqCxGB7UlQWbjTWn4azV15qiw/KcPT2NkXqlETS65cWMmpth52aWq3EgdOtfYwKUIV7PiKmsGempJcrwYKjxMVNQmgrXuwp6YgK4O5FXmhCq2JpMtafsqNkNI9kKIcL939fg3uTHEaO3opi7GnBuAN88txCTy+W5eglNhT23J2NWGbRHhqzoqpycukuasvqWqXOQUVNQnA9tQUZJ+9jrvEChZOtOt9rJ4aQJegUpzGjtgvP0HwZL54SiFbjiRXhW2noI1ZR48/YKhr7x2U+QTBlG6A+vbYF+ALxdRkh8fUeDFGC5uOh6iImvF+kcL2TxORDhH512jYk+y0RYipAVg6rZjmrn4O1nckwqwQdkp33igChe0voi5BpS7GGJo6+yjJi72ogeAJ3Y47U0aPNmYdGw0dvfgDhkmFg+srFWVn4HEJp+PgqWnt7sclZ18P7CxDTeseOxMWNRP8Itl8F3hkorY4hTbbU5N1tqfm8nnB0vmP7Uqs690OFI5UUXggtqdG7yhSl15fgD5/YNB8jRW5mZ7QHFTGhDZmHQN24b3JEZafXC6hLC8zLstPzVaLBLtLOIRVFdamlmMmGp6acX+RAETkBuAwwS9SWmDfhQ68SEwuymbptCIeebU2EWaFCBXfyxi9qFFPTerS0RvZsxgr8rM8tKuoGQ8xb8zq5KasA7EL702KIGogfgX4Wrr6Q4X3bOzK3ZoBNXaiIWrG/UUSkTzgi8B/RsEOx2DH1ORFuEi8aXEVr55o41hj/NreD6Sz10emx4XHPfL0sCP2U6BVgjIEHZYIH81yZDTI9aqnJlGM1JjVyU1ZBzJU4T2b8vz4eGpau/sHxVdqrOL4SXSg8K3A94wxIwaRpNIdQnuPj1yvG7drcKXKaxZPAkiot6azz0fuKC9gRdaXMQWaWipDYHtq4iVq8rI8dPX5NfNj7Ghj1jEwVOE9m/I4LT+1dQ/21NjP1QM+dqIhaibyRVoJfFtEjgCfBf5NRG6J9CapdIfQ1t1/Vo2acKpLcjhvaiEPv5I4UdPV6ydnFIX3INip2+txqZs0hQmJmjgtP9niyc7CU0aNNmYdA7VDFN6zqSjIpLEz9mnVrRFETVaGm6wMl3pqxkE0RM24v0jGmEuNMTOMMTOA/wW+aYz5URRsSmrae3wUZA99gbhqYSU7a1oT1nq+s89H7ijL4YsIlQWZnG6Lfeqjkhjs5af8zPgECtuipkMzoMaENmYdG7VDFN6zKc/PxB8wMb9hC4qawefbomyvLuuPgwnfehljfJZ35THADdxpf5GALcaYjQS/SHdZX6QmgsInbWnvHdpTA3DxnDJ4/DVeONTEm8+riqNlQbr6/KMqvGdTmZ/FKRU1KYvtqRnLnJgI9tKnxtWMHW3MOnpOtfawcmbJkPvL884U4LOL8UUbY8xZHbrDKczWau3jISr+5PF+kQaMvzUatjiBtm7foG7H4Zw7pZBcr5vnDzUkRNR09o4+pgagsjCLPVrWPil48VAjbpewYsbQJ+ux0h7v5SfrfTQDSokV/oDhVFtPxMJ7NqGqwhE85g0dvTywrYb5kwpCpTjGQ6cVOxZR1GhfvXGR6EDhtKS9p3/Ymh8ZbhcXzCzh+YONcbTqDF19o4+pgTOemkRXQk53/AHDp+7Zztf+vDuqx03U8pN6apRYMVzhPZuhWiXct+U4F//XU3zz4b18fYLfNVu0RBI1RdkqasaDipoE0N7jG7Hmx+tmlXKwvjMhsSodvaOPqQGYVJhJV58/tEyhJIYXDzVyur2Xg6c7oiowO3t9uF1CVkZ8ThcaU6PEmtNtQaFSOcyykt0q4URzd2hbvz/Adx7fxzmTC/jgxTM4cLpjQufo1q7IhVhBmwWPFxU1ccYYY4ma4e96Xzc7WA/r+UPx99Z09flHVU3YptIKtqvTuJqE8uCOYNJhZ58/quXdO3p95GV6hswSiTYhUaMiWYkRp62eThXDBApne93MqchjZ01LaNtju05R19bLZ1bN4R3LpwITO0cP66nJ8dLSrYHCY0VFTZyxS86P5KlZNLmQ/CwPLyRA1HSO0VNzRtRoSe9E0dPv55FXTzG1OOhOP3g6ev3D2nt8catRAypqlNhji/6KEQKAl00rYvux5pDnc8NzR5hWksPl8yo4p6qAwuwMnjswcVEzsPgeBIVOT3+Ann7/uI+fjqioiTNtQ3ToHojbJSyfXsz2Yy1xsOoMPn+AXl9gVB26bWxRY5cdV8bPX1+rp88XGPPrHt9dR3uPj0+vmgvAwYbOqNnU0dsfV1Gj2U9KrLGXn8ryRhI1wSbDhxs62X2yjc1Hmnn/RdNxuwS3S7hoVgnPHWoYtx1tw3pqtKrweFBRE2fO9H0a+SIxuzyPI42dBOJYWbXLuisYU0p3QfDEUNeuomYiHDjdzgd/+RK33L1t1MKmubOPz9y7nc/cu53qkmzetnQKOV53VD01Hb2+uGU+AXg9Lrwel2Y/KTHjdHsPxTkZeD3DXwKXTS8GYNuxFn7z4lGyMlzcuOJMrdmLZ5dxvKmb403ja2sTWn6KUNW4KNtuQaOiZiyoqIkztjIfTXPAGWW59PQHohofMRJdvbaoGf1FLMfrIT/LQ516aibEnIp8br1uEY/vruNT92zD5x9Z2Pznn3bx8Cu13HzZLB78xCV43C5mledyKJqemjgvP0FwCUo9NUqsON3eGwoEHo455XnkZ3l4dn89D20/wVvOm3yWAAnFPo4zU7W1ux+XQF4Ez/iZZsEaVzMWVNTEmaE6dEdiZmkuAIejeIEaCbs0/VhSugEmFWRpTE0UWHvxDL7yloU8tquOe146NuzYbceaeXDHSW6+bBZfuvYcSi1X+uzyPA7VO9dTA0FRo9lPSqw43d5LRcHIBfVcLmFJdREP7ThJZ5+f966cdtb+uRVB0fPKidZx2WE3s3RF6ANYGKGv3k+fOciPnz4wrvdKF1TUxBlb1IyU/QQwoywHgCONcRQ1dvXYMcTUQDCuRqsKR4cPXTKDC2eW8P0n9w/prQgEDP/5p91U5GfyiSvmnLVvVlkeJ1q6oxZg2NHrIz/OnprcTI8GCisxo76tZ9RVgpdNCy5BLawqYEl10Vn7RIRpJTnUNI9/+WmoG1xb1ITH1Pz6+SPc8ffDWhNsGFTUxBk7UHg0y0+TC7Pxelwciaenxlp+GktKNwRFjfZ/ig4iwrprF9DQ0ccdfz8cccxvXzzKzuMtfOGaBYOWCmeV52JM0MM3miWskUjE8lO+iholDJ8/wPpH9kbFA2mMob5jdMtPAMutuJr3rJwWsazB1OJsasJq2YyFtp7BzSxtQoHCVkzN6fYealt7aOrs42AUPbGphoqaONM+BlHjcgnTS3LiuvzU1TdeT00mp9t74xrUnMosm1bMNYsm8X9/Pchrde1n7Tva2Mk3H97LpXPLePuyKYNeO7s8D4D/fmwfC7/6GI/vOjVuO/wBQ2eff0wxVtEgN9OtokYJ8fS+en7214M88ur457JNc1c//X4zYjq3zaVzy7j9/cu56YLqiPunFudQ09w9Lu9JpA7dNnmZHtwuCdWqefn4mSWuFw83jfm90gUVNXGmvceHS0YvGqaX5sZ3+alv7NlPAJMKs/AFDI0x7mibTnzluoXkZHr40C838+KhRtY/spdP37OdD/1qMx638O13nBfxznFmWTAW66m9p/H5A/xu8/Fx22DHWI1GhEeTvKyMkNdQcRZfvP9lvvTAK2zceXJc5QkicfeLRwGi4g0+U3hvdKJGRLhq0SQ87siXy+ribLr7/SN28+71+bnx/57nxbDaY8OJGhGhKKyp5cs1LbgESnO9vKSiZkhU1MQZu5BZpMCwSMwsy+FoY1fcPCBdvXag8NguYrYr18lVhUXkGhHZJyIHRGRdhP2ZIvI7a/+LIjIjbN+XrO37ROTqaNgzuSibX37wApq7+njX7S/wi2cPWSc24X/eeT5VQ/Styfa6+cyqufzX287lw6+fyd/214dc2GPFDtaNf/aTOxR/pjiLjl4ff955kk/fs511D7w84eMdb+rimdfqgcjNJceKXaNmtMtPIzG1OBj7eHyEJaia5m5eOtx0VgXiNitQeCgKczJCgcI7a1qZV5nP62aX8uKhJo2rGYKoiJrxXgxEZLWIbBWRV6zfb4yGPclMU2cfRTlDd+geyIyyXHp9AWrjJBbOeGrGdhGbVBg8QZxsGd/acqIRETfwY+BaYCHwbhFZOGDYh4FmY8wc4HvAt6zXLgRuAhYB1wA/sY43YRZPKWTDP13Irdct5IV/W8Uz/+8NbPr85Vy9aNKwr/vc6nm8+8JpvOW8yfT7DY/vHp/bviPOHbptNKXbufz4vcvY8dWr+Pjls3lg2wme3nt6Qse7d/MxBJhVlhsSJBNhtNWER8vUkuDNxUjBwnZxUjtL1BgzrKcGrKaWXf0YY3i5poXzphaycmYJp9p6xh3Hk+pM+EwVdjFYDdQAm0VkozEmvH1p6GIgIjcRvBi8C2gArjPGnBSRxcBjwOAggRTidHvPmL5Mdlr3kYZOphQN3VE2Wpzx1Iztmjy3Ig+v28XmI01cNcIFN0m5EDhgjDkEICL3AmuA8Hm8BrjVenw/8CMJrv+sAe41xvQCh0XkgHW856Nh2AUzSrhgRsm4Xnve1EKqS7L588u1vHNF5JiA4QiJmgRkP3X3+/H5A0O6/WNNc2cf//XIHrr7o7OEMlHOm1LIRy+blWgzRoXbJXxu9Vye3FPHv/3xFf7y6UspyR39zVw4f9x2givmV5CX6WHH8ZYJ22Z33R7t8tNI2J6akURGrSVq7CW07n4//X4zvKjJ8XK6PShgmrv6OW9qUehc8OLhJqpLcqLxEVKKaJypxn0xMMZsDxuzC8gWkUzr4pCS1Lf3Mn9S/qjHzyg7U6vmkjllsTIrRGefH6/HRcYYLyS5mR5Wzirhqb2n+fKbBzo4HMEUIDz4pAZYOdQYY4xPRFqBUmv7CwNeO0ici8jNwM0A06ZNG7g7JogIbz53Mr949hDNnX0Uj/HC0tGToJgau1VCn5/C7NHNxX5/gI07TrJmyeSoCKGXjjRx35YaphRlkzlC5dl4UJgd3//BRMn0uPnvd57PjT97nut++Hd++J6lLJlaNOqld4Da1m5OtvZw82WzqGnupr69F2PMhJqrnm7vIS/TM+Yl9qHIy/RQnJMxCk9Nt/X+wcvbcM0sbYqyM3itrp2Xa4JBwudPLWJuRR55mR5eqWkJNdVUzhCN/+pELgbhTTPeDmxLZUEDQVHz+jGIk0kFWWTGMa072MxyfCsnb5hfwdf+vJtjjV1MK9U7iIEYY24HbgdYsWJF3BbEr5hfzs/+epAdNS28YX7FmF5re2rinf1ki6iOXt+wJ/1w/rC1hnUPvEJhdgZXLqycsA12TM89H71I5/M4WVJdxO8//jo+8dttvO0nz+ESmFeZzz9fMZu3nDcZ9wgCx+59t3RaMX3+AN39/mDdpFHU+RqKYDXh6HhpbKYW53C8aXSeGjvu8Ewzy6G/WwXZGbR29/P47lN4PS7mT8rH5RLK8zM1KWMIEn/7AYjIIoJLUh8bZszNIrJFRLbU19fHz7go0tPvp63HN+qiTxBM654/KZ9dJ9tiaNkZOvt8476DeeOC4AXzqb110TQpXpwAwtdnplrbIo4REQ9QCDSO8rUJY4HlGXztVPsIIweTqEDh8TS1vNfK8tpTG53vSodVfiHe8USpxvnVRfz5U6/na2sW8c9XzAbgM/fu4P13vEivb/gMt+3HmvF6XJxTVRA6b9ZPsG1MfVvvmM7BoyFYq2Z0MTUNHb34/IFQ8P7wy08ZtPf4eGjHSW6+dFaoV1VhWFaUcjbREDUTuRggIlOBPwIfMMYcHOpNjDG3G2NWGGNWlJeXR8Hs+BNayx1j1P35U4t4uaYFfxwyoDp6fGNO57aZUZbLrLJcntrnSNG5GZgrIjNFxEsw8HfjgDEbgbXW43cAT5lgCsJG4CYrIH4mMBd4KU52j0hRjpfKgkz2jUPU2E0l8zPHf2c8HmwRNdoMqL2n2kLxFntORUfUtCdI0KUixblePvC6Gfy/qxfw8Kcv5ba3Lua5g4186Q+vDJvFs/1YC+dOKcTrcYXOmxPthXe6vYeKguhkPtlUl4xcq8auuB4w0NjZN+rlJwh6vD5z5dzQ9uKcjFD9GuVsoiFqxn0xEJEi4C/AOmPMP6JgS1JjpyOO9S5hSXURnX3+uFSRrG3tYdIQqcKj4Q0LKnjhUGOoiJ9TMMb4gFsIBqvvAe4zxuwSka+JyPXWsDuAUisQ+PPAOuu1u4D7CMaRPQp80hiTVEVW5k8qYO8EPDXjFbrjJW+MnprfbT6O1+3iwpkl7Kkd++eMREevj0yrY7gSPVwu4b0rp/P51fN4YPsJPnPvDmpD8SY9fO53O7juh3+ntaufV060stRqTWCfNycqaho6+ijPi76nptcXoKFjaKFxqvVMa4a6th7arO/WcKLmvOoiFkzK5wc3LT0rzrE4x0tzp3pqIjHhWxArRsa+GLiBO+2LAbDFGLOR4MXgLuti0ERQ+EDwIjIH+IqIfMXadpUxZmI5gEmKnY44ZlEzrQiAHcdamFc5+iDj8XCsqYvzqwvH/foLZhRzx98Pc6i+k8VTxn+cRGCMeRh4eMC2r4Q97gHeOcRrbwNui6mBE2DBpHx+dahxzNlEnX0+sjPccc9AyguLqRmJfn+AP24/wVWLKplTkcf3n9xP1wSWUW3aeiYWu6EMz6feOAdfwPCzvx7k0VdPUZ6fSXNXHz6/oc8f4FP3bqfXFwid/yqisPzU6wvG5JTmjS8TayimFgdvBI83d0U8v/f0+2ns7GP1wkqe2F1HXVsvzVZMzHCiZtm0Yh797GWDthfleEfs3t3V58PjSj9RHhW/6ngvBsaYbwDfiIYNTsD21Iw1SG1maS4FWR62H2/hxiFKdUeD1q5+Wrv7mV6SO+5j2G7dYNVOZ4maVGZeZT59vgBHGruYU5E36te198S/Qzecqbg9GlHT0NFLS1c/F88uozTPizGw71Q7S61GhOMlGJCqS0+xQkT4/Op5vHP5VH79/BGau/rxelx85PUz+dHTB3hgWzCKwf4/FmZn4HW7QhWBx4Mdh1I8hlphoyE8rXtZhHln39CeP7WQJ3bXcbq9h72n2inL8446ED6c4pwMOvv89PkCEUVLT7+fa7//LMunFfPddy0Z8/GdjH5j40h9ey8ijLleg8slnF9dxM4o1GgYjuNWoNtEah/Ygi0aRbKU6GEHC+871T4mUdPRG/9mlhCW/TSKmBrbDV+Sm8HCqgIA9tROXNS09/SrqIkD1SU5g8pArLt2AY/vqiPH62ayVdhTJJj1MxFPjd3KoCQ3uh64Svtmbogiqfby2qIphYgEC/DtOtnK4imF40pPL7KuIS1dfRHjg+74+2GONnZxormbL73pnKgHRicz6eWXSjD17T2U5maOy5W/pLqIfXXtdPfFLlTjWFNQ1EybgKiJ1rq3El3mVOThEthXN7Z4k46e/oSImrFkP9lu+KIcL1OLs8nL9EQlAyoR3cknSrK1+hgvFflZ/PDdS/nqdYvOuuhHS9RE21NTkOUhK8M15HnPDhKuLs6mLC+TY42d7D/dwbnjXKIvtjp4N0fIgKpv7+UnTx/g/KmF+AKG328df+83J6KiJo7Ut48/lXBJdRH+gOGVE60jDx4ntqipLhl/oHCmx01hdsaE0y6V6JKV4WZGWS77xpgZlChPTYbbRabHFcoQGY7msCUFEWHBpHz2RiEDqr3HWctPydrqY7y8YUEFbz6v6qxt5fmZE/ICn/HURFfUiAgV+VlD9r6z07knFWZTWZDJ3/Y34A+Ycccd2qKsOUJczU+fOUivL8D33rWE180q5e4Xj8Wtd2Ay4JxvbApwegKi5nwrA2DH8WYunDm+kvkjcbSxi5Jc74SDIyvyMye07q3EhvmV+WP2YLR1+5ieoMJzcyvzRpWebZ/Y7bvXc6oKeHD7iQlXng0KOkcFCidtq49oUZGfydajzeN+fWiuRFnUAFQWZA4pampbe8jP9JCX6aEyP4tXTwTn9Xg9NUXWXI8ULLyzpoXl04uZVZ7He1ZO41P3bOdDv9qcFF7HrAw337nx/Ji+R+I/ZRpR39477uylsrxMphZns7Mmdp6a401dUeklUlGQqctPScicijwe3XUKf8CMWMnVpqGjl+UzJhabMl6WVhfzx+0nRrQ3fPkJYNHkAu564SiHGjqZXT76+KGBtDkvpiYlW32EU56fSVNn35ABsiNhe2qKxhGcOxIV+VlDivBTrT2hpr92DExprpeqwvHVyznjqRnsyTza2MUqqxDq1YsmcencMk4kSaPheFQmd9Q31skEAoaGjolVsox1sHAwnbtowsepyM/ipcNNEzdIiSqF2RkYE0zTLhiFN84fMDR19VEWg7va0bB0WhF3vXCUA6c7hu2X1tzVT67XHbrIrbAa/m0+3DRuUWOM0eynCCSq1YeNXYCvsbOXqnHU02ru7KMwOyMmJQoqCjL562uRb+Zq286ImkqrkeZ4g4ThjKdm4PJTV5+Pho7eUFsPr8fFXR8eqGtTG42piRMt3f30+82Eeo4smVpETXM3DR3R94L4/AFOtHQzbQLxNDYVVjDfcNU1lfhju59Hk1EEwbtaY6AsQZkTdvbS9mPDLzc0d/WFvDQAs8tzKc318tKR8Qvrzj4/xsS/kecESdlWHzblE8yubOrqj3o8jU1FfhYdvb6Iwe11rT2hDClbmI136QkgOyMo4lsHeGqikezhdFTUxAk7cHainhqAl2taomDR2dS29uAPmKh8GcrzM+nzB0YV5KnEj7H2U7LFc1mUq6+OlhmlORTlZISaGg5FS1d/6M4VgkGbF8woYfMERM2ZnleOiqlJ2VYfNmVW0bzGzvGJmubOvlDsVbSxPTADl94HeuknFZ7x1IwXEaE4J2OQp+ZYo4qalBE17T39bHjuyIgN0hJFSNRM4AKxeEoBLoEdx6MfV3Mm8ykaMTXR6dGiRBe7iF77KEVNo1XyvTRBy08iwtLqIrYfH9lTMzBF94KZJRxv6g5lnYyVdquZpZM8Nane6gOgNDd4/mwcph3BcDR19sXMU2N7YgYGC7d29+MLmNDNwSVzyvjym87hDQsm1sOwOMc7KKbGPo8nKrg/GXDON3YEHnnlFF/duIufP3uIL1yzgOvPn5xok4DghP7yH1/heHMwUGsijdRyvB7mVebHJK4mmm7L8HLmsW7roIyesfZTCnlqEli4a+m0Yp55rZ62nv4h44BauvqZUnT2sumFVlzNS0eaxnUusIWf0zp0p3KrD4ASy1NjB/yOleauPhZNLoimSSEqwvo6hXPG4xm0PdPj5qOXzZrw+xXlZAzKfjrW1EV+lmdcVYpThZTx1Nx4QTV3ffhCCrMz+PQ927l14y58/kCizeKZfaf588u1tHf3s3JmyaCT71hZUl3EzpqWqMer7D7ZRnaGe1zBdwMJVRXWtO6kYqwxNaGTcW4iRU0RxjCskI/kqTmnKp9cr5uXDjeO633tDt0FDhM1qU6u143X7RqXqDHG0BhDT419wzqwRtd4GxmPRCRPzdHGLqaX5kyolIHTSRlRA3Dp3HI23vJ6Pvz6mfzquSN87K6tCV+O2nq0mVyvmyc+fzm/+9jrJtxc7PzqIlq6+jlqrZ1Gi7/tr+fi2aWjTvUdjtDyk7ZKSCpCombUnpo+MtxCQXbiLuwLJgXvqg/Vd0bc7w8YWrv7B8VJeNwulk0v5h8HGsdVeMyhMTUpj4hQkusdl6jpsnolxaJGDQQFcKbHFcFTE7Q12p3BIzW1PN7UNaHefalASokaALdL+I+3LOTrNyzmyb2n+fQ92+lPoMdm27FmlkwriopYAEKprQdOd0TleABHGjo52tjFFfMntsZrk+t1k53h1piaJGOsoqaxo5fS3MyE3vWV5XnxelycHKLORlt3P8ZwVvaTzduXTeVwQycP7Rx7Eo8TY2rShfGKmlA14Si3SLARESoLsqgbcDPX0B6bgPvinAxauvpDXnt/wHC8OTq1xpxMyokam/dfNJ2vXreQx3bV8R8PvpoQG7r6fOypbY/YtXW8VFvdYO3mkxPB/jI8s+80AJfPq5jwMcEqGa4F+JKO8WQ/leUnJkjYRkSYXJjFySECfs9UiB3sUbn+/MmcO6WQ/350Hz39Y/PYdjg0piYdKM3z0jgOUWPPlVgtP0HkauoNHb14XBL1OJfiHC++gAnFf51q66Hfb9I6SBiiJGqStYnahy6ZySeumM29m49z/9aaaB56VOw83oo/YFg2PXqipizPS3aGm5rmiVWIrGvr4dJvP82G547w19fqmVmWGyrYFA0q8jOH7FirJAavx4XX4xp19lNDR1/C0rnDqSrMHtJTY8cURPLUuFzCv73pHE629nDnPw6P6T3b7OUnr4qaZKM4Z2KemlgtP0EwA2rgsntDRy+leV5cUfLW2xTarRKsLvVHG4NLtOmczg1REDXJ3kTt86vncdGsEv79wVc4cHpsHYonyjaraNiy6uiJGhFhanE2x5sm5qm5f2sNNc3dfHXjLp7d38Dl86Kz9GRTkZ8VKsCXLD9KcAlqtJ4ae/kp0VQVZVE7hKhpCfV9inyhet3sUi6dW8avnzuKfwyxNXaH7mhfiJSJU5LrpTlZPTUR+j/F6ubAnvMt3cHPdVwL7wHRSelO6iZqHreLH7x7KZd/+xl++Y8j3PbWc6N16BHZerSZ2eW5IUUdLapLckIp4uPBGMPvtxxn+fRi3CK8dKQp6qKmsiCLv7xSy8wvPTzy4Djw728+h49cOvE0SqeTl+kZVfaTMSZ4Mk7w8hPAlKJsTrX14PMHBpW3P9Ohe+jv2LsuqOaWu7fz3MEGLp07unne7ry+T2lDaa6X9l4fvT4/mZ7R3wM3WR6NWMXUQPBmrrPPf1Z3+4aO3hiJGrtVgu2p6cLjknH3k0oVovGtjXkTNZhYI7WK/CyuXFjJI6+e4tbrF5ERg74fAzHGsP1YM6sXVkb92NXF2Ww+3DRiF+IXDzXyud/t4NHPXXZWjY/NR5o50tjFp944l9WLKnl67+moi5oPXTKDopwMAkniIVkaxbgmJ5Ob6aGjd+T4kvZeH33+QELTuW2qCrMJmGAxx8kDSiIMbGYZiSvPqaQgy8MfttaMWtRo36fkxV4+au7sZ1Lh6EVNc2cfbpfE9P9qVxWua+shz+o71tDey9yK6Nfrsue8/R041tTF1OLsmPS1chKO+dZOtJHadedV8aedJ/nHgQaumB+dgNjhONLYRXNXf0wuptUlObT3+mjt7h/2ZP7qyTZOtvZwuL4z1GLBHzD89sWj5GV6uPbcSeR4PaxZElFHTtjGT6+aG/XjKhMjP9NDR+/I7StCGRtJ4KmZXBS88zzZ0j1I1DR3BS9Uw9WTycpwc935k/nDthrLAzOy57S958ydtpJc2BWuGzt7Q00iR0NTV7BFQiyXFO2ln6ONwQ7xsfR42p/djq883qSZTxCdQGFHNFG7fH45+Vke/rSzNhaHH4TdhG/ptKKoH3uqnQHVNPwSVKul4Gtbg+N+8ewhln39CR7acZK3Lp1CjgZBph25mW46R+GpsbNLkiFQ2BYykTKgmrv6KcrOGDHt/B3Lp9LTH+DPL4/u+9/e6xuV+FHiT0mYp2YsNHUMLtIYbeyu8AdPB4N227qDHs9o16iB4FJyZUFmqIbTsaautI+ngeiIGkc0Ucv0uLlm0SQe33VqzOmd42HH8RZyve6YuB2rrU7aI6V1t1gNJWuti8F9W45Tnp/JD9+9lH9/yzlRt0tJfvKyMkZVp8b21CRFoHDhGU/NQFq6+s5qZjkUS6qLWFhVwO1/OzSqSuPtPf2azp2klI6zqWVTZ19MM58guDRWnJPBoYZgHbH6GDeFnVWWx6GGDtp6+mnu6ldRQxREjZOaqL35vCrae328cGh8pdPHwvZjLZw3NXpF98KxXYwjZUC1dJ0RNcYYjjd1c/m8cq47f/KYAuyU1CEv0z06UdORPMtP+VkZ5Gd5ImZANXf2j+ruW0T49Kq5HG7oZOPOkyOO7+jxaYuEJMX+f481rbuuvSfUdDKWzC7P46DlPYl1p/tZ5bkcqu/U7txhROVb65QmahfOLMHjEjYfaYppXE1Pv589tW1RaVoWiYKsDAqzM8bkqWno6KO730918cR7OynOZbTZTw0dfYjENlNkLEwpyuZES6Tlp77QcuxIXLWwknOqCvjhUwfIznDT3uvjHcumRoyx0Jia5KUox4vI2ESNMYbTbb2hvnSxZFZ5Lk/tDRY0bYhR36cz75VHa3c/O6zeaBpTk8IVhSOR4/WwaEohmw83x/R9dp1sxRcwLLWCc2NBdUn26GNqWrpDAkgnfXqTm+mhu98/Ys2Wxs5eirIzkiaToqowKxQbFk6kvk9D4XIJn7G8Nf/822184f6Xedqqph2Ozx+gu9+vMTVJitslYy7A197ro7vfH8pOiiWzy/No6Oijtbs/rEVCbG4OZpUH+zw9s68eIKoFVJ1Kcpyx4sjKmSXsON4S07ia7cdaAFgSgyBhm+rinBE9Na1hnhp7qUpFTXoz2v5Prd2+YTPr4s3koshVhVu6+sdUfv7qRZVs+KcLeeiTlzCpIItfPXdk0JhQiwT11CQtY+3/ZFf5jcfy0ywrWPhQfQcNHX0hERYLZpcF3+u5gw0U5WScVbojXUk7UXPBjBL6/AFermmN2XtsP97ClKJsKvJj9wWqLsmhprl72A7E9vJTXVtPaM11qi4/DUJESkTkCRHZb/2OmIcvImutMftFZK21LUdE/iIie0Vkl4isj6/1YyNvlP2fkq343OSibJq7+unuO3Mz0ucLelTGImpEhMvnlXN+dRHvf910nt3fwP66syuN24H1ZXFYqlDGR0nO2Po/2S1bYrUMFM5sy3tysL6Tho5eSnKj3yLBZkpxNl6Pi64+P9P1hhVIQ1GzwurDtPlIU8ze4+WaFs6vLozZ8QEmF2bR5wvQ1BX5ix0ImJBr3hcwbD/eQlmeV9O4I7MOeNIYMxd40np+FiJSAnyVYGHJC4Gvhomf/zHGLACWApeIyLXxMXvs2Bk9I3lqOpIspiSUARW2BGV7IkeT/RSJmy6oxutxseH5I2dt33uqDYAFk6KfuahEh7F6auqsJpPx8NRUl+TgcQkH6zvYc6o9pnE8bpcwszQ39L5KGoqa4lwv8yvzefFwbERNd5+fmuZu5lcWxOT4NhXWl7N+iE7Y7T0+jIFzqoJ2bDnSNOqAyjRkDbDBerwBuCHCmKuBJ4wxTcaYZuAJ4BpjTJcx5mkAY0wfsI1gvaWkJHeUy0/hZd6TATsV187ogzOipmCc3Y9L8zJ5y3lVPLj95Fkez7217XjdLmaV5U7AYiWWlOSNrf9TPJefMtwuppfmcM9Lx9h5vIX3rBxbBfyxYsfVaOZTkLQTNQAXzCxm29HmUdWrGCuHGzoxBmZXxPaEaLtRhxI1dpOzBZOCoqatx6dKfmgqjTF2VbZTQKTeFpHagZxVillEioDrCHp7BiEiN4vIFhHZUl9fP2Gjx0O+LWpGyIBq70mu4nMFETxMrdYcH8vy00AumllKR6+PY2HlEfacamduZV7SBEkrgynN9dLc1Tfs8ns4dW295HjdcRPqs8rzaOnqZ8X0Yt59gYqaeJI8t2JxZMX0En7zwjH2n+4IeTKixcH6YNElu7JkrKgYSdRYd7TnVJ1xoadzOveVV17JqVOnIu0qCn9ijDEiMuY2HFal7HuAH9jNXQcy0VYf0WC0nppki6nJywwKl/aewZ6aiYiaBdb3Y09tGzMsz8ze2rZR94hSEkNxjpeACcYNjqbr9uk41aixOaeqgGf2nea/3nZuzDu9z7KChfWmNUjynLXiyBIr1Xr7sZaoi5oDpzsQgZkxdl3bxZxODyFq7BP+zLJcMj0uen2BtJ70mzZtirhdRFoAv4hUGWNqRaQKGJznG2zfcUXY86nAM2HPbwf2G2P+NwrmxozRZD8ZY5KuoaNtS7iHKRqiZl5lPi4JemeuPbeKxo5eTrf3nnUzoCQfdlXh5q6+0Ymatt64BAnb/PPls3n7silML439EubqRZV8qnEOF8woifl7OYG09K9OL82hOCeDHcejX6/mYH0H1cU5ZGXEtmJvbqaHXK97mOWnM0GUdpBltcbUDEV4G4+1wEMRxjwGXCUixVaA8FXWNkTkGwT7mX029qZOjEjiYCBdfX4CJrlSmm2728NFTdfERU1WhpuZZbnsrQ0GB+87FcyEspdtleRkrFWF41VN2Cbb646LoIFgMdZ/uWo+Xk9aXs4HkZZ/BRFhSXVRqApjNDlY3xlK6Ys1FQVZod4iA7EL7xVme0PdXO2eUcog1gOrRWQ/cKX1HBFZISK/ADDGNAFfJ9jrbDPwNWNMk4hMBb4MLAS2icgOEflIIj7EaMgdRUp3qE5LEnlqcr0eRAYuPwXtnIioAVhQVcBeS8zstsSNemqSG9s7MxpRY1cTrtQU/bQgec5acWZJdTHPvFZvxQ5EJyDSHzAcqu/gktmlUTneSJTnZYbqLwykJewudnJhNi450+1YORtjTCOwKsL2LcBHwp7fCdw5YEwNENtF8yiS4XaR6XENu/xkC4dkChR2uYQ8r4f23rOXn/IyPRMO6D1nUj5/ebmWjl4fe0+1U56fSWkSdCdXhqY41Kl7ZFFjVxOuiEM1YSXxpKWnBmDptCKMIapF+E62dNPrCzCnIrZBwjblBZlDempauvvJ9brxelysWTqFmy+bTYZmcyhY/Z+GFTXBfflJtPwEwSWo8OWnlu6+CXtp4MxS075TbeypbXNqfRp3uhSQhDM9yYaq0xWOfeMXz+UnJXGk7VXu/FCwcOS4mr/vb2DFNzbRMoovjc0BO/MpXqImL5P6tqGzn+wy95fPK2fdtQviYpOS/ORljVLUJNHyEwTtDl9+auvuH3eNmnDsDKjvP3mAXSfbuGhWfDytUaaKNCkgCcGYlewM96g8NXaNmngGCiuJY0Kixsnl5QuzM5hVnjtkXM3ju0/R0NHLrpNtoz7mwdPxSee2Kc/PDLpW+wb3sWrtHltPHCV9yPV6HBdTA8HlsI4By0+F2RO3cUpRNvlZHv72Wj3LphXx0UtnTfiYCaCINCkgaVOSO7pWCfGsJqwknol6ahxdXn5JdRE7h1h+2nIk6ME5YAmV0XCwvoOSXO+oUgyjgV2rpiHCElRrlFzzSuqRN2AZZyB2ZlQyZT9B0J72ASndRdkT/66JCOdOKaQk18uP37vMqVkknnQpIGlTnJsxJk+Nipr0YKJnrTWcqd2xgWDdji8OGBO6OwAQEfvu4B4gdHcgInG/OzhnUgEPbDtBU+fZtQ6CAYNBD83+0+1DvXwQx5u641rV0Xannm7vGVSDpqWrP26xPYqzyMv0UDdEgDlAWxIGCkNwOex4WOXfaHojv3Pj+fj8hqrC5A2mH6qA5G233XbW81QvIGlTnOOlKaxtxlCcauuJazVhJbFM9L8c7fLy3x/qjUTkZuBmgGnTolN2er4VELj3VBsXzy4Lbd9xrIWAAa/bNSZPTW1rN/Mq4xdkOFyrhJbu/nE3+lNSm+IcL3tqh15WDS0/JdlFID8rg7YBnprCKM3xZBYzNkMVkLTwpUsBSZuSXC9HG7tGHHeiuZupaVxNPd0Y0c8qIptE5NUIP2vCxxljDBCTuwPr+LcbY1YYY1aUl0enhLmd5WAX3LLZcrQJEVi9sHJMoqaurTeuLs6K/MhNLY0xtHb1UxgF17ySepTle2ns6CP4lR1MR4+PHK8bd4zLu4+V/LBA4Z5+Pz39AV1iPUMLaVJA0qYkd3RNLWuau7WZbxoxoqgxxlxpjFkc4echoM66K2CEu4PqsOdTrW02Cbs7KM/PpDgnY5Co2Xq0mfmV+SydVkRDR9/oaiH09NPR6wtV740HJbleXDK4VUJ3v58+f0A9NUpEyvMy6fMHaOuOHFcTbGaZXF4aCKaY9/oC9PkCtE2wQ3cKUkuaFJC0Kcnx0t7ro883fGPimuYupmiNrrRhomcuu7z8eoa/O/hmWHDwVcCX4Ky7g4R8gUSE+ZPyQ9VEIVhAb/uxFm5YOjkUk3KgvoMLcofvq2HHKEyKo6hxu4TSvMxBnppo9MRRUhe7b1h9R2/E5ZuOXl/SLT1BWIuHXp/O8cH4jTFpUUDSJlSAr6tvSA95W08/bT0+XX5KIyYa5u/48vILJhXwWl17qIX9wfoOOnp9LJtWHBI1++tGXoKqbbVETZwj7CvyMwd5alqi0BNHSV1sURMpaw6CF4JkCxIGyMs606nbFjVFOsfTltG0SjjR3A2gy09pxIRux1KhvPyCSfl09fmpae5mWmkONc3BwLMZZblMLswmx+seVQbUqdb4e2oguIQ2lKdGT/hKJMrygxeDoURNsnXotglvaqmeGsVuajlceEBNSNSopyZdcGRBhmgSngEFcKo1eKKfVJCFyyXMqcgbVbBwXYJKcVfkZw5Kz7VreWi8gRKJkKdmiA7vHT3JvfykokYBKM0buVWCfZOqoiZ9SHtRY6dg28HCp1q7ETmTLj2nPI+Xa1rZdbKVFw41cvl/P80vnh2cpFXb2kNxTgZZGe74GU9QRDV09OLznwmWCwVRJuESgpJ4inOCAeYNHZEvBskbKByczx29Pl1iVUbtqcnKcMWtIKqSeNJe1ORmephWksPeOkvUtPVQnpcZav649uIZwaaQP/oH7/n5Cxxt7OJ3m48POk5dW09CKlZWFmQRMJxVLvxM8bTkuzApicftEkpyM4ddfsrLTD6xcMZTcyamRr2R6Yud3dnUOXQBvhNWOrdIwiMdlDihVz1gVnkuRxs7ATjV1ntWXMz51UU88bnL+K+H9+JyBfvE/M/jr3G0sZPppbmhcbWtPXFN57axhdSp1jOiyk7VVVGjDEVZnjeiqPEHTFDUJOHcyRuw/JSf6Um6WjpK/MhwuyjI8tDUGVmcA9S0dOnSU5qR9p4agBmluRxt6MIYQ13rYI9LUY6Xb73jPP7rbeexZkmwGPKmPWeX5Klr64l7kDCcybYKj6tp6+kn1+vG49Z/rxKZ8vxM6iMsP3X2WfFYSShqwlO626JYTVhxLiW5w7dKqNFqwmmHXvWA6aU5tPf6aOrso7a1e1iPS3VJDvMq89i0uy60rdfnp6Gjj0kF8f/yVBYEY3/OEjXd/eqWV4alLC8zYqBwsjazBMj0uPF6XLT19HO6vVfjJBSKh6kqbMdeaTp3eqGihqCnBmDvqXbaenwjxsZceU4lLx1potW6Q7C7wE4qzIytoREozcvE7RLq2s5coNp6+jVIWBkWe/lpYKsEO3MuGevUQLCqcFt3PztrWlg8pTDR5igJpjTXO2SdGrtGjVYTTi9U1BD01AC8eKgRGLmA3qpzKvEHDP/+0Ks0d/ZxKlRNOP5fHrdLKM/LDNkAwZiaguzku9NWkoeyvEx6fYFQ80qbjt6gUE/GmBoILkFtP9ZCe0+wQKaS3pTmDi4+aqPp3OlJcp654szU4hxcAi8cagIYMeB32bQiPvXGOfzkmYM8d6CB188NdviOdzVhm8rCrEExNYmyRXEGZ6oK953llWlL4uUnCHqQXjnRCsDy6Spq0p1ppTk0dPTS1ecjx3v2nN1+rAW3S5hVnpcg65REoJ4awOtxMbkomx3HW4CgSBgOEeFfrprPnz/1emaU5fLQjpNA/KsJ21QOKMDX3uPTmBplWMqsOkxHGju5/kd/56+v1QNnYmqSMVAYzoitklwvM0o1ViLdmVYSnAPHmroG7Xtq72mWTyvWWkZphooaixmlufRZBexG6+U4p6qA+z/+Or574/l89sq5CbsQTCrMCrVpADumJjkvSkpyUG55an769EFermll8+Ggl9Jejkrm5ScIeku19ohihw4cbTxb1Jxq7WF3bRtvWFCRCLOUBJKcZ64EML00h78fCJ40c8fgehcR3rZsagwtG5nKgizaenx09/nJynDR1p2cDQmV5MHu//TSkaCYsT19dtBlUXZyZhbZ83qpxtMowPSSYJLHsQGi5ul9wZIbb1RRk3aop8bCzoByYixKZVitms4+PwGDBgorw1KS48V2dHhcEgq2PNXaQ2F2Btne+Lb7GC1nPDUqahQozMmgMDuDI1bxVJun9p5mSlE28yo1nibdmJCoEZESEXlCRPZbvyOeaURkrTVmv4isjbB/o4i8OhFbJortxkxUXMxECK9Vo32flNHgcbsoyfEypSibS+eWhTw1p9p6klrYVxRkkpXh4vxqTedWgkwvzTkrpqbX5+cfBxp4w4JyXaJMQybqqVkHPGmMmQs8aT0/CxEpAb4KrAQuBL4aLn5E5G3AyG2wY8yMMud6amybT7X1hPo+aaCwMhKfXT2P2966mKqibOrDPDXJLOw/ePEMHvnMZYMyXZT0ZVpJzlkxNbtPttHV5+f1c8oTaJWSKCYqatYAG6zHG4AbIoy5GnjCGNNkjGkGngCuARCRPODzwDcmaMeEmVaSg9ftCkXTO4kKS9ScbusN9X1ST83oSSWP41h4/0XTuWJ+BRX5mTR29tHnCySsh9loyfF6mFmWO/JAJW2YXprDiZZu+q1ED9trM7tc50k6MlFRU2mMqbUenwIqI4yZAoS3ta6xtgF8HfgOMDgfbwAicrOIbBGRLfX19RMwOTJZGW4e+MTFfPCSGVE/dqwpyPKQneEOempC3Yv1TnYMpIzHcTzYMVm1rd00dvYmpNu8ooyX6aW5+AOGky3BCsLHm+yie867QVUmzoiiRkQ2icirEX7WhI8zwXrrZojDRDruEmC2MeaPoxlvjLndGLPCGLOivDw2bsXFUwodmTUkIqG07tDykwM/RwJJGY/jeKiwata8eqINY0YuPqkoycT0krPTuo83dVOen5m0we5KbBnxdt4Yc+VQ+0SkTkSqjDG1IlIFnI4w7ARwRdjzqcAzwOuAFSJyxLKjQkSeMcZcgTJmZpXlsv90OxfMCDoPNKZmTMTF4ygiNwM3A0ybNm0i9kYV2zPzck1L8LmKGsVBTLcyV49aHppjTV2ODCNQosNE1yg2AmuB9dbvhyKMeQz4Zpir/irgS8aYJuCnACIyA/izCprxM39SPn99rT5UZyQ/SYunJYorr7ySU6dORdpVFP7EGGNEZDwex89Z83hIjDG3A7cDrFixYtTvEWtsT81OS9Sop0ZxEhX5mWR6XByz0rqPNXVx4cySBFulJIqJXvnWA/eJyIeBo8CNACKyAvi4MeYjxpgmEfk6sNl6zdcsQaNEkfmT8vEFDNuPt5Cd4SbDrSWIwtm0aVPE7SLSAvjT2eNYmpeJS4LLT+DMDEAlfXG5hOmlORys77SC3bupVk9N2jKhK58xptEYs8oYM9cYc6UtVowxW4wxHwkbd6cxZo7188sIxzlijFk8EVvSnQWTCgDYerRZg4THju1xhOE9jleJSLHldbwKeMwY81NjzGRjzAzg9cBrThI0EOz0XpaXSUevj6wMl/bKURzH+VOL2Hq0mZrmLgIGXX5KY/R2PkWYVZ5Lhlvo6vNrkPDYWQ+sFpH9wJXWc0RkhYj8AsAS7LbHcTMp5nG042omFWRpwTJn4k7HsgQ2K2eV0trdz6Y9dYCKmnRGRU2KkOF2Mbs8WBJcg4THhnocz8TVJHPhPWVYqkjjsgQrrRia32+pAVTUpDMqalKI+ZPyAbRDtzJm7AKOVYXZCbZEGSdFpHFZguqSHKYUZbP/dAdejysk0pX0Q0VNChESNeqpUcaIfRHQwnuOxROvsgSxLII6EWxvzdTibFwuXUJNV/SWPoVYEPLUqKhRxkZlyFOjoiZZGaoswW233XbW83QsSwCwclYJD2w/oUtPaY6KmhTCzoDS7CdlrNid3tVTk7wMVZbAwpfOZQkALpxZCmg8Tbqjy08pRFVhFv90yUxWL5yUaFMUh7FiRglvXTqFi2Zp0TKH0kIalyUAmFGawz9dMpM1SyYn2hQlgegtfQohInzluoWJNkNxIIXZGXzvXUsSbYYyfmoJliVI20Koev5TQEWNoihKKuA3xqwauNEYswU4qywBcOdQBzHGHAEcWZZAUUCXnxRFURRFSRFU1CiKoiiKkhKoqFEURVEUJSUQY5Kq1MCoEJF6gsFwAykDGuJsTjRwot1OtBkG2z3dGFMebyOGmcPgzL+t2hw/Itkd93msczhpcKLdMZvDjhQ1QyEiW4wxKxJtx1hxot1OtBmcYbcTbByI2hw/nGC3E2wciBNtBmfaHUubdflJURRFUZSUQEWNoiiKoigpQaqJmtsTbcA4caLdTrQZnGG3E2wciNocP5xgtxNsHIgTbQZn2h0zm1MqpkZRFEVRlPQl1Tw1iqIoiqKkKSpqFEVRFEVJCVJG1IjINSKyT0QOiMi6RNsTCRGpFpGnRWS3iOwSkc9Y228VkRMissP6eVOibR2IiBwRkVcs+7ZY20pE5AkR2W/9Lk60nTYiMj/s77lDRNpE5LPJ/LfWORxbdA7HHifMYXDuPNY5PIr3TIWYGhFxA68Bq4Eagl1o322M2Z1QwwYgIlVAlTFmm4jkA1uBGwh21O0wxvxPIu0bDhE5AqwwxjSEbfs20GSMWW+dwIqNMV9MlI1DYc2PE8BK4EMk4d9a53Ds0TkcW5wyh8G581jn8MikiqfmQuCAMeaQMaYPuBdYk2CbBmGMqTXGbLMetwN7gCmJtWpCrAE2WI83EDwpJCOrgIPGmKGqnyYDOocTg87h6OGIOQwpN491DoeRKqJmCnA87HkNST5BRWQGsBR40dp0i4i8LCJ3JpP7MAwDPC4iW0XkZmtbpTGm1np8CqhMjGkjchNwT9jzZPxb6xyOPTqHY4vj5jA4bh7rHB6BVBE1jkJE8oA/AJ81xrQBPwVmA0uAWuA7ibNuSF5vjFkGXAt8UkQuC99pguuYSbeWKSJe4Hrg99YmJ/ytkx6dw/FD53DscOA81jk8Aqkiak4A1WHPp1rbkg4RySD4JfqtMeYBAGNMnTHGb4wJAD8n6MZNKowxJ6zfp4E/ErSxzlqbtteoTyfOwiG5FthmjKmDpP5b6xyOMTqHY45j5jA4cx7rHB6ZVBE1m4G5IjLTUoQ3ARsTbNMgRESAO4A9xpjvhm2vChv2VuDVeNs2HCKSawXTISK5wFUEbdwIrLWGrQUeSoyFw/JuwlyeSfy31jkcQ3QOxwVHzGFw5jzWOTw6UiL7CcBKCftfwA3caYy5LbEWDUZEXg88C7wCBKzN/0bwH76EoNvwCPCxsDXShCMiswjeFQB4gLuNMbeJSClwHzANOArcaIxpSpCZg7C++MeAWcaYVmvbXSTp31rncOzQORwfnDCHwZnzWOfwKN8vVUSNoiiKoijpTaosPymKoiiKkuaoqFEURVEUJSVQUaMoiqIoSkqgokZRFEVRlJRARY2iKIqiKCmBihpFURRFUVICFTWKoiiKoqQEKmoURVEURUkJVNQoiqIoipISqKhRFEVRFCUlUFGjKIqiKEpKoKJGURRFUZSUwJNoA8ZDWVmZmTFjRqLNUFKArVu3NhhjyuP9vjqHlWiSiHmsc1iJJtGaw44UNTNmzGDLli2JNkNJAUTkaCLeV+ewEk0SMY91DivRJFpzWJefFEVRFEVJCVTUKIqiOIxHH32U+fPnM2fOHNavXz9ov4hkisjvROSAiLwoIjPC9n3J2r5PRK4O236Nte2AiKyLzydRlOiiokZRGPmELiKXicg2EfGJyDsG7FsrIvutn7Xxs1pJR/x+P5/85Cd55JFH2L17N/fccw9A1oBhHwaajTFzgO8B3wIQkYXATcAi4BrgJyLiFhE38GPgWmAh8G5rrKI4CkfG1MSTjTtP8sTuOr7/riW4XJJoc0ZNT7+frAz3Wdt8/gAAHrdq2XDCTuirgRpgs4hsNMbsDht2DPgg8K8DXlsCfBVYARhgq/Xa5njYHmvq2nro7PUxuSibB7ef4KEdJ7l6USXvu2i6zqME8dJLLzFnzhxmzZoFwE033cTLL79cNGDYGuBW6/H9wI9ERKzt9xpjeoHDInIAuNAad8AYcwhARO61xu5GcSyBgOFDv9rM8aauRJsCQG6mhz996vUxfQ8VNSPw02cOsqe2jevOq+KqRZMSbc6oeHJPHR+7ayvnVxfx1qVTyM/ysOtkG/dtOc65Uwq568MrAXjpcBO1rd0U53i5dG4ZwXNeWnIhI5zQjTFHrH2BAa+9GnjCGNNk7X+C4B3wPbE3O7rsON7Cr/5xmO3HWzh/ahE9/X427akjYEAEjIHKgkxu/VMjG54/yvzKfGZX5LL24hlU5A90FCix4sSJE1RXV4eeT506FcA7YNgU4DiAMcYnIq1AqbX9hbBxNdY27PFh21cOfG8RuRm4GWDatGkT+RhKHOjs8/HX1+pZWFXA7Iq8RJtDdkbsb4TGJGpE5Brg+4Ab+IUxZv2A/ZnAr4HlQCPwLmPMERFZDawn+MXrA/6fMeYp6zXLgV8B2cDDwGeMMWYiHypaHDjdzp7aNgB+9PQBVi+sTPoL/4HTHXz23h1ML82hqbOPf3/wVQDcLmF+ZT7P7m/gpcNNNHf18bG7toZe9/+uns8n3zAnUWYnmtAFwCLiCX0Mr50SaWCyXhBOtHTz9T/t5tFdp8jL9LByZgnPH2rEHzB87PLZzCrL5XBDJ8unF/PGBRU8tusUd71wlEMNHTyxp447/n6Y91w4nbUXT8frcfFKTSvLphdTlpdJT7+ftu5+KgpU9KQCxpjbgdsBVqxYkRTnaWVofP7gv+jGFVP54CUzE2xNfBi1qBmliz60jisiNxFcx30X0ABcZ4w5KSKLgcc4c+L/KfBR4EWCouYa4JGJfazosHFnLS6Bz145j+8+8RrP7m/gsnnBNPo+XwCvJ7nc7209/dx81xa8Hhe//vBKqgqyONbUhd8YinO8ZGe4ufTbT/HdJ/ZxvKmb+ZX5/Pi9y/jhU/v5n8f3cU5VPm9cUAlAr89PhsvlqCW3ZCdZLgi9Pj+1LT30+QPct/k4d71wFBH416vm8cFLZpKX6cG+r4gk4q9ZXMU1i6sAONzQyQ+f3M+vnz/Cnf84HBqT43Vz5TmV/G1/Pa3d/fzTJTP5f1fPH7QkOlqMMZxu7yXH6yYv05P0NxexZMqUKRw/fkZH19TUQPBmMZwTQDVQIyIeoJDgjaa93WaqtY1htisOpT8NQw7G4qkZ0UXPEOu4xpjtYWN2AdmWV6cEKDDGvGAd89fADSSBqDHG8KedJ7loVikfv3w297x0jP96ZC8Xzizh1ROtvP+Ol/jhu5dy5cLKRJsKBNdOP3fvDo41dvGbj6xkSlE2ADPKcs8a9+HXz+Jbj+4F4Hc3X8Scijy+9fbzOFjfwcfv2sabzp1EXpaHP247wfxJ+fzs/cvTYWlhuBP9aF57xYDXPhMVq8bJ8aYuHtt1ij5/gNJcLx6XC49bmFWWx9GmTm77yx5qW3sAcAm8delUPn/VvNCcgchiJhIzy3L57ruW8MVrF3D/1hoyPS7OqSrgd5uP8+iuU7xxfgVFORnc8ffD/PbFo5TlZTK3Io8LZpZwormbV060MqkgiwVVBVw0s4TZFXk0dfbx6olWth5tZkpRNhfMLOGnzxzkr6/VA1Cck8E1iydx8ewyyvIymVaaw+TCLNp7fZxs6WZmWS6ZnvGJJydwwQUXsH//fg4fPsyUKVO49957AVoGDNsIrAWeB94BPGWMMSKyEbhbRL4LTAbmAi8BAswVkZkE5/RNwHvi8oGUmNEfCN6ceFXURGQ0Lvqh1nEbwsa8HdhmjOkVkSnWccKPGdF1H0/8gaCgOdzQyc2XzcLrcXHbWxfz4Q1b+PQ929l2rIXufj9bjzUnjaj5302v8eTe03xtzSIumlU65Lj3XTSNXz13mCvmVbDSGpeV4eaXH7yQHz99gD9sraHXH2DVggqe2VfPmh/9g99+ZCWzyhO/HhtDNjP+E/pjwDdFpNh6fhXwpfEY0dXn4+4XjwHgEqE4N4PsDA/NXX109voAmFORxyVzyshwuzDG0Nbj41RrD6/VtbPzeAvPHWxkt7VkOhQLqwr43JXzyPAIS6qLmTlA+I6HyoKss5YvL5lTdtb+65dM5qk9p2no6OXlE608va+evEwP504p5GB9B5v21PGDAb6r/CwP7T3Bz52X6eGzV84lx+tm18k2HtpxknteOnM6yspw0dMfCD1eWl3Mgqp8SnO9NHb2sWhyIW9bOiUlPI8ej4cf/ehHXH311fj9fv7pn/6Jl19+uUdEvgZsMcZsBO4A7rICgZsIzmmMMbtE5D6CN6M+4JPGGD+AiNxCcD67gTuNMbsS8PGUKNLvsz01zp/3oyWugcIisojgktRV43htXOIRTrf1cNPtL3CooZPqkmzeZLnZ37igkn9ZPY//efw1cr1uyvIyOXC6I2Z2jIVHX63lB08d4J3Lp/L+i6YPOzY/K4On//UKsgcsA5TnZ3Lr9Yv44jUL8AUC5Gdl8OqJVt59+wt8/8n9fP+mpbH8CAnFEuCDTujhFwkRuQD4I1AMXCci/2mMWWSMaRKRrxMURgBfs4OGx0pnr59v/GXPiOMKsjxke900dfbR7z+jBLweF8umFfHFaxbwlvOqKM3z0tTZRyAAPT4/B093EDBwzeJJuON8cb94dhkXzz4jdJo7+8jP8oTc4u09/bx0uImTrT2U5HiZXZHLvIp8TrX18OLhRi6ZXXZWXE5Pv5+jjV00dPRypLGTg6c7Kcv3UlWYxc7jrWw71sy9Lx2nu98fEjx3vXCUN587ifL8zEHHcxpvetObeNOb3hR6/u///u8YY75iPzfG9ADvjPRaY8xtwG0Rtj9MMARASRF8AV1+Go7RuOiHWsdFRKYSvCh8wBhzMGz81BGOCcQvHuFnfz3E0aYuvn/TEt50bhUZYZPhk2+Yg4iwdFoRdz1/lH2n2mNlxqh5ra6dz9+3kyXVRXz9hsWjWjbI8Q79b8/2ugle12HxlELetmwK97x0nK9e10dJ7sAEi9Qh0gl9wEViM2fP1fBxdwJ3TtSG0lwvL98a1Pt+v6G5q4+uPj+led7g/8zAlqNNPL6rDoOhJDeTsjwv5fmZzK0IZiINXHYJ/1/Pq8yfqIlRo3jAXMrPymDVOYO9npOLsnnr0sF/9qwMN/Mn5TOf/EFeIXt8IGDo8wfI9Lh4cMcJ1j+yl28+HFx6FYGl1UVMLc5hQVU+/3TJzFC8j+0BK8zOiMpnVZREYd/0eNVTE5HRuOiHWsctAv4CrDPG/MMebIypFZE2EbmIYKDwB4AfjvfDTJTGjl7ufukoNyyZwpolg1fBRCTkYn/+YCOP766j1+dP6Pr9tx/dS1aGm5+9b/m4gzCH490rp7Hh+aP8YWsNH71sVtSPr5zB5RIKss5cSAde+AFWnVMZ8eKvDMblErJcwe/EW5dO5YYlU+js83OsMRhz9I8DDeysaWHjzpP8cdsJrlk8iZcON7H7ZBvtvT5uf/9yx5RxUJRIhAKFXeqpGcRoXPQMsY4L3ALMAb4iIvbd71XGmNPAJziT0v0IcQ4S7uj18ZUHX8UXMLgEen0BPvGG2SO+bk5FHv6A4UhDF/MnJeYOuLvPz7P7G3j3hdOYVBgbV/qCSQUsm1bEPS8d4yOXzkzrrBPF2YgIeZkeFk4uYOHkAj63eh4Af32tni/e/zI/evoAiycXsnpRJQ9sO0Fde2+CLVaUiWF7ajSmZghG4aKPuI5rjPkG8I0hjrkFWDwWO6LF8aYuPrJhCwfqO8jxumnv8fHmc6uYPYqgWHvMgdMdCRM1fz/QQK8vwOoYByu/Z+V0/vX3O3lox0luWJrwOG5FiSqXzyvnb194A70+P/lZGTR19vHAthP4/QPrLCqKs7A9NZr9lAa8dLiJj/9mKz5/gF996AJWTC/h6X2nWTmzZFSvn12ehwgJDRZ+ck8d+ZkeLpgxOpvHy5olk7n3pWOse+Bl5lXms3ByQUzfT1HijdfjCtWdsoOofQGtLac4G1/IU5M+oiZ9PmkYj+86xXt/8QJF2Rk8+MlLuHRuOdleN286t4rSvMxRHSPb62ZqcTYH6hMjagIBw6Y9p7lsfnnMiwBmuF385H3LKMr2cvNdW2jqHFjnS1FSB48lavwqahSH029lP2Wk0fJTyoqa0209EbOT9p1q57O/28HCyYX88ROXTKj+ypzyPPbXJSYD6uUTrTR09HLlORVxeb+K/Cx+9v7lnG7v5VP3bAs1x1SUVMP21PiTo1uLoowbu05NhnpqnM9Hf72Fq//3b7z1J//ggW019PT72XasmZvv2kJupofb37+cwpyJpWzOqcjjUENnQu7o7nr+KBlu4Yp58RE1AEuqi/jGDYv5x4FG/uuRvXF7X0WJJyFPjV9FjeJs7CVUDRR2OK/UtLKzppVrFk0K1XH50gOv0OsLkJ/l4VcfupDKKBTemlORR58vwNHGzrhW3N18pIk/bKvhn6+YHTHtN5bcuKKa3SfbuOPvh1k0uYC3LYtYukVRHIvG1Cipgh0onE6empQUNXe/dJSsDBffesd5FGR5eO5gI39+uZbFUwpYs2QKeZnR+dgrZwbbDPzl5Vo+tWpuVI45Ej5/gP948FUmF2bxqTcmpqv2l998Dntq21j3wCvMqcjjvKlFCbFDUWKBiOASjalRnI+d0p2RRnVqUu6TdvT6eGjHSa47bzKF2RmICJfMKeO/3nYu7105PWqCBoLNIi+eXcq9m4/H7QT4wLYT7D3Vzn+8ZeGwlYFjSYbbxU/eu4ySHO+oyvoritPwuFzqqVEcj8+ffr2fUkbU/O21et78g2e57od/p6vPz3tWxq4/VDjvXTmdEy3d/G1/fczeI2CdXH3+AD955gDnTinkmsWJrXRampfJO1dMZcuRJpo1G0pJMdwuwR/QYHjF2aTj8lPKfNJsr5uqwixml+dy82WzWFJdFJf3Xb2wkrI8b6i7crRp7Ohl1Xf/yto7X+Kel45xpLEr1IMq0Vx5TiUBA0/vO51oUxQlqnhcop4axfGElp/SyFOTMjE1F8woiXkRukh4PS7edUE1P3nmIH99rZ7L55VH7dj9/gCfvHsbJ1q6qWnu4q+v1TOvMo+rYlxBeLScO6WQ8vxMntxzWgOGlZTC7ZaQh1RRnIovoJ4aZRx88g1zmF+Zz6fu3saRhs6oHLPfH+DfHniFFw41sf5t53L3Ry9iXmUe665dgMuVHKrb5RJWLajgr6/V0+dTV72SOqinRkkF0rH3k4qaKJDj9fDzD6zA5RI+fe92zASLdrV29fPBX77E77fW8OlVc3nbsqlcMKOExz93OW9ckBxeGpsrz6mko9fH++54kXn//gjv+8WLPLmnLtFmKcqECMbUqKhRnE0opkazn5SxUl2Sw5ffdA4v17ROOMbkP/+0i5cON/Hf7ziPz1udhJOVS+aUUZDlYd+pdq47bzKH6jv48IYtHG/qSrRpijJuNPtJSQV8foPbJUnj3Y8HKmqiyA1LpzClKJsfPnVg3N6ao42dPLTzJGtfN4N3rqiOsoXRJ9vr5ql/vYIX/20V37nxfL785oUAdPf7E2zZ2BGRa0Rkn4gcEJF1EfZnisjvrP0visgMa3uGiGwQkVdEZI+IfCnuxitRxeXSOjWK8+n3B0IVstMFFTVRJMPt4uNXzGb7sRaeP9g4rmP87K8HcbuEj142K8rWxY6yvEyyMtzAmWqs/Q7rDSUibuDHwLXAQuDdIrJwwLAPA83GmDnA94BvWdvfCWQaY84FlgMfswWP4kzUU6OkAv1+k1ZBwqCiJuq8c/lUKvIz+eFTB8b82pMt3dy/tYZ3raiOShuHRODgDscXAgeMMYeMMX3AvcCaAWPWABusx/cDqySYW2+AXBHxANlAH9AWH7OVWKB1apRUoN8fSKt0blBRE3WyMtzcfNksnj/UyNajTWN67e1/O4Qx8LHLneOlGYgdZd/vvGaAU4DjYc9rrG0RxxhjfEArUEpQ4HQCtcAx4H+MMYP++SJys4hsEZEt9fWxK9aoTByPS/A5bw4ryln4AgE86qlRJsp7Vk6jOCeDH43BW1Pf3ss9Lx3jrUunMLU4J4bWxRaPFWXvQE/NRLgQ8AOTgZnAv4jIIGVqjLndGLPCGLOivDx69YyU6ON2CYEJZjHGgqamJlavXs3cuXNZvXo1zc3NEceJyFoR2W/9rA3bvtyK/TogIj+wPI2IyK0ickJEdlg/b4rTR1JiSL/f4FVRo0yUHK+Hj1w6i6f31fPqidZRveaOvx+m3x/gn6+YHWPrYovtqfE5z3V/AgiPzJ5qbYs4xlpqKgQagfcAjxpj+o0xp4F/ACtibrESM5K1Ts369etZtWoV+/fvZ9WqVaxfv37QGBEpAb4KrCQouL8qIsXW7p8CHwXmWj/XhL30e8aYJdbPwzH9IEpc6PcH0qpGDaioiRnvf9108rM8/Pjpkb01B063c9fzR3jzeZOZVZ4XB+tihx1T40DX/WZgrojMFBEvNhfUYgAAWO1JREFUcBOwccCYjYB91/sO4CkTTHM7BrwRQERygYuAvXGxWokJyVqn5qGHHmLt2uAUXLt2LQ8++GCkYVcDTxhjmowxzcATwDUiUgUUGGNesObtr4Eb4mK4khB8fqPZT0p0KMjK4IMXz+DRXafYX9ce2l7f3ntW1eGn953mrT9+jmyvm89dOTcRpkYVe/02GS8Iw2HFyNwCPAbsAe4zxuwSka+JyPXWsDuAUhE5AHwesNO+fwzkicguguLol8aYl+P7CZRo4k7SmJq6ujqqqqoAmDRpEnV1EQtdDhUfNsV6PHC7zS0i8rKI3Bnm2VEcTDBQOL0u8ynT+ykZ+dAlM/nFs4f58dMHePeF0/jNi8d49NVaAG674Vzaevr55sN7WDCpgJ+vXcGUouwEWzxxPA5N6QawXO4PD9j2lbDHPQTTtwe+riPSdsW5JNJTc+WVV3Lq1KlB22+77baznotINBvb/hT4OsFMvq8D3wH+aeAgEbkZuBlg2rRp0XpvJUaoqBkBEbkG+D7gBn5hjFk/YH8mQZfmcoKxBu8yxhwRETtD5ALgV8aYW8Je8wxQBXRbm66y4hIcT0mul/ddNI2fP3uYB3ecpCDLw/svmsFrde184Q/BG/lrF0/iOzeeT443NfSl27kp3YoSwuNy0eXzJeS9N23aNOS+yspKamtrqaqqora2loqKikjDTgBXhD2fCjxjbZ86YPsJAGNMyOUjIj8H/hzpwMaY24HbAVasWKFf8iTHFzBpF1Mz6itpWHGy1QTdlptFZKMxZnfYsFBxMhG5iWBxsncBPcB/AIutn4G81xizZZyfIan5+OWzqW/v5eLZZVx3/mSyvW58/gDff3I/OV4PH7tsVkqVsM4IBQrr+U5xLskaU3P99dezYcMG1q1bx4YNG1izZmApJSC4hPrNsCWkq4AvGWOaRKRNRC4CXgQ+APwQQESqjDG11vi3Aq/G9pMo8aDPp56a4QgVJwMQEbs4WbioWQPcaj2+H/iRiIgxphP4u4jMmbjJzqI0L5P/vWnpWds8bhf/ctX8BFkUW9xWSrcDs58UJYTHJfiTMKV73bp13Hjjjdxxxx1Mnz6d++67z96VIyK/MMZ8xBIvXycY3wXwtbC6SZ8AfkWwSOQj1g/At0VkCcHlpyPAx2L/aZRY4wsYsjJU1AxFpOCzlUONMcb4RMQuTtYwwrF/KSJ+4A/AN0yExkm6lusMHJz9pCghkjVQuLS0lCeffDLSri5jzEfsJ8aYO4E7Bw6yPOKDvOXGmPdH004lOfD5A2RkpUZow2hJBgn3XqtnzqXWT8QvlxYucwYeXX5SUgCPOzmXnxRlLPT5Taggarowlk87keJkQ2KMsQPV2oG7CS5zKQ7FDhRWUaM4GbfLpaJGcTw+7f00LBMpThYREfGISJn1OAN4Cxqg5mgy7JgaB6Z0K4qNW1SYK87HF0i/Lt2jXmyzYmTs4mRu4E67OBmwxRizkWBxsrus4mRNBIUPACJyBCgAvCJyA8GI/KPAY5agcQObgJ9H44MpicHt1pRuxfmop0ZJBfp86dcmYUwRROMtTmbtmzHEYZePxQYluQl5avSCoDiYYO8n9TYqzsYXCITOyelCen1aJeaEYmp0+UlxMG63oFNYcTr9fkOGJ708NSpqlKji0UBhJQXwuAS/emoUh9PvD2j2k6JMBJdLcInWqVGcjdslKswVx+PzG7ye9LrMp9enVeKCx+XSC4LiaDxJ2iZBUcZC0FOjy0+KMiGChcvUda84F5d6ahSHY4yxGlqm12U+vT6tEhfcLqFfl58UB6OeGsXp2KLcm2Yp3SpqlKiT4dYaH4qzsevUDFM7VFGSmn4rfU89NYoyQdxa40NxOHYcgmpzxanY3vJ0qyicXp9WiQueJO1wPBwico2I7BORAyKyLsL+TBH5nbX/RRGZEbbvPBF5XkR2icgrIpIVV+OVqHOmh5mKc8WZ2J4a7f2kKBPEaR2ORcQN/Bi4FlgIvFtEFg4Y9mGg2RgzB/ge8C3rtR7gN8DHjTGLgCuA/jiZrsQI21PjpHmsKOHYN5Zap0ZRJojH5aLfWReDC4EDxphDxpg+4F5gzYAxa4AN1uP7gVUiIgR7mL1sjNkJYIxpNMb442S3EiO027zidNRToyhRwoHVWKcAx8Oe11jbIo4xxviAVqAUmAcYEXlMRLaJyBeGehMRuVlEtojIlvr6+qh+ACW6hDw1DltGVRSbM6ImvS7z6fVplbiQZindHuD1wHut328VkVWRBhpjbjfGrDDGrCgvL4+njcoYUU+N4nTsuZtuXbpV1ChRx2kxNcAJoDrs+VRrW8QxVhxNIdBI0KvzN2NMgzGmi2AX+2Uxt1iJKW4rDsFh81hRQvT51FOjKFHB43KFXJ8OYTMwV0RmiogXuAnYOGDMRmCt9fgdwFMmWMTkMeBcEcmxxM7lwO442a3ECI9mPykOx/bUpFtMjSfRBiiph9OqsRpjfCJyC0GB4gbuNMbsEpGvAVuMMRuBO4C7ROQA0ERQ+GCMaRaR7xIURgZ42Bjzl4R8ECVq2MtPqmkUp+JL05gaFTVK1PG4ndc3xxjzMMGlo/BtXwl73AO8c4jX/oZgWreSIthxCOqpUZxKn11RWFO6FWVieFyu0F2CojgRd5LWqWlqamL16tXMnTuX1atX09zcHHGciKwVkf3Wz9qw7beJyHER6Rgwfsjikooz8fnTc/lJRY0SddwOW35SlIF4kjT7af369axatYr9+/ezatUq1q9fP2iMiJQAXwVWEqzB9FURKbZ2/8naNpCIxSUV52J7GdNt+Sm9Pq0SFzLcaZXSraQgLklOT81DDz3E2rVBx8vatWt58MEHIw27GnjCGNNkjGkGngCuATDGvGCMqY3wmqGKSyoOpc+nKd2KEhXUU6M4nTMxNck1j+vq6qiqqgJg0qRJ1NXVRRo2mmKSQ75mQHHJs9ACks4hXT01GiisRB2P26UBloqjOVOnJv7z+Morr+TUqVODtt92221nPRcR4u1MMcbcDtwOsGLFiuRSfMpZpGtF4TGJGhG5Bvg+wbTXXxhj1g/Ynwn8GlhOsDDZu4wxR0SklKBL8wLgV8aYW8Jesxz4FZBNMPvkM1b9D8WheFzOy35SlHDONLSM/3tv2rRpyH2VlZXU1tZSVVVFbW0tFRUVkYadINhY1WYq8MwIb2sXl6wZUFxScSj9oYaWuvwUkYl0MgZ6gP8A/jXCoX8KfBSYa/1cM5YPoCQfbpeEIu8VxYm4k7T43vXXX8+GDcHQlw0bNrBmzcC+q0Cw3tJVIlJsBQhfZW0bjqGKSyoOxT4Hez3p5akZy6cddydjY0ynMebvBMVNCBGpAgqs4DVD0Mtzwzg+h5JEZLh0+UlxNp4kTelet24dTzzxBHPnzmXTpk2sW7fO3pUjIr8AMMY0AV8nWBByM/A1axsi8m0RqbHG14jIrdbr7wBKreKSnwfWoTia/lCdmvTy1Ixl+SlS8NnKocZYVVrtYLOGYY5ZM+CYEQPaRORm4GaAadOmjcFsJd64ndf7SVHOIlkbWpaWlvLkk09G2tVljPmI/cQYcydw58BBxpgvAIM6yQ9XXFJxJiFRk2YxNY75tNrh2DlkaEyN4nDsKqx+XUZVHIp9DvaqqBmSiXQyHu6YU0c4puIw3C6XxtQojsauLK/iXHEq/T7bU5Ney09jETUT6WQcEasIVJuIXGQVevoA8NAYbFKSkAy3aEyN4mhCnhoVNYpD6Q+kZ/bTqGNqJtLJGEBEjgAFgFdEbgCuMsbsBj7BmZTuR6wfxcFo9pPidEK9nzQBSHEo/f4AGe741zJKNGOqUzPBTsYzhti+BVg8FjuU5MauU2OMSbsvlJIanMl+Uo+j4kx8/kDadegGBwUKK87BjrZX173iVELZT+pxVBxKv9+kXYduUFGjxIBkTYdVlNFiB1eqMFecSnD5Kf0u8en3iZWYk+HQC4KIXCMi+0TkgIgMKj4mIpki8jtr/4siMmPA/mki0iEikSpnKw5ChbnidHx+k3aZT6CiRokBdjNAJ7nuJ9gGxOa7aKB7SuAWZwpzRbHpD6inRlGigidJ++aMwLjbgABYGX2HgV3xMVeJJXaApXpqFKcSjKlJv0t8+n1iJebYLk+HXRAitQEZ2LLjrDYgQCvBfjl5wBeB/xzuDUTkZhHZIiJb6uvro2a4En3cbs1+UpxNMPtJl58UZcJ40i8e4Vbge8aYjuEGaasP53AmpTvBhijKOEnXQOEx1alRlNHg0L45Y2kDUjOgDchK4B0i8m2gCAiISI8x5kcxt1qJCW6tU6M4nHRN6VZRo0Qde/mp31kXhFAbEILi5SbgPQPG2G1AnufsNiCX2gNE5FagQwWNs7EDhdPI26ikGL40DRRWUaNEnTN3uc65IEy0DYiSWrhcgkucNYcVJZx+X3qmdKuoUaKOvfzU77CAhIm0AQkbc2tMjFPijsflUk+N4lj6AwHyMtLvEp9+vikl5ngc6KlRlIG4XDqHFefi05RuRYkODk3pVpSz8LhcjiogqSjh9GtKt6JEB48DKworykDcLiFgdA4rzuJ4Uxf3vHSMxs6+tPTUpN+CmxJzznhqnBVToyjheFyic1hxFHVtPbztp89R394LQFVhVoItij8qapSoEyq+p54axcG4XaIxNYpj6PX5+dhdW+ns9fGHf34dU4pyqMjPTLRZcUdFjRJ1nJjSrSgD8bhEhbniGH781AF2HG/hZ+9bxvLpJYk2J2Gk34KbEnPsdVwNFFacjNudXJ6apqYmVq9ezdy5c1m9ejXNzc0Rx4nIWhHZb/2sDdt+m4gcF5GOAeM/KCL1IrLD+vlIjD+KEmVau/v55T+O8KZzJ3HN4qpEm5NQVNQoUccdWn7SeATFubhFkkqYr1+/nlWrVrF//35WrVrF+vXrB40RkRLgqwRbd1wIfFVEiq3df7K2ReJ3xpgl1s8vYmG/Ejvuev4I7b0+PvmGOYk2JeGoqFGiToamdCspQLLF1Dz00EOsXRt0vKxdu5YHH3ww0rCrgSeMMU3GmGbgCeAaAGPMC8aY2jiZq8SJrj4fd/z9MG9cUMGiyYWJNifhqKhRoo7bTunWzBHFwXhcrqQSNXV1dVRVBZcWJk2aRF1dXaRhU4DjYc9rrG0j8XYReVlE7heR6qEGicjNIrJFRLbU19ePwXolVtz94jGau/rVS2OhgcJK1NHsJyUVcLviv/x05ZVXcurUqUHbb7vttrOeiwgiUSus9ifgHmNMr4h8DNgAvDHSQGPM7cDtACtWrNAveILp6fdz+98OcfHsUpZPLx75BWnAmESNiFwDfJ9gw79fGGPWD9ifCfwaWA40Au8yxhyx9n0J+DDgBz5tjHnM2n4EaLe2+4wxKybweZQkQCsKK6mAxy344+xt3LRp05D7Kisrqa2tpaqqitraWioqKiINOwFcEfZ8KvDMcO9pjGkMe/oL4NujtVdJLL/fWsPp9l7+96YliTYlaRj18pOIuIEfA9cCC4F3i8jCAcM+DDQbY+YA3wO+Zb12IcGOxosIru/+xDqezRusADUVNClAKFBYRY3iYBLhqRmO66+/ng0bNgCwYcMG1qxZE2nYY8BVIlJsBQhfZW0bEhEJT5e5HtgTHYuVWNLvD/CzZw6ybFoRr5tVmmhzkoaxxNRcCBwwxhwyxvQB9wIDv1VrCLouAe4HVknQR7oGuNcY02uMOQwcYOgofMXhZFgxNX7NflIcjCfJAoXXrVvHE088wdy5c9m0aRPr1q0DYMuWLQDTAYwxTcDXgc3Wz9esbYjIt0WkBsgRkRoRudU69KdFZJeI7AQ+DXwwjh9LGSd/3H6CEy3d3PLGOdFcinQ8Y1l+ihSAtnKoMcYYn4i0AqXW9hcGvNYOXjPA4yJigP+z1mwHISI3AzcDTJs2bQxmK/HGrctPSgqQbJ6a0tJSnnzyyUHbV6xYAXDUfm6MuRO4c+A4Y8wXgC9E2P4l4EvRtHUg3X1+Xj3ZyorpxXoBjgL+gOGnzxxk0eQC3jA/4jJk2pIM2U+vN8YsI7is9UkRuSzSIGPM7caYFcaYFeXl5fG1UBkTHl1+UlKAZEvpdiqBgOGTd2/jnT97nt++eCzR5qQEf3mllsMNndzyBvXSDGQsouYEEJ7qN9XaFnGMiHiAQoIBw0O+1hhj/z4N/BFdlnI8Z7p0O2f5SUSuEZF9InJARNZF2J8pIr+z9r8oIjOs7atFZKuIvGL9jpg1ojgPt8ulwjwKfG/Tazy19zTTSnK4deMunt53mtrWbmpbuznV2qPCcYwEAoYfP3WAORV5XL1oUqLNSTrGsvy0GZgrIjMJCpKbgPcMGLMRWAs8D7wDeMoYY0RkI3C3iHwXmAzMBV4SkVzAZYxptx5fBXxtQp9ISThO89SEBcGvJrg0ullENhpjdocNCwXBi8hNBIPg3wU0ANcZY06KyGKCQZmjqQuiJDkelxBwyBxOJh7bdYppJTksmJTPT545yA+fOsCNK6by5Tcv5IYf/4MP/XLzWeOnFmfz3pXT+cilM0MtVpSh2bSnjn117XzvXefjcqmXZiCjFjVWjMwtBE/abuBOY8wuEfkasMUYsxG4A7hLRA4ATQSFD9a4+4DdgA/4pDHGLyKVwB8t95kHuNsY82gUP5+SAFwuwSWOamgZCoIHEBE7CD5c1KwBbrUe3w/8SETEGLM9bMwuIFtEMo0xvbE3W4klyRZT4wQCAcOtG3dR29rDtJIcjjV1sWbJZL5+w2IyPW5+97GLeHrvaYz1Z+3zB3jklVN869G9uAQ+dvnsxH6AJMcYw4+ePsC0khyuO29yos1JSsZUp8YY8zDw8IBtXwl73AO8c4jX3gbcNmDbIeD8sdigOAOPy0W/c4rvTSQIviFszNuBbUMJGg12dxbB7CfnLKEmAy6X8PCnL+X3W4/zx+0n+eI1C/j45bNCcR8V+Vm864Kz5/4HXjeDG3/2PHe/dIyPXjor6b0P97x0jNv/dij0XATevmwqn7hidszjW57d38DLNa3819vOxaNerYjoX0WJCe40uyCIyCKCS1IfG2qMBrs7C/XUjI/iXC83XzabRz5zKf88ygv9e1ZO42hjF88dbBxxbCLp7PWFvErnTink3CmFVORn8t+P7eNT92ynubMvZu8dCBi++8RrTCrI4m3LdIV7KLRNghITPG5xkqdmLEHwNQOC4BGRqQSD3D9gjDkYe3OVeJBsdWpSmWsWT6L4Txn8+vkjNHT0cqSxkzVLpjCzLDfRpp3F3S8eo6Wrnzs/eAHLpgXbEhhj+L+/HeJbj+7l8d11rDl/Ml9+8zkU5Xij+t73b61hx/EWvvPO88n0uEd+QZqiokaJCQ67IEwkCL4I+Auwzhjzj/iZrMQat8ul/cviRFaGm3csn8rPnz3M47uDjTr/d9N+3rtyGre99dwEWxekp9/P7c8e4pI5pSFBA8E+XB+/fDZvXFDBXc8f5Xebj7P5SBO/WLuCORX5UXnvlq4+1j+6lwtmFKuXZgR0+UmJCR63c9JhjTE+wA6C3wPcZwfBi8j11rA7gFIrCP7zgJ32fQswB/iKiOywfrQaVgrgdjkq2N3xfPSyWXzw4hnc9eELefHfVvHuC6v57YvHeHrv6USbBsB9W45T3947ZDfseZX5fP2Gxdxz80V09Pq48f9eoKvPF5X3/r+/HaK1u5+vrVmsdWlGQD01SkzwuMRRdWrGGwRvjPkG8I2YG6jEHbfLhd+oqIkXFflZ3Hr9otDz/7x+MS8dbuKrG3fxutmlZGUkbsmlzxfg//56iOXTi0fss7R8ejE/fs8y3nX7C/x5Zy03XlA97PjR8Nirp7hkThnnVBVM+FipjnpqlJgQ7HCsFwTFuThsCTXl8HpcfH3NYo41dfGDJ/cn1JYHx9hn6cKZJcypyOO3L028gvKh+g4ONXRy5TnqAB4NKmqUmOBxuejXC4LiYNwO8zamIhfPKeNdK6r5yTMHeXzXqYTY4A8YfvrXgyyeUsAV80aXtSgivOfCaew83sKuk60Tev8n9wSX3964QEXNaFBRo8SEdEvpVlIP9dQkB/+5ZhHnTy3kc7/bwf669ri/v91n6ZNXjK3P0tuXTSXT4+IHT+6naQKp3k/sqWPBpHymFueM+xjphIoaJSZ4XI5K6VaUQbjdWqcmGcjKcPOz9y8n2+vh5ru20trdH7f3nkifpcKcDP7p9TN5bFcdF/3Xk/zPY/sGtd3wBwz/89g+brl7G61dgz9XS1cfW482c+U5lRP6HOmEBgorMUFjahSno56a5KGqMJufvm8Z7/n5C3zm3u385L3LyPHG7vJ1uKETAfbVtbOvrp3/fdeScVU6/uI1C3jr0in85OkD/OjpA+w91X5WSvbvtxzn6X31uAR2n2zjF2tXMKs8D4B+f4BvPboXf8CwSuNpRo2KGiUmeLTDseJw3KKemmTighkl3Hr9Ir78x1dZ+c0ned9F0/l/V82PeluFXp+fd/7sORo6+sjP9DC9NIe3nFc17uPNq8zne+9awtJpxXztz7vZtKcutM/tEr5xw2LmT8rn43dt5YYf/4Mfv3cZRdlevvnwHp4/1MjHLp/FkuqiKHyy9EBFjRITnJbSrSgDcbuCq/OBgEn6fkTpwntXTmd+ZT6/ePYwP33mIDNLc6OSMh3OY7vqaOjo453Lp7LjeAufWz1vwn2WRIS1F8/g2sWTaA5bZirOzaAiPwuABz95CR/99Rbef8dLAOR43fz3O87jnSui+/lSHRU1SkzQvjmK0/G4g0LGFzB4VdQkDStmlLBsWjE3/t/zrH90L1ctqoxqS4K7XzzKtJIcvvX286IuZisKsqgoyIq4r7okh/v/+WK+98RrVBdn87blUynIyojq+6cDGiisxIQMt0s9NYqjcVsXNI2rST5cLuHrNyymtbufb/xlDyZKRRIP1nfwwqEmbrqwOiHeubxMD//xloV88JKZKmjGiYoaJSa4NchScTgel+2pUXGejJxTVcDNl83i/q01fPreHfT0+yd8zLtfPIbHJbxj+dQoWKgkAl1+UmJChqbDKg5HPTXJzxeunk9BVgbffmwvGS7hu+9aMu5jNXf2ce9Lx3jzeVWhOBfFeainRokJwWqsejFQnMsZT43O42RFRPjnK2bzvpXT+csrtXT0jr+B5C+fO0Jnn59PXBG5YaXiDFTUKDEhmNKtbnvFudjZT8nkqWlqamL16tXMnTuX1atX09zcHHGciKwVkf3Wz1prW46I/EVE9orILhFZHzY+U0R+JyIHRORFEZkRn08UHdYsmUyvL8Cm3XUjD45Ae08/v/rHYa5eVMn8SflRtk6JJypqlJjg0eUnxeHYWbzJJGrWr1/PqlWr2L9/P6tWrWL9+vWDxohICfBVYCVwIfBVESm2dv+PMWYBsBS4RESutbZ/GGg2xswBvgd8K9afJZosm1bMlKJsNu48Oa7X3/XCUdp6fNzyhrlRtkyJNypqlJigy0+K00lGT81DDz3E2rVrAVi7di0PPvhgpGFXA08YY5qMMc3AE8A1xpguY8zTAMaYPmAbYEfErgE2WI/vB1bJWBodJRiXS3jLeVX87bV6WrrG1mepq8/HL549zGXzyjl3amGMLFTihYoaJSZkuFxJdTFQlLGSjDE1dXV1VFUFq9tOmjSJurqIyy1TgONhz2usbSFEpAi4Dnhy4GuMMT6gFSgdeGARuVlEtojIlvr6+gl9lmhz3fmT8QUMf3q5dkyvu+el4zR19vHpN2osTSqg2U9KTAg2A9SYGsW5nMl+iu88vvLKKzl16tSg7bfddttZz0VkTF2jw17nAe4BfmCMOTSW1xpjbgduB1ixYkXyqD1g0eQCllQX8f1N+1mzZPKo6rz0+vzc/reDrJxZwooZJXGwUok1Y/LUiMg1IrLPCiZbF2H/kMFmIvIla/s+Ebl6tMdUnEmGAysKx2J+K84lUZ6aTZs28eqrrw76WbNmDZWVldTWBj0RtbW1VFREbHR4AgivrT/V2mZzO7DfGPO/kV5jiZ5CoDFqHyoOiAhfX7OYxs5evvv4a2ft6+z18bvNx3j01TNenNbufj6yYQt1bb18epXG0qQKo/bUiIgb+DGwmqA7c7OIbDTG7A4bFgo2E5GbCAabvUtEFgI3AYuAycAmEZlnvWakYyoOxO1y0e8L0NU3/hTLaJLhdpExTP+WWMxvY8zEq4EpCcP21HT2+pJiHrtEuP7669mwYQPr1q1jw4YNrFmzJtLQx4BvhgUHXwV8CUBEvkFQsHxkwGs2AmuB54F3AE+ZaJXpjSPnTi3kfSun8+vnj7D16JnMsCMNnbRb6d4fvXQmcyry+OkzBznR0s36t53LJXPKEmWyEmXGsvx0IXDAdleKyL0Eg8vCT/prgFutx/cDP7KCzdYA9xpjeoHDInLAOh6jOKbiQDIzXHT2+Vn4lccSbQoA//7mc/jIpbOGGxKL+f18VD+EElcyM9wAvP2nyfFvXL2wkvXr1nHjjTdyxx13MH36dO677z57d46I/MIY8xFjTJOIfB3YbO37mrVtKvBlYC+wzVq6+pEx5hfAHcBd1txtIijSHcm/Xj2fXp+fho4zAcPnVOXzjuXV/GnnSX7+7GEA5lfm85sPr2TlrEGhQ4qDGYuoiRR8tnKoMcYYn4jYwWZTgBcGvNYOXBvpmEAwQA24GWDatGljMFtJBB943XRKcrwEkuRm78KZI66Xx2p+h9A57CxWzizh62sW0dWXHA636aW5lJaW8uSTT0ba3WWMCXlfjDF3AneGDzDG1AARg3CMMT3AO6NobsIozM7g2+84P+K+C2eWcOncMopyvFwwo3hcMUlKcuOYQOFkDlBTBlNVmM1HLxvWM5J26Bx2FlkZbt7/uhmJNkOJMlctmpRoE5QYMpZA4ZGCz84aMyDYbKjXjuaYihIPYjG/FUVRlDgyFlGzGZgrIjNFxEtwzXXjgDF2sBmcHWy2EbjJyh6ZCcwFXhrlMRUlHsRifiuKoihxZNTLT1YMwS0EI+vdwJ3GmF0i8jVgizFmI0MEm1nj7iMYdOkDPmlnhkQ6ZvQ+nqKMjljNb0VRFCV+iAOz9lixYoXZsmVLos1QUgAR2WqMWRHv99U5rESTRMxjncNKNInWHHakqBGReuBohF1lQEOczYkGTrTbiTbDYLunG2PK423EMHMYnPm3VZvjRyS74z6PdQ4nDU60O2Zz2JGiZihEZEsi7ronihPtdqLN4Ay7nWDjQNTm+OEEu51g40CcaDM40+5Y2qwNLRVFURRFSQlU1CiKoiiKkhKkmqi5PdEGjBMn2u1Em8EZdjvBxoGozfHDCXY7wcaBONFmcKbdMbM5pWJqFEVRFEVJX1LNU6MoiqIoSpqiokZRFEVRlJQgZUSNiFwjIvtE5ICIrEu0PZEQkWoReVpEdovILhH5jLX9VhE5ISI7rJ83JdrWgYjIERF5xbJvi7WtRESeEJH91u/iRNtpIyLzw/6eO0SkTUQ+m8x/a53DsUXncOxxwhwG585jncOjeM9UiKkRETfwGrAaqCHYx+fdxpjdCTVsACJSBVQZY7aJSD6wFbgBuBHoMMb8TyLtGw4ROQKsMMY0hG37NtBkjFlvncCKjTFfTJSNQ2HNjxPASuBDJOHfWudw7NE5HFucMofBufNY5/DIpIqn5kLggDHmkDGmD7gXWJNgmwZhjKk1xmyzHrcDe4ApibVqQqwBNliPNxA8KSQjq4CDxpihqp8mAzqHE4PO4ejhiDkMKTePdQ6HkSqiZgpwPOx5DUk+QUVkBrAUeNHadIuIvCwidyaT+zAMAzwuIltF5GZrW6UxptZ6fAqoTIxpI3ITcE/Y82T8W+scjj06h2OL4+YwOG4e6xwegVQRNY5CRPKAPwCfNca0AT8FZgNLgFrgO4mzbkheb4xZBlwLfFJELgvfaYLrmEm3likiXuB64PfWJif8rZMencPxQ+dw7HDgPNY5PAKpImpOANVhz6da25IOEckg+CX6rTHmAQBjTJ0xxm+MCQA/J+jGTSqMMSes36eBPxK0sc5am7bXqE8nzsIhuRbYZoypg6T+W+scjjE6h2OOY+YwOHMe6xwemVQRNZuBuSIy01KENwEbE2zTIEREgDuAPcaY74Ztrwob9lbg1XjbNhwikmsF0yEiucBVBG3cCKy1hq0FHkqMhcPybsJcnkn8t9Y5HEN0DscFR8xhcOY81jk8OlIi+wnASgn7X8AN3GmMuS2xFg1GRF4PPAu8AgSszf9G8B++hKDb8AjwsbA10oQjIrMI3hUAeIC7jTG3iUgpcB8wDTgK3GiMaUqQmYOwvvjHgFnGmFZr210k6d9a53Ds0DkcH5wwh8GZ81jn8CjfL1VEjaIoiqIo6U2qLD8piqIoipLmqKhRFEVRFCUlUFGjKIqiKEpKoKJGURRFUZSUQEWNoiiKoigpgYoaRVEURVFSAhU1iqIoiqKkBCpqFEVRFEVJCVTUKIqipDBWF+TTIhKxFL0E+YGIHLC6Ji+Lt42KEi1U1CiKoqQ2vwKuGWb/tcBc6+dmgh2UFcWRqKhRFEVJYYwxfwOG6wW0Bvi1CfICUDSg4aCiOAZPog0YD2VlZWbGjBmJNkNJAbZu3dpgjCmP9/vqHFaiyQTn8RTgeNjzGmvboAaDInIzQW8Oubm5yxcsWDDOt1SUs4nWudiRombGjBls2bIl0WYoKYCIHE3E++ocVqJJvOaxMeZ24HaAFStWGJ3DSrSI1hzW5SdFUZT05gRQHfZ8qrVNURyHihpFUZT0ZiPwASsL6iKg1RgzaOlJUZxAzJafRKQa+DVQCRjgdmPM9weMuQJ4CDhsbXrAGPO1WNmkKIqSbojIPcAVQJmI1ABfBTIAjDE/Ax4G3gQcALqADyXGUkWZOLGMqfEB/2KM2SYi+cBWEXnCGLN7wLhnjTFviaEdcaPPF8AfMGR73Yk2RVGUCdDR6yPX60ZEEm3KhDHGvHuE/Qb4ZJzMUZSYErPlJ2NMrTFmm/W4HdhDMKI+ZfnC/Tt58w+epbvPn2hTFEUZJy8dbmLZ157gsV11iTZFUZQxEpeYGhGZASwFXoyw+3UislNEHhGRRcMc42YR2SIiW+rr62Nl6rjp8wXYtOc0hxo6+e4T+xJtjqIo46C+vZdb7t5Gnz9AfUdvos1RFGWMxFzUiEge8Afgs8aYtgG7twHTjTHnAz8EHhzqOMaY240xK4wxK8rL415WZES2HG2io9fH3Io87vj7YXYeb0m0SYqijAGfP8Cn79lOU2cfAH5/IMEWKYoyVmIqakQkg6Cg+a0x5oGB+40xbcaYDuvxw0CGiJTF0qZY8fTe03jdLu768EpK8zL53qbXEm2Soihj4HubXuP5Q4186U3nAOA3CTZIUZQxEzNRI8EIuzuAPcaY7w4xZpI1DhG50LKnMVY2xZKn99WzclYJkwqzeNuyKfx9fwPN1h2foijJzVN76/jx0we56YJqbrogWLLFH1BPjaI4jVh6ai4B3g+8UUR2WD9vEpGPi8jHrTHvAF4VkZ3AD4CbrEh8R3G8qYsDpzu4Yn4FANedNxlfwPDYrlMJtkxRlJE43tTF5363k4VVBdx6/SLcrmDGky/guFORoqQ9MUvpNsb8HRg2H9IY8yPgR7GyIV4881owcPkN84OxPosmFzCjNIc/v1zLTRdOS6RpiqIMQ6/Pzyfv3kbAGH76vmVkZbjpt2JpAipqFMVxaEXhKLDtaDMV+ZnMLMsFQER4y3mTee5gAw2aQZFSiMgREXnF8jxusbaViMgTIrLf+l2caDuV0fH1P+/m5ZpWvvPO85leGvz+ukU9NYriVFTURIGdx1s4v7rorEJdbz6vioCBTbu11kUK8gZjzBJjzArr+TrgSWPMXOBJ67mS5Dy4/QS/eeEYN182i6sWTQptd7kEEfXUKIoTUVEzQVq7+znU0Mn5UwvP2j6/Mh+PSzjW1JUgy5Q4sgbYYD3eANyQOFOU0fBaXTtfeuAVLpxRwheunj9ov8cl6qlRFAeiomaCvHqiFYDzphadtd3lEkpyvTR2aAZUimGAx0Vkq4jcbG2rDGsAeIpgv7NBJHsByXSho9fHx3+zldxMNz96z1I87sGnQZcIfhU1iuI4Ytn7KS3YWdMCwHkDPDUApXmZNHZqTE2K8XpjzAkRqQCeEJG94TuNMUZEIl4NjTG3A7cDrFixQq+YCcAYw5ceeIUjDZ385iMrqSjIijjO41JRoyhORD01E+Tl463MKM2hKMc7aF9Znpd69dSkFMaYE9bv08AfgQuBOhGpArB+n06chcpw/Pr5o/xp50n+5ar5XDx76Dqfbl1+UhRHoqJmguysaRm09GRTlpdJo2Y/pQwikmt1nEdEcoGrgFeBjcBaa9ha4KHEWKgMx/9v787D5CqrxI9/T1X1mk7T6WyErCxtICBZJ4CoMGxCZgzwgEAMI4w4yKbgNgPqj3FcngFnBEdZHBQeUUISiAxEBZmAMMgMINlDglkIZCN00mt6r67q8/vj3uoUnV5qu3Wrqs/nefrpWm5XndyuVJ8673nfd93uRr7/+y2ce+I4bjzr+EGPDVqlxpi8ZMNPaTjQ0sn+5s5+h54ARltPTaEZD/yXO8stBDyuqn8QkTeBJ0TkOmAXcIWPMZp+NLaFuXnJWsZXlvKjK2YSCAy6hBbBQIBo/q0DasywZ0lNGjbucZqEZ06u6vf+0RUldHRHaeuKMKLETnW+U9WdwMx+bq8Hzs1+RCYRPT3KbcvXU9caZsWNZ/Q7VNxXKCBEbfMnY/KODT+lYd2eRkIB4ZRj+q/UjKlw3jytWmOMf+57aQf/s+0g/7xwxoBDxX1ZT40x+cmSmjSs39PESRMqKSsO9nv/mIoSAOpsBpQxvnh1ex33vrCNS2dP5LNJbFkSDAg9NvxkTN6xpCZF0R5lw55mZg0w9AQw2q3U1LVYUmNMtu1v7uDLy9ZRM66CH1x6yodW/B6KVWqMyU+eJjUiMllEXhKRLSKyWURu7ecYEZGfiMgOEdkoInO8jClTdhxopbUrwuwpVQMeE6vU1LfZ8JMx2dQd7eGWx9fR2R3lgcVzKS9OrqctGJCC2iZBRC4Uka3u++wR23iIyBT3vXqd+z68wI84jUmX15WaCPA1VZ0BnA7cLCIz+hxzEVDjfl0PPOhxTBmxbncjALOnDLx3YfWIWE9N7lVqOsJR7nhqE7vq2/wOxZiMu/u5v7BmVyN3X3YqJ4yrSPrnnW0SejyILPtEJAjcj/NeOwNY1M/78LeBJ1R1NnAV8EB2ozQmMzxNalR1v6qudS+3AG8DE/scdjHwK3W8DlTFFjLLZet2N1FVXsS00eUDHlNaFGRkSYi6HGwUXrF2L0v/vJun1u7zOxRjMuq5Tfv5xavvcu3HpvHpmcek9BgFtk3CfGCHqu5U1TCwDOd9N54Cle7lo4D3sxifMRmTtZ4aEZkGzAbe6HPXRGBP3PW9HJn45Ny+Oev3NDGrz87c/RkzsoS6HKvURHuUh/+0E4C1bsXJmELwbl0b31ixkVmTq/jmgpNSfpxQsKCSmkTeY78DXC0ie4FngS/190C59j5sTF9ZSWpEpAL4DXCbqh5K5TFU9SFVnaeq88aOHZvZAJPU2hVh24EWZk8eeOgpJhcX4Hvh7Vreq29nYlUZ63c3FdKbtxnGOsJRbnxsDUVB4f7FcygOpf72NgwbhRcBv1TVScAC4NcicsQJzKX3YWP643lSIyJFOAnNElV9qp9D9gGT465Pcm/LWe/VtaEK048eOeSxoyuKc65S8/Cf3mXSqDJuPa+Glq4I2w+0+B2SMWm785m32Frbwr1XzmJiVVlajxUsrOGnRN5jrwOeAFDV14BSYODNsYzJUV7PfhLgYeBtVb1ngMNWAp9zZ0GdDjSr6n4v40rX3sZ2ACaNGvqNc0xFSU7NfjrU2c2f32vg8rmTmD+tGoC1u5r8DcqYNC1/czdPrtnLl86p4ezp49J+vALb++lNoEZEjhWRYpxG4JV9jtmNuyq2iJyEk9TY+JLJO15Xas4E/g44R0TWu18LROQGEbnBPeZZYCewA/g5cJPHMaVtb2MHQEKfBkdXlNDYHiYSzY2ZFLGtHeZOHcXU0eWMHlHMml3WV2Py1+b3m7nzmc18omYMt55bk5HHLKSkRlUjwC3A8ziTNZ5Q1c0i8l0RWege9jXgH0RkA7AUuFbVVh80+cfTDYlU9VVg0E5a9z/OzV7GkWn7mjooLw5SVV405LFjKopRhYb2MONGlmYhusFt2NsEwKmTnCbnOVNHWbOwyVvNHd3ctGQto8qL+fGVswgOsVFlogqtp0ZVn8X5ABl/251xl7fgfAg1Jq/ZisIp2NfYwcSqsoRWKO1dgC9HmoXX7W7iuLEjOKrMScjmTBnFu3VtObmWjjGDUVW+8eQG9jV2cP/i2Yx2/69lQsi2STAmL1lSk4J9TR1MTKCfBg4vwNeQA301qupMRY/b1O9Et9l5V0O7T1EZk5qf/2kn/72lltsvOpG5U6sz+tjBgBCxXbqNyTuW1KRgX1NHQk3CAOXuZpft4aiXISXk/eZO6lq7mBW3tUOlW7E51NHtU1TGJO/P7zZw9x+2suCjR3Pdx4/N+OPbhpbG5CdLapLU2hWhqb2biVUDryQcr7TISWo6u/1PatbvbgJgZlyl5qgyp63qUGfEh4jyk4gE3T1yfudeP1ZE3nD31VnuzjAxHjnQ0sktj69lSnU5d192alIbVSYqFAgUVE+NMcOFJTVJ2heb+ZRgpaY05CQ1XRH/Zz9t2NtEcTDASRMqe2+LVWqarVKTjFtxZpHE3A3cq6onAI04a34YD0SiPdy6dD2HOrt58Oo5jCwdulk/FYECmv1kzHBiSU2S9jU5vSeJLu5VWuSc4lyo1Gx+v5mTJoz80EqrlaU2/JQMEZkE/A3wC/e6AOcAK9xDHgUu8SW4YeCeVdt4bWc937/ko5x4dOXQP5CikCU1xuQlS2qSFKvUJNpTU5JDw08Nbd2Mq/zwtPLSoiAloYAlNYn7MfCPQKz0NhpoctcCgQH2LgPbNyddL75dywMvv8NVfzWZy+dO8vS5CmxDS2OGDUtqkrS3qYPiYICxCU4fLXGrIrkw/HSoo7u3MhOvsqyIQ52W1AxFRP4WOKCqa1L5eds3J3V7Gtr5yvL1nHxMJd9ZeLLnzxcKCJEe///PGmOS4+nie4VoX2MHE6pKCSS4yFdJKIBIblRqDnV0U1l25K+8sjRkPTWJORNYKCILcJaRrwT+A6gSkZBbrcn5vcvyTVckyk1L1qLAA4vn9DbfeykYFHJkEXBjTBKsUpOkZKZzA4gIJaGA70lNtEdp6Yr0W6k5qqyIQx02+2koqnqHqk5S1Wk4++f8UVUXAy8Bl7uHXQM841OIBel7v9vCpn3N/OgzM5k6ekRWntPZ0NKyGmPyjSU1SdrrriacjNKiIJ3d/r5BtrpTtmMrCcerLCuySk16/gn4qojswOmxedjneArG0+v28djru7n+k8dxwclHZ+15C22bBGOGCxt+SkJHOMrBli4mj0psjZqY0lCQroi/lZpY0lLZT1JzVFkR79a1ZTukvKaqLwMvu5d3AvP9jKcQbatt4Y6nNjF/WjX/+KnpWX3uUEDosaTGmLzjaaVGRB4RkQMi8tYA958tIs1xO3jf2d9xuWJvozOde8roJJOaooDvlZpYI3BlaX89NVapMbmltSvCjY+tYURJiPs+O5tQMLtFZavUGJOfvK7U/BK4D/jVIMf8SVX/1uM4MmJXvZvUVCeb1AR976k5NESl5lBHN6rqyeqsxiRDVbnjqU28W9fGY1847YhlCLLBtkkwJj95+vFHVV8BGrx8jmza3ZBaUlNSFKTT5yndsUpN/z01IXrU+XRsjN9+9doufrvhfb52wXQ+dvwYX2KwSo0x+SkXGoXPEJENIvKciAy4AEUuLFy2u6GdipJQ787bicqF2U9D9dSA7f9k/LdudyPf//0WzjlxHDeedbxvcQQDgirWV2NMnvE7qVkLTFXVmcBPgacHOjAXFi7b3dDO5OrypIdoSouCdPk+/OQkLAP11AA0t1tfjfFPY1uYm5esZXxlKfdcMTPhtaC8EHKfO2pDUMbkFV+TGlU9pKqt7uVngSIR8afenIDdDe1MTXLoCaA0lBuNwgGBipIjk5rDlRpLaow/enqU25avp641zAOL51BV7u9G57GEyrZKMCa/+JrUiMjR7oaAiMh8N556P2MaSE+PsruhPemZT+BWanJgSndlWVG/VSbbqdv47b6XdvA/2w5y56dncOqkKr/D6a3UWF+NMfnF6yndS4HXgOkisldErhORG0TkBveQy4G3RGQD8BPgKtXcrPceaOkiHOlhciqVmlyY0j3Avk8QV6mxpMb44NXtddz7wjYunT2RxadN8TscAIIB562xUCo1InKhiGwVkR0icvsAx1whIltEZLOIPJ7tGI3JBE+ndKvqoiHuvw9nynfO21XvLE6X0vBTUZBOnys1hzoj/e77BId7aqxR2GTb/uYOvrxsHSeMreAHl56SM0sKBN0wCiGpEZEgcD9wPs4u8m+KyEpV3RJ3TA1wB3CmqjaKyDh/ojUmPX43CueNVKdzQ+6sU9PfdG6ACrd52IafTDZ1R3u45fF1dHVHefDquZQX584C50F3sb8C2al7PrBDVXeqahhYBlzc55h/AO5X1UYAVT2Q5RiNyQhLahK0u6GdgMDEJDazjIk1Cvs5stY8yPBTMCCMLA3Z8JPJqrue+wtrdjVy12WncsK4Cr/D+ZBYT01h5DRMBPbEXd/r3hbvI8BHROR/ReR1EbmwvwfKhaU1jBmMJTUJ2t3QzjFVZRSlsFx7SVEQgC4fF+A71DlwUgPOEJQlNSZbntu0n4dffZdrPzaNT888xu9wjhCUWKNwYWQ1CQgBNcDZwCLg5yJS1fegXFhaw5jBWFKToH0p7M4dUxJyTnOXj83ChzoG7qkBd6sEm9JtsmDnwVa+sWIjsyZX8c0FJ/kdTr+ChVWp2QdMjrs+yb0t3l5gpap2q+q7wDacJMeYvGJJTYIa2sOMqShJ6WdLeys1/vTVhCM9dHRHB+ypAWerBOupMV7rCEe5aclaioLC/YvnUBzKzbegYKCgKjVvAjUicqyIFANXASv7HPM0TpUGd62wjwA7sxijMRmRm+8oOaixLZz09ggxsaTGr2ndvTt0D5LUOJta2uynwYhIqYj82d3WY7OI/It7+7Ei8oY7XXa5+4fD9KGqfPvpt9ha28K9V85KufKZDcECWnxPVSPALcDzwNvAE6q6WUS+KyIL3cOeB+pFZAvwEvANVc3JNcOMGUzuTDfIYdEepamjm1EpJzVO7ujXtO7eHbqH6KmxSs2QuoBzVLVVRIqAV0XkOeCrwL2qukxEfgZcBzzoZ6C56InVe/jN2r18+ZwTOHt6bs8YLrRtEtwV25/tc9udcZcV53X81SyHZkxGWaUmAc0d3ahCdfnAScFgSkOxSo1PSY27/oz11KRHHa3u1SL3S4FzgBXu7Y8Cl2Q/uty2+f1m/t8zm/lEzRhuPe8jfoczpNg2CZFoYSQ1xgwXltQkoKEtDJBGpcbf4adYBWawnpqK0hDt4SiRaEH0EHhGRIIish44AKwC3gGa3BI/9D9dNvazw3I6bHNHNzc+tpbq8mJ+fOWs3qGdXBYqoOEnY4YTS2oSEEtqUu+pcYef/KrUJDD8VJoD087zgapGVXUWzgyS+cCJSfzssJsOq6p8/ckNvN/Uwf2LZzM6xWb7bAsW2PCTMcOFJTUJ6K3UpLhz8OFKjV/DT0M3CpeG/E288o2qNuE0VJ4BVIlIbGyvv+myw9bP/7STVVtquWPBScydWu13OAkrpEZhY4YTrze0fEREDojIWwPcLyLyE3fWyEYRmeNlPKlqbHeSmtEVqSU1vevU+FQFic1qGmz4ySo1QxORsbEFyUSkDGcvnbdxkpvL3cOuAZ7xJcAc88bOeu7+w1YuOuVoPn/mNL/DSYolNcbkJ69nP/0SZ8PKXw1w/0U4CzzVAKfhzBg5zeOYkpavlZrV7zVwy+PriKpSHAz0Jlf98bualCcmAI+6GwQGcKbG/s6dBrtMRL4PrAMe9jPIXHCwpYsvLV3HlOpyfnj5qTmzUWWiYisKW1JjTH7xepfuV0Rk2iCHXAz8yp1O+LqIVInIBFXd72VcyWpsC1NeHOz9w5+skt4p3dmtgmzZf4gPDnVy0oRK5k4ZNegflsN9P1apGYiqbgRm93P7Tpz+GgNEoj18eek6DnV28+jn5zNykF6uXBUKxhbfs6TGmHzi9zo1A220dkRSIyLXA9cDTJkyJSvBxTS0h1Ou0kDc0E6WqyAdYef5VtxwBiNKBv9Vl8Smnfu0lo4pHPes2sZrO+v598/M5KQJlX6Hk5JgwEnyeyypMSav5E2jsJ8zR9JZTRj8W6em3U1qEqkwxapJfu5PZfLfi2/X8sDL73DlvMlcPneS3+Gk7PCGlpbUGJNP/E5qEtlozXcN7amvJgxQFBQCkv2hnc7uKCWhQELrgvT21FilxqRoT0M7X1m+nhkTKvmXi0/2O5y0HG4UtiTfmHzid1KzEvicOwvqdKA51/ppABrauhidRlIjIpQWBbO+oWV7OEpZcWJ9QLFqUraHyExh6IpEufnxtSjws6vnptx/litiPTW2FqUx+cXTnhoRWYqz8+sYEdkL/DPO0vKo6s9w9iJZAOwA2oG/9zKeVDW2dafVUwNOJSTblZqO7ijlCf5xKbFGYZOG7/1uCxv3NvPQ381lyuhyv8NJW0AKapduY4YNr2c/LRrifgVu9jKGdHVForR2Ragekd4MjpJQIOs9NR3JVGp616mxSo1JztPr9vHY67v54lnHccHJR/sdTkbYNgnG5Ce/h59yXlO7sxpvOj014FZqsjyluz0cSWL4ySo1Jnnbalu446lNzJ9WzTcumO53OBlji+8Zk58sqRlC775PaQ4/+VKp6Y5SXpRYMc4W3zPJau2KcMNjaxhREuK+z84mFCyctxNLaozJT4XzLuSRxjR36I5xemqyP/xUmmClpsQqNSYJqsodT23ivbo2frpoNuMqS/0OKaNCtqGlMXnJkpohNMT2fUo7qQlkfQ2YZBqFQ8EAoYDYlG6TkF+/vovfbnifr10wnTOOH+13OBkXsEqNMXnJkpohNGSwUpPLU7rBjdEqNWYI6/c08b3fbeHcE8dx41nH+x2OJ2KVmkjUkhpj8oklNUOIJTVVg+xwnYjSUPandHd2J5vUBKxSYwbV0BbmpsfWML6ylB9dMbO3olFoYj01PTb8ZExesaRmCLWHuqgeUZx2E6QfCUN7OPHhJ3D2f7JGYTOQnh7ltuXrqWsN8+DiuVSl2Tyfy2JJTaFskyAiF4rIVhHZISK3D3LcZSKiIjIvm/EZkymW1Axhx4EWThhbkfbjZLtRWFXpSLJSU+JD34/JH/e9tINXth3kzk/P4KOTjvI7HE8V0uwnEQkC9wMXATOARSIyo5/jRgK3Am9kN0JjMseSmkGoKls/aKFmfPpJjTOlO3sJQ1ekB1WSG34KZb/vx+SHV7fXce8L27h09kQWnzbF73A8F3J36S6EpAaYD+xQ1Z2qGgaWARf3c9z3gLuBzmwGZ0wmWVIziAMtXRzqjPCR8SPTfqxsV2piO3SXJTH8VFqU3cQr34jIZBF5SUS2iMhmEbnVvb1aRFaJyHb3+yi/Y82k/c0d3LpsHTXjKvjBpacgUph9NPFirUIFMvw0EdgTd32ve1svEZkDTFbV32czMGMyzZKaQWyrbQHITKWmKOhWT7LzJtnhJlDlSc5+sp6aQUWAr6nqDOB04Ga3jH878KKq1gAvutcLQne0h5uXrKWzO8oDi+dSXuzpzio5Q0QIBmRY7NItIgHgHuBrCRx7vYisFpHVBw8e9D44Y5JkSc0gttW2AmSoUuOc6q4sbZXQEY4AUJbEHyFnKwdLagaiqvtVda17uQV4G+cT78XAo+5hjwKX+BKgB/712b+wdncTd112KieMSz+5zydBkULZpXsfMDnu+iT3tpiRwCnAyyLyHk7CvrK/ZmFVfUhV56nqvLFjx3oYsjGp8TSpGarjXkSuFZGDIrLe/fqCl/Eka3ttC9UjihlTUZL2Y5WG3A0jszS80xF2nieZ4aeSkDUKJ0pEpgGzcZoqx6vqfveuD4DxA/xMXn3KfXbTfh7533e55oypfHrmMX6Hk3UFVKl5E6gRkWNFpBi4ClgZu1NVm1V1jKpOU9VpwOvAQlVd7U+4xqTOs6Qm0Y57YLmqznK/fuFVPKnYVttCTYY+ncYadtu7Ixl5vKG0u5WapIefrFIzJBGpAH4D3Kaqh+Lvc3ee73eMMZ8+5e482Mo/rtjIrMlVfOtv+vtvW/hCgcKo1KhqBLgFeB6nuviEqm4Wke+KyEJ/ozMms7wcIO/tuAcQkVjH/RYPnzNjVJXtta1cOmfi0AcnIJZcxBp4vRbrqSm1RuGMEpEinIRmiao+5d5cKyITVHW/iEwADvgXYfo6wlFuWrKWUFC4f/EcikPDc5Q6UDiVGlT1WeDZPrfdOcCxZ2cjJmO84OW71ZAd967LRGSjiKwQkcn93A9kv3S/v7mTlq4INRnop4HDw0Ad2Upqwsk3Ctvie4MTZ9rPw8DbqnpP3F0rgWvcy9cAz2Q7tkxRVb799FtsrW3hx1fOYmJVmd8h+SYUkEKZ/WTMsOH3R7DfAtNU9VRgFYebLY+Q7dJ9bObTRzI0/BSbNZKtSk17KklNUSBrjcx56kzg74Bz4vrAFgB3AeeLyHbgPPd6Xlr+5h5+s3YvXzqnhrOnj/M7HF8FA2LbJBiTZ7wcfhqq4x5VrY+7+gvghx7Gk5Q9jR0AHDtmREYer7wkNvyUnZ6a2PBTUuvUhIKEIz309GjB7umTDlV9FRjoxJybzVi88Na+Zu5cuZmPnzCGW8+t8Tsc3wUDYhtaGpNnvKzUDNpxD+D2H8QsxGliywn1rV1A+rtzx8QqJtkefkp2l27I3rRzkzuaO7q5aclaqsuL+Y+rZvVuEzCcObOfLKkxJp94VqlR1YiIxDrug8AjsY57YLWqrgS+7HbfR4AG4Fqv4klWQ1uYqvIiitLcyDKmvMg51W1ZbhROdkVhSH53b5PfVJWvP7mB95s6WHb96YzOwBIGhSAYEKI2/GRMXvF0edChOu5V9Q7gDi9jSFV9a5jqDFVp4HDFpCNLw0/t4SjFwUBSu4vHKjU2rXt4eeiVnazaUsv/+9sZzJtW7Xc4OSNojcLG5B2/G4VzVn1bF2NGZO4Ta7andHd2R3srL4kqcafu2gJ8w8cbO+v54fNbWfDRo/n8mdP8DienhAJC1HpqjMkrltQMIOOVmqLsJjXt4UjS+/RYpWZ4OdDSyS1L1zGlupy7Lzt1WGxUmYyA2PCTMfnGkpoB1LeFGV2RuaQmEBBKiwK9vS5e6+juSWo6N8T31FilptBFoj3cunQ9LZ3dPHj1HEaWFvkdUs4JBa1R2Jh8Mzy23E1StEdpbA8zOoOVGnDWqsnalO5wJKnVhOHw/lS2AF/hu/eFbby2s55//8xMTjy60u9wclIwELCkxpg8Y5WafjS2h1El47NAyouDWV18L9lKTYlN6R4W/viXWu5/6R0WzZ/M5XMn+R1OzgoKltQYk2csqelHQ1sYIKM9NeAmNV3Zm9Kd7LTsWKOwVWoK156Gdm5btp4ZEyr550+f7Hc4OS0UCBApkL2fjBkuLKnpR5278F4me2oAyopDtGerpyYcTWqNGohrFLakpiB1RZyNKhX42dVzkx6eHG4CAbCcxpj8YklNP2KVmtEZnNINUF4UzNo6NR3dyQ8/xRqFbUp3Yfrub7ewaV8z91wxiymjy/0OJ+dZpcaY/GNJTT/qW92kJsOVmmz31CQ7/HR4mwSr1BSa/1q3lyVv7OaLnzyO82eM9zucvGDbJBiTfyyp6Ud9WxgRGFWe6eGnYNb2fuoMRykrSm5y2+GeGvt0Wki21bbwzafeYv6x1XzjU9P9Didv2DYJxuQfS2r6Ud/axajy4oxv6petSo2q0t4dpaw4uV+v9dQUntauCDc8toYRJSF+umh2UttmDHe2S7cx+cfe4frR0JbZ1YRjyotDtGWhpyYc7SHao0mvKFwUDBAMiK0oXCBUldt/s5H36tr46aLZjK8s9TukvBKy4Sdj8o7nSY2IXCgiW0Vkh4jc3s/9JSKy3L3/DRGZ5nVMQ6lvzfzCe+BUarIx/NQZdoaPUpndUhoK2PDTAETkERE5ICJvxd1WLSKrRGS7+32UnzHG+9Vru/jdxv187YLpnHH8aL/DyTsBG34yJu94mtSISBC4H7gImAEsEpEZfQ67DmhU1ROAe4G7vYwpEXVtXRlvEgYnqYn0KGGPF7dr7470Pl+ySouC1ig8sF8CF/a57XbgRVWtAV50r/tu3e5Gvv/7LZx74jhuPOt4v8PJS4VUqUngw+VXRWSLiGwUkRdFZKofcRqTLq+3SZgP7FDVnQAisgy4GNgSd8zFwHfcyyuA+0REVP37iNTQFs74dG5w1qkBZw2Z4pB3+WSsGpRKUlOSoUrN+j1N/PyVnfTkyCfdy+dO4tyT0pv1o6qv9FNJvBg42738KPAy8E9pPVGaGtrC3LxkLeMrS/nRFTMJZLg3bLgISmEkNXEfLs8H9gJvishKVY1/H14HzFPVdhG5EfghcGX2ozUmPV4nNROBPXHX9wKnDXSMqkZEpBkYDdTFHyQi1wPXA0yZMsWreOmO9tDU3u1ZpQacSspReLeBYKwZOaXhp6JgRhqFn163jz9s/oDjx45I+7Eyoam926uHHq+q+93LHwADZk7ZeA339Ci3LV9PXWuYFTeeQVWGZ/ANJwU0pXvID5eq+lLc8a8DV2c1QmMyJG82tFTVh4CHAObNm+fZO01je2zhPQ+TGo/7apo7nD/gVWXJJ04lRcGMVGoa2sJMHlXGf3/lrLQfK1+oqorIgK/NbLyG73tpB69sO8gPLj2FUydVefEUw0YoKEQKI6lJ5MNlvOuA5zyNyBiPeN0ovA+YHHd9kntbv8eISAg4Cqj3OK4BxRbeq/Zi+MmtnHi9/1N9GntXlRYFMtJT09geZpQHiWEOqhWRCQDu9wN+BfLq9jrufWEbl86eyGfne1fNHC4CIvQURlKTMBG5GpgH/NsA918vIqtFZPXBgwezG5wxCfA6qXkTqBGRY0WkGLgKWNnnmJXANe7ly4E/+tlPE0tqxngy/OQUxto9ntbdmEZSUxIKZGSbhIa2MNXDY+gj/vV7DfCMH0Hsb+7gy8vWUTOugh9cegoi1keTrlCgYCo1iXy4RETOA74FLFTVrv4eSFUfUtV5qjpv7NixngRrTDo8TWpUNQLcAjwPvA08oaqbReS7IrLQPexhYLSI7AC+is+zR+rbYptZetEoHOup8b5SI0JK/RSlRcGMrFPT2FZ4lRoRWQq8BkwXkb0ich1wF3C+iGwHznOvZ1V3tIdbHl9HV3eUBxbPTXp9ItO/YCBQKD01Q364FJHZwH/iJDS+VRuNSZfn736q+izwbJ/b7oy73Al8xus4EnWwxUlqxnqQ1IwocZIar9eqaWwLU1VWlNKKyBUlIXbXt6cdQ0O7NwsY+klVFw1w17lZDaSPu577C2t2NfLTRbM5YVyFn6EUlGCAgkhq3AkYsQ+XQeCR2IdLYLWqrsQZbqoAnnSrfLtVdeGAD2pMjrKPdH3Ut4UpCgqVZZk/NeVFseEnb5OahjSqJGNHlnCgpd/Kc8I6wlE6u3syvneWOdJzm/bz8Kvvcu3HpvHpmcf4HU5BKaBKTSIfLs/LelDGeMC2SeijrqWL0SNKPOlJiA0/dXjcU+Oss5NaQjG+spTWrgitXanH2NAe6+nxbtq6gXfr2vjGio3MmlzFNxec5Hc4BcfpqbHVtY3JJ5bU9FHfFvZkjRo4PKW7LRuVmhSrJONGOsNuBw51pvz8sUZlq9R4pyMc5cbH1lAUFO5fPMfTxRyHq0BA6FFnDy1jTH6wd8I+6lq7PGkShrgp3V4nNe2pJ2axTQ/TGYJqSGP2lRmaqvLtp99ia20L9145i4lVZX6HVJBCbk9aoQxBGTMcWFLTR31r2JPp3OB88istCng6/KSqzsyjFKsk4yudhK42nUqNO/xUaLOfcsXyN/fwm7V7+dI5NZw9fZzf4RSsWKO9bWppTP6wpCaOqlLX2sUYjyo14KxV42Wl5lBnhEiPplwlGTvSqdQczESlxoafMu6tfc3cuXIzn6gZw63n1vgdTkELWqXGmLxjSU2c1q4IXZEezyo14AxBeTmlO92hn8rSEKVFgfQqNW1hAgKVKWzTYAbW3NHNTUvWUl1ezI+vnJXSlH2TuNjwU4EswGfMsGBTuuPEVhP2YofumBElQU8rNekmNSLCuJGl6fXUtIepKi+2P7oZpKp8/ckNvN/UwfIvnu5Z35c5LODOgBxuWyUYk8+sUhOnrtX5Qz5mpHd/MMqKQ7R52FOTiSbd8ZUlaVZquhlVblWaTHrolZ2s2lLLHQtOYu7Uar/DGRZCQavUGJNvLKmJU9fq3Q7dMeUeDz+ls+9TzLiRpRw4lF5Pjc18ypw3dtbzw+e3suCjR/P5M6f5Hc6wYT01xuQfS2rixCo1Yz2s1JQXezv8lM4O3THjKtNbVbixPfXZV+bDDrR0csvSdUypLufuy061jSqzKCiW1BiTbyypiRPrqfGyylBWHKTDww0tG9vDlBYF0trUcNxIZ1XhthRXFbZKTWZEoj18eek6Wjq7efDqOYwstSG9bLJKjTH5x7OkRkSqRWSViGx3v48a4LioiKx3v1b2d0y21Ld1UVVeRFHQu1xvZGmIhrawZ2+U9a3htKdSx9aqSaVao6pOpcaSmrTds2obr+9s4PuXfJQTj670O5xhJ2izn4zJO15Wam4HXlTVGuBF93p/OlR1lvvl666wda1dnvbTAHyiZizNHd38aftBTx6/sT1MdZpT0se5a9Wk0izc2hWhO6q2Rk2aXny7lgdefoer/moyl8+d5Hc4w5JVaozJP14mNRcDj7qXHwUu8fC5MqKuNez5VNlzTxrHqPIinlyz15PHT2ffp5h0KjWNbd3A8FtNWEQuFJGtIrJDRAZK4BOyp6Gdryxfz8nHVPKdhSdnKkSTpFDAeXu0pMaY/OFlUjNeVfe7lz8Axg9wXKmIrBaR10XkkoEeTESud49bffCgN1WOutYuxnqc1JSEglw8ayKrNtfS5G4nkEnp7NAdMy62/1MKlZrhuEO3iASB+4GLgBnAIhGZkcpjdXZHuXHJGhR4YPEcSt39wkz2xUahLakxJn+ktfieiLwAHN3PXd+Kv6KqKiIDvTNMVdV9InIc8EcR2aSq7/Q9SFUfAh4CmDdv3hGPta+pg7W7GpP+N8Q7cKiLj5/gfYXhM/Mm8cv/e497V21j3rSh1xzpUWXNrkb+7516jh87grOnj6OipP9fXV1rV9pVksrSECWhAK/vbOjd4DIRPaps2tsMDLsduucDO1R1J4CILMOpVG5J9oG+97stvLXvED//3Dymjh6R4TBNMoJupealrQd452Crz9HA0UeV8lcJvF8YM5ylldSo6nkD3ScitSIyQVX3i8gE4MAAj7HP/b5TRF4GZgNHJDVDWbe7kS8tXZfsjx1hWhb+kJx8zFHMnFzFo6/t4tHXdiX0M6VFAeYfO5oNe5p5fnPtoMdOrS5PKz4R4fixFbzwdi0vvD34c/WntCjA5DRjyDMTgT1x1/cCp/U9SESuB64HmDJlyhEPsuNAC8ve3MMXzzqO82cMVNg02RLb6f7fnt/qcySO82eMt6TGmCF4uU3CSuAa4C73+zN9D3BnRLWrapeIjAHOBH6YypOd9ZGxvPDVT6YRrrMsejaSGoDHv3Aa+5s7Ej5+0qhySouCqCq7G9rpjvb0e1ym/g3Lv3h6Co3CQnlxkFHlxZQV27BJX0NVG08YN5KnbvwYJx9jM51ywZwpo3j1n/6aTg+XYEhGOss0GDNcePm/5C7gCRG5DtgFXAEgIvOAG1T1C8BJwH+KSA9Of89dqpp0yR5gZGlRXq3jMaIkxAnjRib9cyKSlWGJfDufPtsHTI67Psm9LWkzJ1dlIh6TIZNGDauKozF5z7OkRlXrgXP7uX018AX38v8BH/UqBmOy5E2gRkSOxUlmrgI+629Ixhgz/NiKwsakSVUjwC3A88DbwBOqutnfqIw5bKglB0SkRESWu/e/ISLTfAjTmLTZIK0xGaCqzwLP+h2HMX3FLTlwPk4T+5sisrLPUP91QKOqniAiVwF3A1dmP1pj0mOVGmOMKWy9Sw6oahiILTkQL36x1BXAuWK7p5o8lJeVmjVr1tSJSH9zoccAddmOJwPyMe58jBmOjHuqH0EM8hqG/Dy3FnP29Bf3YK/jRJYc6D1GVSMi0gyM7vs88csSAF0i8lZyoXsuF3+nFlNipmfiQfIyqVHVsf3dLiKrVXVetuNJVz7GnY8xQ+7EPdBrGHInxmRYzNnjZ9zxyxLk4vmzmBKTqzFl4nFs+MkYYwpbIksO9B4jIiHgKKA+K9EZk0GW1BhjTGHrXXJARIpxlhxY2eeY2GKpAJcDf1RV2/TK5J28HH4axEN+B5CifIw7H2OG/Ig7H2Lsy2LOnqTidntkYksOBIFHVHWziHwXWK2qK4GHgV+LyA6gASfxyWgcWWIxJaZgYxJLxo0xxhhTCGz4yRhjjDEFwZIaY4wxxhSEgklqhloGPBeIyGQReUlEtojIZhG51b39OyKyT0TWu18L/I61LxF5T0Q2ufGtdm+rFpFVIrLd/T7K7zhjRGR63PlcLyKHROS2XD7X9hr2lr2Gk3rulLdVEJE73Nu3isinshjTV93X5UYReVFEpsbdF407X32bpL2M6VoRORj33F+Iu+8a93W3XUSu6fuzHsVzb1ws20SkKe4+r87RIyJyQAZY00gcP3Fj3igic+LuS/4cqWref+E0v70DHAcUAxuAGX7H1U+cE4A57uWRwDZgBvAd4Ot+xzdE7O8BY/rc9kPgdvfy7cDdfsc5yOvjA5wFynLyXNtrOCux22s48eca9LUI3AT8zL18FbDcvTzDPb4EONZ9nGCWYvproNy9fGMsJvd6q0/n6Vrgvn5+thrY6X4f5V4e5XU8fY7/Ek7TuGfnyH3cTwJzgLcGuH8B8BwgwOnAG+mco0Kp1CSyDLjvVHW/qq51L7fgbH440d+o0hK/tPqjwCX+hTKoc4F3VHWgFXxzgb2G/WGv4SOls63CxcAyVe1S1XeBHe7jeR6Tqr6kqu3u1ddx1uPxUjr/Zz8FrFLVBlVtBFYBF2Y5nkXA0jSfc0iq+grOjLqBXAz8Sh2vA1UiMoEUz1GhJDX9LQOe02+0brl2NvCGe9MtbuntkVwqgcdR4L9FZI04S6UDjFfV/e7lD4Dx/oQ2pKv48H/eXDzX9hr2nr2GE5PIa/FD2yoAsW0VvHodJ/u41+F8+o8pFZHVIvK6iFySgXiSieky93e1QkRiiyB6cZ4Sfkx3aO5Y4I9xN3txjhIxUNwpnaNCSWryiohUAL8BblPVQ8CDwPHALGA/8CP/ohvQx1V1DnARcLOIfDL+TnXqhTm3PoA4i40tBJ50b8qHc53z7DWcPfYaTo6IXA3MA/4t7uap6mwL8FngxyJyfJbC+S0wTVVPxak0PDrE8dlyFbBCVaNxt/l1jjKqUJKaRJYBzwkiUoTzx2CJqj4FoKq1qhpV1R7g52SmXJtRqrrP/X4A+C+cGGvdMiHu9wP+RTigi4C1qloLOX2u7TXsMXsNJyydbRW8eh0n9Lgich7wLWChqnbFbo/73e8EXsapMHoek6rWx8XxC2Buoj/rRTxx+lb+vDpHiRgo7pTOUaEkNYksA+47d8z5YeBtVb0n7vYJcYddCuTUzrciMkJERsYuAxfgxBi/tPo1wDP+RDioD40b5/C5ttewh+w1nJR0tlVYCVwlzuyoY4Ea4M/ZiElEZgP/iZPQHIi7fZSIlLiXxwBnAluyFFP872ohTg8aOKs7X+DGNgrn9fi81/G4MZ2I03j7WtxtXp2jRKwEPufOgjodaHaHhFM7R150O/vxhdNBvQ2n+/tbfsczQIwfxylvbwTWu18LgF8Dm9zbVwIT/I61T9zH4XTSbwA2x84vzhj6i8B24AWg2u9Y+8Q9AufT41Fxt+XsubbXsKdx22s4uec94rUIfBcnYQAoxRkO24GTtBwX97Pfcn9uK3BRFmN6AaiNe12udG//mHu+Nrjfr8tiTP/qvt42AC8BJ8b97Ofd87cD+PtsxONe/w5wV5+f8/IcLcUZJu3G6Yu5DrgBuMG9X4D73Zg3AfPSOUe2TYIxxhhjCkKhDD8ZY4wxZpizpMYYY4wxBcGSGmOMMcYUBEtqjDHGGFMQLKkxxhhjTEGwpMYYY4wxBcGSGmOMMcYUhP8Pk2dBYxwBei0AAAAASUVORK5CYII=\n",
-      "text/plain": [
-       "<Figure size 648x1080 with 15 Axes>"
-      ]
-     },
-     "metadata": {
-      "needs_background": "light"
-     },
-     "output_type": "display_data"
-    }
-   ],
-   "source": [
-    "def pplot(dd):\n",
-    "    x = dd.shape[0]\n",
-    "    fix = int(x/3)+1\n",
-    "    fiy = 3\n",
-    "    fig, axs = plt.subplots(fix, fiy, figsize=(3*fiy, 3*fix))\n",
-    "    \n",
-    "    for i in range(x):\n",
-    "        axs[int(i/3)][i%3].plot(dd[i])\n",
-    "        \n",
-    "pplot(dd)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 30,
-   "id": "68b447f0",
-   "metadata": {
-    "tags": []
-   },
-   "outputs": [
-    {
-     "data": {
-      "image/png": "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\n",
-      "text/plain": [
-       "<Figure size 648x1080 with 15 Axes>"
-      ]
-     },
-     "metadata": {
-      "needs_background": "light"
-     },
-     "output_type": "display_data"
-    }
-   ],
-   "source": [
-    "conv = Conv2D(1, (3,3), input_shape=train_shape, activation='relu', padding='same')\n",
-    "c = conv(d)\n",
-    "cc = pd.DataFrame(c.numpy().reshape(d.shape[1], d.shape[2]))\n",
+    "def predict(model, entry):\n",
+    "    print(model.predict(entry)[0].shape)\n",
+    "    prediction = np.argmax(model.predict(entry), axis=-1)\n",
+    "    p = [0 for i in range(52)]\n",
+    "    p[prediction[0]] = 1\n",
+    "    return np.array(p)\n",
     "\n",
-    "pplot(cc)"
+    "p = predict(model, np.array([x]))\n",
+    "lb.inverse_transform(p)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 31,
-   "id": "20c0eb4e",
+   "execution_count": 103,
+   "id": "ea020844",
    "metadata": {},
    "outputs": [
     {
      "data": {
-      "image/png": "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\n",
       "text/plain": [
-       "<Figure size 648x648 with 9 Axes>"
+       "'Q'"
       ]
      },
-     "metadata": {
-      "needs_background": "light"
-     },
-     "output_type": "display_data"
-    }
-   ],
-   "source": [
-    "pol = MaxPooling2D(pool_size=(2, 2))\n",
-    "p = pol(c)\n",
-    "pp = pd.DataFrame(p.numpy().reshape(p.shape[1], p.shape[2]))\n",
-    "pplot(pp)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 32,
-   "id": "b1095a11",
-   "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/html": [
-       "<div>\n",
-       "<style scoped>\n",
-       "    .dataframe tbody tr th:only-of-type {\n",
-       "        vertical-align: middle;\n",
-       "    }\n",
-       "\n",
-       "    .dataframe tbody tr th {\n",
-       "        vertical-align: top;\n",
-       "    }\n",
-       "\n",
-       "    .dataframe thead th {\n",
-       "        text-align: right;\n",
-       "    }\n",
-       "</style>\n",
-       "<table border=\"1\" class=\"dataframe\">\n",
-       "  <thead>\n",
-       "    <tr style=\"text-align: right;\">\n",
-       "      <th></th>\n",
-       "      <th>0</th>\n",
-       "      <th>1</th>\n",
-       "      <th>2</th>\n",
-       "      <th>3</th>\n",
-       "      <th>4</th>\n",
-       "      <th>5</th>\n",
-       "      <th>6</th>\n",
-       "      <th>7</th>\n",
-       "      <th>8</th>\n",
-       "      <th>9</th>\n",
-       "      <th>10</th>\n",
-       "      <th>11</th>\n",
-       "      <th>12</th>\n",
-       "      <th>13</th>\n",
-       "    </tr>\n",
-       "  </thead>\n",
-       "  <tbody>\n",
-       "    <tr>\n",
-       "      <th>0</th>\n",
-       "      <td>0.001282</td>\n",
-       "      <td>0.412811</td>\n",
-       "      <td>-0.256378</td>\n",
-       "      <td>0.000854</td>\n",
-       "      <td>-0.409180</td>\n",
-       "      <td>0.262451</td>\n",
-       "      <td>0.006683</td>\n",
-       "      <td>0.005463</td>\n",
-       "      <td>0.008514</td>\n",
-       "      <td>0.017822</td>\n",
-       "      <td>0.096680</td>\n",
-       "      <td>-0.016479</td>\n",
-       "      <td>-0.561079</td>\n",
-       "      <td>0.0</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>1</th>\n",
-       "      <td>-0.016144</td>\n",
-       "      <td>0.409729</td>\n",
-       "      <td>-0.254547</td>\n",
-       "      <td>0.009766</td>\n",
-       "      <td>-0.405640</td>\n",
-       "      <td>0.272217</td>\n",
-       "      <td>0.005798</td>\n",
-       "      <td>0.004425</td>\n",
-       "      <td>0.008240</td>\n",
-       "      <td>0.017822</td>\n",
-       "      <td>0.096680</td>\n",
-       "      <td>-0.016602</td>\n",
-       "      <td>-0.561079</td>\n",
-       "      <td>1.0</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>2</th>\n",
-       "      <td>-0.006683</td>\n",
-       "      <td>0.408264</td>\n",
-       "      <td>-0.252441</td>\n",
-       "      <td>0.011963</td>\n",
-       "      <td>-0.401001</td>\n",
-       "      <td>0.274780</td>\n",
-       "      <td>0.004425</td>\n",
-       "      <td>0.002502</td>\n",
-       "      <td>0.006561</td>\n",
-       "      <td>0.018555</td>\n",
-       "      <td>0.096558</td>\n",
-       "      <td>-0.016357</td>\n",
-       "      <td>-0.561079</td>\n",
-       "      <td>2.0</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>3</th>\n",
-       "      <td>-0.011963</td>\n",
-       "      <td>0.407440</td>\n",
-       "      <td>-0.253662</td>\n",
-       "      <td>-0.002930</td>\n",
-       "      <td>-0.400635</td>\n",
-       "      <td>0.282837</td>\n",
-       "      <td>0.002716</td>\n",
-       "      <td>0.001312</td>\n",
-       "      <td>0.005707</td>\n",
-       "      <td>0.018433</td>\n",
-       "      <td>0.096680</td>\n",
-       "      <td>-0.015991</td>\n",
-       "      <td>-0.561079</td>\n",
-       "      <td>3.0</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>4</th>\n",
-       "      <td>-0.011292</td>\n",
-       "      <td>0.401672</td>\n",
-       "      <td>-0.246674</td>\n",
-       "      <td>-0.006226</td>\n",
-       "      <td>-0.399658</td>\n",
-       "      <td>0.289795</td>\n",
-       "      <td>0.000549</td>\n",
-       "      <td>0.001709</td>\n",
-       "      <td>0.005615</td>\n",
-       "      <td>0.018311</td>\n",
-       "      <td>0.096680</td>\n",
-       "      <td>-0.016357</td>\n",
-       "      <td>-0.561079</td>\n",
-       "      <td>4.0</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>...</th>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>70</th>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>71</th>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>72</th>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>73</th>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>74</th>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "    </tr>\n",
-       "  </tbody>\n",
-       "</table>\n",
-       "<p>75 rows × 14 columns</p>\n",
-       "</div>"
-      ],
-      "text/plain": [
-       "          0         1         2         3         4         5         6   \\\n",
-       "0   0.001282  0.412811 -0.256378  0.000854 -0.409180  0.262451  0.006683   \n",
-       "1  -0.016144  0.409729 -0.254547  0.009766 -0.405640  0.272217  0.005798   \n",
-       "2  -0.006683  0.408264 -0.252441  0.011963 -0.401001  0.274780  0.004425   \n",
-       "3  -0.011963  0.407440 -0.253662 -0.002930 -0.400635  0.282837  0.002716   \n",
-       "4  -0.011292  0.401672 -0.246674 -0.006226 -0.399658  0.289795  0.000549   \n",
-       "..       ...       ...       ...       ...       ...       ...       ...   \n",
-       "70  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000   \n",
-       "71  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000   \n",
-       "72  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000   \n",
-       "73  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000   \n",
-       "74  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000   \n",
-       "\n",
-       "          7         8         9         10        11        12   13  \n",
-       "0   0.005463  0.008514  0.017822  0.096680 -0.016479 -0.561079  0.0  \n",
-       "1   0.004425  0.008240  0.017822  0.096680 -0.016602 -0.561079  1.0  \n",
-       "2   0.002502  0.006561  0.018555  0.096558 -0.016357 -0.561079  2.0  \n",
-       "3   0.001312  0.005707  0.018433  0.096680 -0.015991 -0.561079  3.0  \n",
-       "4   0.001709  0.005615  0.018311  0.096680 -0.016357 -0.561079  4.0  \n",
-       "..       ...       ...       ...       ...       ...       ...  ...  \n",
-       "70  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.0  \n",
-       "71  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.0  \n",
-       "72  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.0  \n",
-       "73  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.0  \n",
-       "74  0.000000  0.000000  0.000000  0.000000  0.000000  0.000000  0.0  \n",
-       "\n",
-       "[75 rows x 14 columns]"
-      ]
-     },
-     "execution_count": 32,
+     "execution_count": 103,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "pd.DataFrame(d.reshape(X_train[0].shape[0], X_train[0].shape[1]).T)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 33,
-   "id": "3228ce57",
-   "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/html": [
-       "<div>\n",
-       "<style scoped>\n",
-       "    .dataframe tbody tr th:only-of-type {\n",
-       "        vertical-align: middle;\n",
-       "    }\n",
-       "\n",
-       "    .dataframe tbody tr th {\n",
-       "        vertical-align: top;\n",
-       "    }\n",
-       "\n",
-       "    .dataframe thead th {\n",
-       "        text-align: right;\n",
-       "    }\n",
-       "</style>\n",
-       "<table border=\"1\" class=\"dataframe\">\n",
-       "  <thead>\n",
-       "    <tr style=\"text-align: right;\">\n",
-       "      <th></th>\n",
-       "      <th>0</th>\n",
-       "      <th>1</th>\n",
-       "      <th>2</th>\n",
-       "      <th>3</th>\n",
-       "      <th>4</th>\n",
-       "      <th>5</th>\n",
-       "      <th>6</th>\n",
-       "      <th>7</th>\n",
-       "      <th>8</th>\n",
-       "      <th>9</th>\n",
-       "      <th>10</th>\n",
-       "      <th>11</th>\n",
-       "      <th>12</th>\n",
-       "      <th>13</th>\n",
-       "    </tr>\n",
-       "  </thead>\n",
-       "  <tbody>\n",
-       "    <tr>\n",
-       "      <th>0</th>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.164418</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.460755</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.267381</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.322476</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.116835</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>1</th>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.358112</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.394801</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.405149</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.045922</td>\n",
-       "      <td>0.236685</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.052844</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>2</th>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.356213</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.403690</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.408704</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.045495</td>\n",
-       "      <td>0.236544</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.562665</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>3</th>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.355599</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.399693</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.410803</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.045657</td>\n",
-       "      <td>0.236834</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>1.072485</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>4</th>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.360133</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.382545</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.400446</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.045980</td>\n",
-       "      <td>0.237183</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>1.582306</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>...</th>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "      <td>...</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>70</th>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>71</th>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>72</th>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>73</th>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>74</th>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.000000</td>\n",
-       "      <td>0.0</td>\n",
-       "      <td>0.000000</td>\n",
-       "    </tr>\n",
-       "  </tbody>\n",
-       "</table>\n",
-       "<p>75 rows × 14 columns</p>\n",
-       "</div>"
-      ],
-      "text/plain": [
-       "     0         1    2         3    4         5    6    7    8    9         10  \\\n",
-       "0   0.0  0.164418  0.0  0.460755  0.0  0.267381  0.0  0.0  0.0  0.0  0.000000   \n",
-       "1   0.0  0.358112  0.0  0.394801  0.0  0.405149  0.0  0.0  0.0  0.0  0.045922   \n",
-       "2   0.0  0.356213  0.0  0.403690  0.0  0.408704  0.0  0.0  0.0  0.0  0.045495   \n",
-       "3   0.0  0.355599  0.0  0.399693  0.0  0.410803  0.0  0.0  0.0  0.0  0.045657   \n",
-       "4   0.0  0.360133  0.0  0.382545  0.0  0.400446  0.0  0.0  0.0  0.0  0.045980   \n",
-       "..  ...       ...  ...       ...  ...       ...  ...  ...  ...  ...       ...   \n",
-       "70  0.0  0.000000  0.0  0.000000  0.0  0.000000  0.0  0.0  0.0  0.0  0.000000   \n",
-       "71  0.0  0.000000  0.0  0.000000  0.0  0.000000  0.0  0.0  0.0  0.0  0.000000   \n",
-       "72  0.0  0.000000  0.0  0.000000  0.0  0.000000  0.0  0.0  0.0  0.0  0.000000   \n",
-       "73  0.0  0.000000  0.0  0.000000  0.0  0.000000  0.0  0.0  0.0  0.0  0.000000   \n",
-       "74  0.0  0.000000  0.0  0.000000  0.0  0.000000  0.0  0.0  0.0  0.0  0.000000   \n",
-       "\n",
-       "          11   12        13  \n",
-       "0   0.322476  0.0  0.116835  \n",
-       "1   0.236685  0.0  0.052844  \n",
-       "2   0.236544  0.0  0.562665  \n",
-       "3   0.236834  0.0  1.072485  \n",
-       "4   0.237183  0.0  1.582306  \n",
-       "..       ...  ...       ...  \n",
-       "70  0.000000  0.0  0.000000  \n",
-       "71  0.000000  0.0  0.000000  \n",
-       "72  0.000000  0.0  0.000000  \n",
-       "73  0.000000  0.0  0.000000  \n",
-       "74  0.000000  0.0  0.000000  \n",
-       "\n",
-       "[75 rows x 14 columns]"
-      ]
-     },
-     "execution_count": 33,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "pd.DataFrame(c.numpy().reshape(c.shape[1], c.shape[2]).T)"
+    "lb.inverse_transform(y_test)[0]"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "cf15d166",
+   "id": "be9f7690",
    "metadata": {},
    "outputs": [],
    "source": []
diff --git a/1-first-project/goat.weights.data-00000-of-00001 b/1-first-project/goat.weights.data-00000-of-00001
index 4be11a0..e67af86 100644
Binary files a/1-first-project/goat.weights.data-00000-of-00001 and b/1-first-project/goat.weights.data-00000-of-00001 differ
diff --git a/1-first-project/goat.weights.index b/1-first-project/goat.weights.index
index ed69ba5..c3e091e 100644
Binary files a/1-first-project/goat.weights.index and b/1-first-project/goat.weights.index differ
diff --git a/2-second-project/tdt/DataViz.ipynb b/2-second-project/tdt/DataViz.ipynb
index a4574e6..06049bf 100644
--- a/2-second-project/tdt/DataViz.ipynb
+++ b/2-second-project/tdt/DataViz.ipynb
@@ -2,7 +2,7 @@
  "cells": [
   {
    "cell_type": "markdown",
-   "id": "ae397d48",
+   "id": "744fe1a3",
    "metadata": {},
    "source": [
     "# Constants"
@@ -11,37 +11,69 @@
   {
    "cell_type": "code",
    "execution_count": 1,
-   "id": "3827a09b",
+   "id": "ddf4a4a9",
    "metadata": {},
    "outputs": [],
    "source": [
     "import os\n",
     "\n",
     "os.environ['TF_FORCE_GPU_ALLOW_GROWTH'] = 'true'  # this is required\n",
-    "os.environ['CUDA_VISIBLE_DEVICES'] = '1'         # set to '0' for GPU0, '1' for GPU1 or '2' for GPU2. Check \"gpustat\" in a terminal."
+    "os.environ['CUDA_VISIBLE_DEVICES'] = '2'         # set to '0' for GPU0, '1' for GPU1 or '2' for GPU2. Check \"gpustat\" in a terminal."
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 2,
-   "id": "654f2682",
+   "id": "46b9b6e1",
    "metadata": {},
    "outputs": [],
    "source": [
     "glob_path = '/opt/iui-datarelease3-sose2021/*.csv'\n",
     "\n",
-    "pickle_file = '../data.pickle'\n",
-    "\n",
-    "cenario = 'SYY'\n",
-    "\n",
-    "win_sz = 50\n",
-    "stride_sz = 25 "
+    "pickle_file = '../data.pickle'"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "ce44c37d",
+   "metadata": {},
+   "source": [
+    "# Config"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 3,
-   "id": "6cc88c90",
+   "id": "cf80b408",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Possibilities: 'SYY', 'SYN', 'SNY', 'SNN', 'JYY', 'JYN', 'JNY', 'JNN'\n",
+    "cenario = 'SYY'\n",
+    "\n",
+    "win_sz = 10\n",
+    "stride_sz = 10\n",
+    "\n",
+    "# divisor for neuron count step downs (hard to describe), e.g. dense_step = 3: layer1=900, layer2 = 300, layer3 = 100, layer4 = 33...\n",
+    "dense_steps = 3\n",
+    "# amount of dense/dropout layers\n",
+    "layer_count = 7\n",
+    "# how much to drop\n",
+    "drop_count = 0.3"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "878942c5",
+   "metadata": {},
+   "source": [
+    "# Helper Functions"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "2ef17323",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -59,7 +91,7 @@
   },
   {
    "cell_type": "markdown",
-   "id": "3c47f127",
+   "id": "10ec6524",
    "metadata": {},
    "source": [
     "# Loading Data"
@@ -67,8 +99,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 4,
-   "id": "9dc8d47e",
+   "execution_count": 5,
+   "id": "b2fb4a5a",
    "metadata": {
     "tags": []
    },
@@ -113,8 +145,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
-   "id": "1294685f",
+   "execution_count": 6,
+   "id": "5272d133",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -128,8 +160,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 6,
-   "id": "5e418dc4",
+   "execution_count": 7,
+   "id": "a1c89dfa",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -143,8 +175,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 7,
-   "id": "7938c466",
+   "execution_count": 8,
+   "id": "55b3f20a",
    "metadata": {},
    "outputs": [
     {
@@ -154,7 +186,7 @@
       "Loading data...\n",
       "../data.pickle found...\n",
       "768\n",
-      "CPU times: user 615 ms, sys: 2.24 s, total: 2.85 s\n",
+      "CPU times: user 612 ms, sys: 2.23 s, total: 2.85 s\n",
       "Wall time: 2.85 s\n"
      ]
     }
@@ -179,13 +211,24 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 8,
-   "id": "e3f38b64",
+   "execution_count": 9,
+   "id": "95aa7e4f",
    "metadata": {
     "tags": []
    },
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "CPU times: user 104 µs, sys: 319 µs, total: 423 µs\n",
+      "Wall time: 427 µs\n"
+     ]
+    }
+   ],
    "source": [
+    "%%time\n",
+    "\n",
     "# Categorized Data\n",
     "cdata = dict() \n",
     "# Sorting, HeightNorm, ArmNorm\n",
@@ -221,17 +264,12 @@
     "            if d['armnorm']:\n",
     "                cdata['JNY'].append(d)\n",
     "            else:\n",
-    "                cdata['JNN'].append(d)\n",
-    "\n",
-    "# for k,v in cdata.items():\n",
-    "#     print(k,': ',len(v))\n",
-    "# test_entry = pickle.loads(pickle.dumps(cdata['SYY'][17]))\n",
-    "# test_entry['data']"
+    "                cdata['JNN'].append(d)"
    ]
   },
   {
    "cell_type": "markdown",
-   "id": "83953c92",
+   "id": "c60aa94c",
    "metadata": {},
    "source": [
     "# Preprocessing"
@@ -239,8 +277,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 9,
-   "id": "583e8c34",
+   "execution_count": 10,
+   "id": "76c12c0d",
    "metadata": {
     "tags": []
    },
@@ -254,8 +292,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 10,
-   "id": "b8a05286",
+   "execution_count": 11,
+   "id": "465e703b",
    "metadata": {
     "tags": []
    },
@@ -270,8 +308,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 11,
-   "id": "fbe90e8d",
+   "execution_count": 12,
+   "id": "3eff3694",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -291,8 +329,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 12,
-   "id": "26059dd4",
+   "execution_count": 13,
+   "id": "79f2f0e3",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -316,8 +354,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 13,
-   "id": "2f2181f0",
+   "execution_count": 14,
+   "id": "940b6076",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -328,8 +366,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 14,
-   "id": "276ecf82",
+   "execution_count": 15,
+   "id": "5a6e6aaa",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -349,19 +387,36 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 15,
-   "id": "dab70ad9",
+   "execution_count": 16,
+   "id": "b546250e",
    "metadata": {},
    "outputs": [
     {
      "name": "stderr",
      "output_type": "stream",
      "text": [
-      "100%|██████████| 96/96 [00:15<00:00,  6.14it/s]\n"
+      "100%|██████████| 96/96 [00:15<00:00,  6.03it/s]"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "CPU times: user 14.2 s, sys: 1.9 s, total: 16.1 s\n",
+      "Wall time: 15.9 s\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n"
      ]
     }
    ],
    "source": [
+    "%%time\n",
+    "\n",
     "classes = 16 # dynamic\n",
     "\n",
     "def preproc(data):\n",
@@ -398,7 +453,7 @@
   },
   {
    "cell_type": "markdown",
-   "id": "ddba89b9",
+   "id": "f133e9d4",
    "metadata": {},
    "source": [
     "# Building Model"
@@ -406,30 +461,33 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 16,
-   "id": "61c34fed",
+   "execution_count": 17,
+   "id": "dd333744",
    "metadata": {},
    "outputs": [],
    "source": [
     "import tensorflow as tf\n",
     "from tensorflow.keras.models import Sequential\n",
-    "from tensorflow.keras.layers import Dense, Flatten, BatchNormalization, Dropout, Conv2D, MaxPooling2D\n",
-    "\n",
-    "def build_model(train):\n",
-    "    s = train[0].shape\n",
+    "from tensorflow.keras.layers import Dense, Flatten, BatchNormalization, Dropout, LSTM\n",
+    "import tensorflow.keras as keras\n",
     "\n",
+    "def build_model(shape, classes):\n",
+    "    \n",
     "    model = Sequential()\n",
-    "    ncount = s[0]*s[1]\n",
+    "    ncount = shape[0]*shape[1]\n",
     "    \n",
     "    model.add(Flatten(input_shape=s))\n",
     "    \n",
     "    model.add(BatchNormalization())\n",
     "    \n",
-    "    model.add(Dropout(0.1))\n",
+    "    model.add(Dropout(drop_count))\n",
     "    \n",
-    "    for i in range(1,6):\n",
-    "        model.add(Dense(int(ncount/pow(3,i)), activation='relu'))\n",
-    "        model.add(Dropout(0.1))\n",
+    "    for i in range(1,layer_count):\n",
+    "        neurons = int(ncount/pow(dense_steps,i))\n",
+    "        if neurons <= classes*dense_steps:\n",
+    "            break\n",
+    "        model.add(Dense(neurons, activation='relu'))\n",
+    "        model.add(Dropout(drop_count))\n",
     "    \n",
     "    model.add(Dense(classes, activation='softmax'))\n",
     "\n",
@@ -439,54 +497,118 @@
     "        metrics=[\"acc\"],\n",
     "    )\n",
     "\n",
+    "    return model\n",
+    "\n",
+    "def build_mlp(input_shape, nb_classes):\n",
+    "    input_layer = keras.layers.Input(input_shape)\n",
+    "\n",
+    "    # flatten/reshape because when multivariate all should be on the same axis \n",
+    "    input_layer_flattened = keras.layers.Flatten()(input_layer)\n",
+    "    \n",
+    "    layer_1 = keras.layers.Dropout(0.1)(input_layer_flattened)\n",
+    "    layer_1 = keras.layers.Dense(500, activation='relu')(layer_1)\n",
+    "\n",
+    "    layer_2 = keras.layers.Dropout(0.2)(layer_1)\n",
+    "    layer_2 = keras.layers.Dense(500, activation='relu')(layer_2)\n",
+    "\n",
+    "    layer_3 = keras.layers.Dropout(0.2)(layer_2)\n",
+    "    layer_3 = keras.layers.Dense(500, activation='relu')(layer_3)\n",
+    "\n",
+    "    output_layer = keras.layers.Dropout(0.3)(layer_3)\n",
+    "    output_layer = keras.layers.Dense(nb_classes, activation='softmax')(output_layer)\n",
+    "\n",
+    "    model = keras.models.Model(inputs=input_layer, outputs=output_layer)\n",
+    "\n",
+    "    model.compile(loss='categorical_crossentropy', optimizer=keras.optimizers.Adadelta(),\n",
+    "        metrics=['accuracy'])\n",
+    "\n",
     "    return model"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 17,
-   "id": "47058299",
+   "execution_count": 18,
+   "id": "d6f48e5c",
    "metadata": {},
    "outputs": [],
    "source": [
     "checkpoint_file = './goat.weights'\n",
     "\n",
-    "def train_model(X_train, y_train):\n",
-    "    model = build_model(X_train)\n",
+    "def train_model(X_train, y_train, X_test, y_test):\n",
+    "    model = build_model(X_train[0].shape, 16)\n",
     "    \n",
     "    model.summary()\n",
     "\n",
+    "    model_checkpoint_callback = tf.keras.callbacks.ModelCheckpoint(\n",
+    "        filepath = checkpoint_file,\n",
+    "        save_weights_only=True,\n",
+    "        monitor='val_acc',\n",
+    "        mode='max',\n",
+    "        save_best_only=True\n",
+    "    )\n",
+    "        \n",
     "    history = model.fit(X_train, \n",
-    "              y_train,\n",
-    "              epochs=30,\n",
-    "              batch_size=128,\n",
-    "              shuffle=True,\n",
-    "              verbose=0,\n",
+    "                        y_train,\n",
+    "                        epochs=30,\n",
+    "                        batch_size=128,\n",
+    "                        shuffle=True,\n",
+    "                        verbose=2,\n",
+    "                        validation_data=(X_test, y_test),\n",
+    "                        callbacks=[model_checkpoint_callback]\n",
+    "                        \n",
     "             )\n",
+    "    return model, history\n",
+    "\n",
+    "def train_mlp(X_train, y_train, X_test, y_test):\n",
+    "    model = build_mlp(X_train[0].shape, 16)\n",
+    "    model.summary()\n",
+    "    \n",
+    "    reduce_lr = keras.callbacks.ReduceLROnPlateau(monitor='loss', factor=0.5, patience=200, min_lr=0.1)\n",
+    "\n",
+    "    model_checkpoint = keras.callbacks.ModelCheckpoint(filepath=checkpoint_file, monitor='loss', \n",
+    "        save_best_only=True)\n",
+    "\n",
+    "    callbacks = [reduce_lr,model_checkpoint]\n",
+    "    history = model.fit(X_train, \n",
+    "                        y_train,\n",
+    "                        epochs=5000,\n",
+    "                        batch_size=16,\n",
+    "                        shuffle=True,\n",
+    "                        verbose=2,\n",
+    "                        validation_data=(X_test, y_test),\n",
+    "                        callbacks=callbacks\n",
+    "    )\n",
+    "    \n",
     "    return model, history"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 18,
-   "id": "6c99e0bc",
+   "execution_count": 19,
+   "id": "b30a9bae",
    "metadata": {},
    "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "CPU times: user 187 µs, sys: 140 µs, total: 327 µs\n",
+      "Wall time: 344 µs\n"
+     ]
+    },
     {
      "data": {
       "text/plain": [
        "(48, 48)"
       ]
      },
-     "execution_count": 18,
+     "execution_count": 19,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "from sklearn.model_selection import train_test_split\n",
-    "from sklearn.preprocessing import LabelEncoder, LabelBinarizer\n",
-    "\n",
+    "%%time\n",
     "train = np.array([x['data'] for x in pdata if x['session'] == 1])\n",
     "test = np.array([x['data'] for x in pdata if x['session'] == 2])\n",
     "\n",
@@ -495,24 +617,28 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 19,
-   "id": "727b89e0",
+   "execution_count": 20,
+   "id": "d9354704",
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "CPU times: user 8.86 s, sys: 3.63 s, total: 12.5 s\n",
-      "Wall time: 4.7 s\n"
+      "CPU times: user 15.2 s, sys: 5.2 s, total: 20.4 s\n",
+      "Wall time: 6.07 s\n"
      ]
     }
    ],
    "source": [
     "%%time\n",
+    "\n",
     "X_train = list()\n",
     "y_train = list()\n",
     "\n",
+    "X_test = list()\n",
+    "y_test = list()\n",
+    "\n",
     "train = list()\n",
     "test = list()\n",
     "\n",
@@ -535,30 +661,61 @@
     "                'data': list()\n",
     "            })\n",
     "        for y in x['data'].unbatch().as_numpy_iterator():\n",
+    "            X_test.append(y[0])\n",
+    "            y_test.append(y[1])\n",
+    "            \n",
     "            test[-1]['data'].append(y[0])\n",
     "\n",
     "X_train = np.array(X_train)\n",
-    "y_train = np.array(y_train)"
+    "y_train = np.array(y_train)\n",
+    "X_test = np.array(X_test)\n",
+    "y_test = np.array(y_test)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 20,
-   "id": "ba64dca4",
+   "execution_count": 21,
+   "id": "7b753a95",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "((14700, 10, 338), (14700,), (11115, 10, 338), (11115,))"
+      ]
+     },
+     "execution_count": 21,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "X_train.shape, y_train.shape, X_test.shape, y_test.shape"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "id": "f8daea45",
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "(5832, 50, 338)\n",
-      "(5832, 16)\n"
+      "CPU times: user 476 ms, sys: 93.3 ms, total: 570 ms\n",
+      "Wall time: 570 ms\n"
      ]
     }
    ],
    "source": [
+    "%%time\n",
+    "\n",
+    "from sklearn.preprocessing import LabelBinarizer\n",
+    "\n",
     "lb = LabelBinarizer()\n",
     "yy_train = lb.fit_transform(y_train)\n",
+    "yy_test = lb.fit_transform(y_test)\n",
     "\n",
     "for e in test:\n",
     "    e['label'] = lb.transform([e['label']])\n",
@@ -566,172 +723,10468 @@
     "    \n",
     "for e in train:\n",
     "    e['label'] = lb.transform([e['label']])\n",
-    "    e['data'] = np.array(e['data'])\n",
-    "\n",
-    "print(X_train.shape)\n",
-    "print(yy_train.shape)\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 21,
-   "id": "399176de",
-   "metadata": {
-    "tags": []
-   },
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      " 4:  53 (53, 50, 338)\n",
-      "14:  35 (35, 50, 338)\n",
-      "12:  65 (65, 50, 338)\n",
-      " 8: 149 (149, 50, 338)\n",
-      " 1:  53 (53, 50, 338)\n",
-      " 3: 107 (107, 50, 338)\n",
-      "11:  53 (53, 50, 338)\n",
-      " 3: 125 (125, 50, 338)\n",
-      " 1:  41 (41, 50, 338)\n",
-      "13:  71 (71, 50, 338)\n",
-      "15:  59 (59, 50, 338)\n",
-      " 3:  77 (77, 50, 338)\n",
-      "10: 119 (119, 50, 338)\n",
-      " 6:  47 (47, 50, 338)\n",
-      "14:  41 (41, 50, 338)\n",
-      " 5: 167 (167, 50, 338)\n",
-      " 8:  89 (89, 50, 338)\n",
-      "14:  41 (41, 50, 338)\n",
-      " 9:  71 (71, 50, 338)\n",
-      "10:  77 (77, 50, 338)\n",
-      " 8:  77 (77, 50, 338)\n",
-      "16:  77 (77, 50, 338)\n",
-      "16:  77 (77, 50, 338)\n",
-      " 2:  59 (59, 50, 338)\n",
-      " 9:  77 (77, 50, 338)\n",
-      "15:  77 (77, 50, 338)\n",
-      " 5: 101 (101, 50, 338)\n",
-      "16:  71 (71, 50, 338)\n",
-      "15:  71 (71, 50, 338)\n",
-      "12:  95 (95, 50, 338)\n",
-      " 6:  71 (71, 50, 338)\n",
-      " 2:  53 (53, 50, 338)\n",
-      "12: 845 (845, 50, 338)\n",
-      " 7:  65 (65, 50, 338)\n",
-      " 2:  65 (65, 50, 338)\n",
-      "13:  95 (95, 50, 338)\n",
-      " 5: 125 (125, 50, 338)\n",
-      "11:  65 (65, 50, 338)\n",
-      " 7:  59 (59, 50, 338)\n",
-      "10:  77 (77, 50, 338)\n",
-      " 6:  59 (59, 50, 338)\n",
-      " 7:  53 (53, 50, 338)\n",
-      " 1: 101 (101, 50, 338)\n",
-      "13:  71 (71, 50, 338)\n",
-      "11:  59 (59, 50, 338)\n",
-      " 4:  77 (77, 50, 338)\n",
-      " 9:  29 (29, 50, 338)\n",
-      " 4: 107 (107, 50, 338)\n"
-     ]
-    }
-   ],
-   "source": [
-    "for e in test:\n",
-    "    print(f\"{lb.inverse_transform(e['label'])[0]:2d}: {len(e['data']):3d} {e['data'].shape}\")"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 22,
-   "id": "75af2444",
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "Model: \"sequential\"\n",
-      "_________________________________________________________________\n",
-      "Layer (type)                 Output Shape              Param #   \n",
-      "=================================================================\n",
-      "flatten (Flatten)            (None, 16900)             0         \n",
-      "_________________________________________________________________\n",
-      "batch_normalization (BatchNo (None, 16900)             67600     \n",
-      "_________________________________________________________________\n",
-      "dropout (Dropout)            (None, 16900)             0         \n",
-      "_________________________________________________________________\n",
-      "dense (Dense)                (None, 5633)              95203333  \n",
-      "_________________________________________________________________\n",
-      "dropout_1 (Dropout)          (None, 5633)              0         \n",
-      "_________________________________________________________________\n",
-      "dense_1 (Dense)              (None, 1877)              10575018  \n",
-      "_________________________________________________________________\n",
-      "dropout_2 (Dropout)          (None, 1877)              0         \n",
-      "_________________________________________________________________\n",
-      "dense_2 (Dense)              (None, 625)               1173750   \n",
-      "_________________________________________________________________\n",
-      "dropout_3 (Dropout)          (None, 625)               0         \n",
-      "_________________________________________________________________\n",
-      "dense_3 (Dense)              (None, 208)               130208    \n",
-      "_________________________________________________________________\n",
-      "dropout_4 (Dropout)          (None, 208)               0         \n",
-      "_________________________________________________________________\n",
-      "dense_4 (Dense)              (None, 69)                14421     \n",
-      "_________________________________________________________________\n",
-      "dropout_5 (Dropout)          (None, 69)                0         \n",
-      "_________________________________________________________________\n",
-      "dense_5 (Dense)              (None, 16)                1120      \n",
-      "=================================================================\n",
-      "Total params: 107,165,450\n",
-      "Trainable params: 107,131,650\n",
-      "Non-trainable params: 33,800\n",
-      "_________________________________________________________________\n",
-      "CPU times: user 32.2 s, sys: 9.61 s, total: 41.8 s\n",
-      "Wall time: 18 s\n"
-     ]
-    }
-   ],
-   "source": [
-    "%%time\n",
-    "model, history = train_model(np.array(X_train), np.array(yy_train))"
+    "    e['data'] = np.array(e['data'])\n"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 23,
-   "id": "1a63ecda",
+   "id": "e6857113",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "(14700, 10, 338)\n",
+      "(14700, 16)\n",
+      "(11115, 10, 338)\n",
+      "(11115, 16)\n"
+     ]
+    }
+   ],
+   "source": [
+    "print(X_train.shape)\n",
+    "print(yy_train.shape)\n",
+    "print(X_test.shape)\n",
+    "print(yy_test.shape)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 24,
+   "id": "e5d4ab4e",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Model: \"model\"\n",
+      "_________________________________________________________________\n",
+      "Layer (type)                 Output Shape              Param #   \n",
+      "=================================================================\n",
+      "input_1 (InputLayer)         [(None, 10, 338)]         0         \n",
+      "_________________________________________________________________\n",
+      "flatten (Flatten)            (None, 3380)              0         \n",
+      "_________________________________________________________________\n",
+      "dropout (Dropout)            (None, 3380)              0         \n",
+      "_________________________________________________________________\n",
+      "dense (Dense)                (None, 500)               1690500   \n",
+      "_________________________________________________________________\n",
+      "dropout_1 (Dropout)          (None, 500)               0         \n",
+      "_________________________________________________________________\n",
+      "dense_1 (Dense)              (None, 500)               250500    \n",
+      "_________________________________________________________________\n",
+      "dropout_2 (Dropout)          (None, 500)               0         \n",
+      "_________________________________________________________________\n",
+      "dense_2 (Dense)              (None, 500)               250500    \n",
+      "_________________________________________________________________\n",
+      "dropout_3 (Dropout)          (None, 500)               0         \n",
+      "_________________________________________________________________\n",
+      "dense_3 (Dense)              (None, 16)                8016      \n",
+      "=================================================================\n",
+      "Total params: 2,199,516\n",
+      "Trainable params: 2,199,516\n",
+      "Non-trainable params: 0\n",
+      "_________________________________________________________________\n",
+      "Epoch 1/5000\n",
+      "919/919 - 4s - loss: 425.4944 - accuracy: 0.1014 - val_loss: 96.3826 - val_accuracy: 0.0489\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2/5000\n",
+      "919/919 - 3s - loss: 273.6042 - accuracy: 0.1466 - val_loss: 80.7553 - val_accuracy: 0.0561\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3/5000\n",
+      "919/919 - 3s - loss: 172.5016 - accuracy: 0.1813 - val_loss: 64.9933 - val_accuracy: 0.0646\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4/5000\n",
+      "919/919 - 3s - loss: 108.2793 - accuracy: 0.2088 - val_loss: 53.3990 - val_accuracy: 0.0722\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 5/5000\n",
+      "919/919 - 3s - loss: 86.3331 - accuracy: 0.2260 - val_loss: 44.7914 - val_accuracy: 0.0721\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 6/5000\n",
+      "919/919 - 3s - loss: 67.5223 - accuracy: 0.2382 - val_loss: 38.8661 - val_accuracy: 0.0732\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 7/5000\n",
+      "919/919 - 3s - loss: 53.6253 - accuracy: 0.2472 - val_loss: 34.1444 - val_accuracy: 0.0681\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 8/5000\n",
+      "919/919 - 3s - loss: 47.5227 - accuracy: 0.2580 - val_loss: 29.6918 - val_accuracy: 0.0650\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 9/5000\n",
+      "919/919 - 3s - loss: 40.8288 - accuracy: 0.2620 - val_loss: 26.7311 - val_accuracy: 0.0686\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 10/5000\n",
+      "919/919 - 3s - loss: 35.6907 - accuracy: 0.2695 - val_loss: 23.6847 - val_accuracy: 0.0631\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 11/5000\n",
+      "919/919 - 3s - loss: 33.6311 - accuracy: 0.2679 - val_loss: 20.8185 - val_accuracy: 0.0623\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 12/5000\n",
+      "919/919 - 3s - loss: 28.5294 - accuracy: 0.2709 - val_loss: 18.0884 - val_accuracy: 0.0609\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 13/5000\n",
+      "919/919 - 3s - loss: 24.5650 - accuracy: 0.2696 - val_loss: 15.6305 - val_accuracy: 0.0608\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 14/5000\n",
+      "919/919 - 3s - loss: 23.0020 - accuracy: 0.2713 - val_loss: 13.3984 - val_accuracy: 0.0614\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 15/5000\n",
+      "919/919 - 3s - loss: 21.2715 - accuracy: 0.2665 - val_loss: 11.7454 - val_accuracy: 0.0605\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 16/5000\n",
+      "919/919 - 3s - loss: 18.3594 - accuracy: 0.2681 - val_loss: 10.0513 - val_accuracy: 0.0633\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 17/5000\n",
+      "919/919 - 3s - loss: 16.2328 - accuracy: 0.2616 - val_loss: 8.8303 - val_accuracy: 0.0650\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 18/5000\n",
+      "919/919 - 3s - loss: 13.8618 - accuracy: 0.2710 - val_loss: 7.8014 - val_accuracy: 0.0659\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 19/5000\n",
+      "919/919 - 3s - loss: 13.7863 - accuracy: 0.2599 - val_loss: 7.1464 - val_accuracy: 0.0662\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 20/5000\n",
+      "919/919 - 3s - loss: 11.1480 - accuracy: 0.2643 - val_loss: 6.5873 - val_accuracy: 0.0766\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 21/5000\n",
+      "919/919 - 3s - loss: 10.4391 - accuracy: 0.2584 - val_loss: 6.0758 - val_accuracy: 0.0802\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 22/5000\n",
+      "919/919 - 3s - loss: 9.7688 - accuracy: 0.2541 - val_loss: 5.7184 - val_accuracy: 0.0877\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 23/5000\n",
+      "919/919 - 3s - loss: 9.3946 - accuracy: 0.2515 - val_loss: 5.3169 - val_accuracy: 0.0910\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 24/5000\n",
+      "919/919 - 3s - loss: 8.3558 - accuracy: 0.2537 - val_loss: 5.0412 - val_accuracy: 0.0963\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 25/5000\n",
+      "919/919 - 3s - loss: 7.9109 - accuracy: 0.2497 - val_loss: 4.8023 - val_accuracy: 0.0988\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 26/5000\n",
+      "919/919 - 3s - loss: 7.7169 - accuracy: 0.2546 - val_loss: 4.6018 - val_accuracy: 0.1003\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 27/5000\n",
+      "919/919 - 3s - loss: 7.0899 - accuracy: 0.2529 - val_loss: 4.4534 - val_accuracy: 0.0595\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 28/5000\n",
+      "919/919 - 3s - loss: 6.3440 - accuracy: 0.2656 - val_loss: 4.2341 - val_accuracy: 0.0611\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 29/5000\n",
+      "919/919 - 3s - loss: 6.6692 - accuracy: 0.2674 - val_loss: 4.1783 - val_accuracy: 0.0618\n",
+      "Epoch 30/5000\n",
+      "919/919 - 3s - loss: 5.8945 - accuracy: 0.2713 - val_loss: 4.0080 - val_accuracy: 0.0618\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 31/5000\n",
+      "919/919 - 3s - loss: 5.8075 - accuracy: 0.2689 - val_loss: 3.9229 - val_accuracy: 0.0623\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 32/5000\n",
+      "919/919 - 3s - loss: 5.3554 - accuracy: 0.2677 - val_loss: 3.8339 - val_accuracy: 0.0615\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 33/5000\n",
+      "919/919 - 3s - loss: 5.2647 - accuracy: 0.2793 - val_loss: 3.7327 - val_accuracy: 0.0614\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 34/5000\n",
+      "919/919 - 3s - loss: 5.3449 - accuracy: 0.2820 - val_loss: 3.6892 - val_accuracy: 0.0622\n",
+      "Epoch 35/5000\n",
+      "919/919 - 3s - loss: 5.4178 - accuracy: 0.2820 - val_loss: 3.6112 - val_accuracy: 0.0635\n",
+      "Epoch 36/5000\n",
+      "919/919 - 3s - loss: 4.8692 - accuracy: 0.2882 - val_loss: 3.5103 - val_accuracy: 0.0636\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 37/5000\n",
+      "919/919 - 3s - loss: 5.2422 - accuracy: 0.2873 - val_loss: 3.5021 - val_accuracy: 0.0638\n",
+      "Epoch 38/5000\n",
+      "919/919 - 3s - loss: 4.8051 - accuracy: 0.2941 - val_loss: 3.4460 - val_accuracy: 0.0644\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 39/5000\n",
+      "919/919 - 3s - loss: 4.5701 - accuracy: 0.2893 - val_loss: 3.3799 - val_accuracy: 0.0640\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 40/5000\n",
+      "919/919 - 3s - loss: 4.4864 - accuracy: 0.2947 - val_loss: 3.3437 - val_accuracy: 0.0641\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 41/5000\n",
+      "919/919 - 3s - loss: 4.2492 - accuracy: 0.2937 - val_loss: 3.2705 - val_accuracy: 0.0641\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 42/5000\n",
+      "919/919 - 3s - loss: 4.4921 - accuracy: 0.2962 - val_loss: 3.2148 - val_accuracy: 0.0641\n",
+      "Epoch 43/5000\n",
+      "919/919 - 3s - loss: 4.6991 - accuracy: 0.2941 - val_loss: 3.1950 - val_accuracy: 0.0646\n",
+      "Epoch 44/5000\n",
+      "919/919 - 3s - loss: 4.0940 - accuracy: 0.2992 - val_loss: 3.1515 - val_accuracy: 0.0647\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 45/5000\n",
+      "919/919 - 3s - loss: 3.9955 - accuracy: 0.3028 - val_loss: 3.1349 - val_accuracy: 0.0641\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 46/5000\n",
+      "919/919 - 3s - loss: 4.0743 - accuracy: 0.3034 - val_loss: 3.1004 - val_accuracy: 0.0641\n",
+      "Epoch 47/5000\n",
+      "919/919 - 3s - loss: 3.7824 - accuracy: 0.3038 - val_loss: 3.0873 - val_accuracy: 0.0641\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 48/5000\n",
+      "919/919 - 3s - loss: 3.8392 - accuracy: 0.3036 - val_loss: 3.0639 - val_accuracy: 0.0641\n",
+      "Epoch 49/5000\n",
+      "919/919 - 3s - loss: 3.7053 - accuracy: 0.3069 - val_loss: 3.0727 - val_accuracy: 0.0643\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 50/5000\n",
+      "919/919 - 3s - loss: 3.8474 - accuracy: 0.3082 - val_loss: 3.0553 - val_accuracy: 0.0647\n",
+      "Epoch 51/5000\n",
+      "919/919 - 3s - loss: 3.9240 - accuracy: 0.3060 - val_loss: 3.0496 - val_accuracy: 0.0650\n",
+      "Epoch 52/5000\n",
+      "919/919 - 3s - loss: 3.5656 - accuracy: 0.3134 - val_loss: 3.0237 - val_accuracy: 0.0648\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 53/5000\n",
+      "919/919 - 3s - loss: 3.5436 - accuracy: 0.3073 - val_loss: 3.0077 - val_accuracy: 0.0649\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 54/5000\n",
+      "919/919 - 3s - loss: 3.5733 - accuracy: 0.3092 - val_loss: 2.9904 - val_accuracy: 0.0646\n",
+      "Epoch 55/5000\n",
+      "919/919 - 3s - loss: 3.4300 - accuracy: 0.3095 - val_loss: 2.9733 - val_accuracy: 0.0644\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 56/5000\n",
+      "919/919 - 3s - loss: 3.4395 - accuracy: 0.3093 - val_loss: 2.9626 - val_accuracy: 0.0646\n",
+      "Epoch 57/5000\n",
+      "919/919 - 3s - loss: 3.5122 - accuracy: 0.3097 - val_loss: 2.9520 - val_accuracy: 0.0645\n",
+      "Epoch 58/5000\n",
+      "919/919 - 3s - loss: 3.7682 - accuracy: 0.3142 - val_loss: 2.9397 - val_accuracy: 0.0648\n",
+      "Epoch 59/5000\n",
+      "919/919 - 3s - loss: 3.3636 - accuracy: 0.3142 - val_loss: 2.9333 - val_accuracy: 0.0647\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 60/5000\n",
+      "919/919 - 3s - loss: 3.4117 - accuracy: 0.3131 - val_loss: 2.9261 - val_accuracy: 0.0648\n",
+      "Epoch 61/5000\n",
+      "919/919 - 3s - loss: 3.4966 - accuracy: 0.3129 - val_loss: 2.9235 - val_accuracy: 0.0646\n",
+      "Epoch 62/5000\n",
+      "919/919 - 3s - loss: 3.3473 - accuracy: 0.3155 - val_loss: 2.9150 - val_accuracy: 0.0644\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 63/5000\n",
+      "919/919 - 3s - loss: 3.3024 - accuracy: 0.3147 - val_loss: 2.9048 - val_accuracy: 0.0645\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 64/5000\n",
+      "919/919 - 3s - loss: 3.2226 - accuracy: 0.3188 - val_loss: 2.8932 - val_accuracy: 0.0641\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 65/5000\n",
+      "919/919 - 3s - loss: 3.1722 - accuracy: 0.3167 - val_loss: 2.8876 - val_accuracy: 0.0641\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 66/5000\n",
+      "919/919 - 3s - loss: 3.0936 - accuracy: 0.3156 - val_loss: 2.8806 - val_accuracy: 0.0639\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 67/5000\n",
+      "919/919 - 3s - loss: 3.1805 - accuracy: 0.3224 - val_loss: 2.8727 - val_accuracy: 0.0638\n",
+      "Epoch 68/5000\n",
+      "919/919 - 3s - loss: 3.0835 - accuracy: 0.3178 - val_loss: 2.8663 - val_accuracy: 0.0638\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 69/5000\n",
+      "919/919 - 3s - loss: 3.0565 - accuracy: 0.3241 - val_loss: 2.8623 - val_accuracy: 0.0641\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 70/5000\n",
+      "919/919 - 3s - loss: 3.0446 - accuracy: 0.3209 - val_loss: 2.8538 - val_accuracy: 0.0639\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 71/5000\n",
+      "919/919 - 3s - loss: 3.0794 - accuracy: 0.3234 - val_loss: 2.8532 - val_accuracy: 0.0637\n",
+      "Epoch 72/5000\n",
+      "919/919 - 3s - loss: 3.0815 - accuracy: 0.3193 - val_loss: 2.8445 - val_accuracy: 0.0645\n",
+      "Epoch 73/5000\n",
+      "919/919 - 3s - loss: 2.9671 - accuracy: 0.3209 - val_loss: 2.8397 - val_accuracy: 0.0648\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 74/5000\n",
+      "919/919 - 3s - loss: 2.9790 - accuracy: 0.3231 - val_loss: 2.8354 - val_accuracy: 0.0649\n",
+      "Epoch 75/5000\n",
+      "919/919 - 3s - loss: 2.9803 - accuracy: 0.3231 - val_loss: 2.8324 - val_accuracy: 0.0650\n",
+      "Epoch 76/5000\n",
+      "919/919 - 3s - loss: 3.0280 - accuracy: 0.3210 - val_loss: 2.8270 - val_accuracy: 0.0651\n",
+      "Epoch 77/5000\n",
+      "919/919 - 3s - loss: 2.8999 - accuracy: 0.3251 - val_loss: 2.8285 - val_accuracy: 0.0652\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 78/5000\n",
+      "919/919 - 3s - loss: 2.9820 - accuracy: 0.3229 - val_loss: 2.8253 - val_accuracy: 0.0653\n",
+      "Epoch 79/5000\n",
+      "919/919 - 3s - loss: 2.9667 - accuracy: 0.3269 - val_loss: 2.8239 - val_accuracy: 0.0656\n",
+      "Epoch 80/5000\n",
+      "919/919 - 3s - loss: 2.7462 - accuracy: 0.3269 - val_loss: 2.8213 - val_accuracy: 0.0657\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 81/5000\n",
+      "919/919 - 3s - loss: 2.9387 - accuracy: 0.3261 - val_loss: 2.8216 - val_accuracy: 0.0654\n",
+      "Epoch 82/5000\n",
+      "919/919 - 3s - loss: 3.1069 - accuracy: 0.3270 - val_loss: 2.8208 - val_accuracy: 0.0653\n",
+      "Epoch 83/5000\n",
+      "919/919 - 3s - loss: 2.9728 - accuracy: 0.3270 - val_loss: 2.8225 - val_accuracy: 0.0653\n",
+      "Epoch 84/5000\n",
+      "919/919 - 3s - loss: 2.8798 - accuracy: 0.3262 - val_loss: 2.8215 - val_accuracy: 0.0651\n",
+      "Epoch 85/5000\n",
+      "919/919 - 3s - loss: 3.2898 - accuracy: 0.3289 - val_loss: 2.8217 - val_accuracy: 0.0653\n",
+      "Epoch 86/5000\n",
+      "919/919 - 3s - loss: 2.7544 - accuracy: 0.3276 - val_loss: 2.8204 - val_accuracy: 0.0654\n",
+      "Epoch 87/5000\n",
+      "919/919 - 3s - loss: 2.7426 - accuracy: 0.3248 - val_loss: 2.8220 - val_accuracy: 0.0652\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 88/5000\n",
+      "919/919 - 3s - loss: 2.8542 - accuracy: 0.3272 - val_loss: 2.8226 - val_accuracy: 0.0651\n",
+      "Epoch 89/5000\n",
+      "919/919 - 3s - loss: 2.8704 - accuracy: 0.3231 - val_loss: 2.8222 - val_accuracy: 0.0651\n",
+      "Epoch 90/5000\n",
+      "919/919 - 3s - loss: 2.7393 - accuracy: 0.3252 - val_loss: 2.8199 - val_accuracy: 0.0652\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 91/5000\n",
+      "919/919 - 3s - loss: 2.8665 - accuracy: 0.3284 - val_loss: 2.8183 - val_accuracy: 0.0653\n",
+      "Epoch 92/5000\n",
+      "919/919 - 3s - loss: 2.6440 - accuracy: 0.3288 - val_loss: 2.8198 - val_accuracy: 0.0651\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 93/5000\n",
+      "919/919 - 3s - loss: 3.0581 - accuracy: 0.3233 - val_loss: 2.8183 - val_accuracy: 0.0651\n",
+      "Epoch 94/5000\n",
+      "919/919 - 3s - loss: 2.8854 - accuracy: 0.3306 - val_loss: 2.8180 - val_accuracy: 0.0651\n",
+      "Epoch 95/5000\n",
+      "919/919 - 3s - loss: 2.6791 - accuracy: 0.3310 - val_loss: 2.8157 - val_accuracy: 0.0650\n",
+      "Epoch 96/5000\n",
+      "919/919 - 3s - loss: 2.6548 - accuracy: 0.3312 - val_loss: 2.8156 - val_accuracy: 0.0650\n",
+      "Epoch 97/5000\n",
+      "919/919 - 3s - loss: 2.8296 - accuracy: 0.3295 - val_loss: 2.8170 - val_accuracy: 0.0651\n",
+      "Epoch 98/5000\n",
+      "919/919 - 3s - loss: 2.7311 - accuracy: 0.3270 - val_loss: 2.8169 - val_accuracy: 0.0650\n",
+      "Epoch 99/5000\n",
+      "919/919 - 3s - loss: 2.9752 - accuracy: 0.3255 - val_loss: 2.8169 - val_accuracy: 0.0650\n",
+      "Epoch 100/5000\n",
+      "919/919 - 3s - loss: 2.7121 - accuracy: 0.3284 - val_loss: 2.8154 - val_accuracy: 0.0650\n",
+      "Epoch 101/5000\n",
+      "919/919 - 3s - loss: 2.5806 - accuracy: 0.3296 - val_loss: 2.8143 - val_accuracy: 0.0650\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 102/5000\n",
+      "919/919 - 3s - loss: 2.6190 - accuracy: 0.3329 - val_loss: 2.8138 - val_accuracy: 0.0649\n",
+      "Epoch 103/5000\n",
+      "919/919 - 3s - loss: 2.6306 - accuracy: 0.3282 - val_loss: 2.8158 - val_accuracy: 0.0649\n",
+      "Epoch 104/5000\n",
+      "919/919 - 3s - loss: 2.9201 - accuracy: 0.3282 - val_loss: 2.8170 - val_accuracy: 0.0649\n",
+      "Epoch 105/5000\n",
+      "919/919 - 3s - loss: 2.6691 - accuracy: 0.3305 - val_loss: 2.8168 - val_accuracy: 0.0647\n",
+      "Epoch 106/5000\n",
+      "919/919 - 3s - loss: 2.5750 - accuracy: 0.3303 - val_loss: 2.8165 - val_accuracy: 0.0648\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 107/5000\n",
+      "919/919 - 3s - loss: 2.6581 - accuracy: 0.3342 - val_loss: 2.8167 - val_accuracy: 0.0647\n",
+      "Epoch 108/5000\n",
+      "919/919 - 3s - loss: 2.5326 - accuracy: 0.3344 - val_loss: 2.8174 - val_accuracy: 0.0648\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 109/5000\n",
+      "919/919 - 3s - loss: 2.5910 - accuracy: 0.3334 - val_loss: 2.8169 - val_accuracy: 0.0648\n",
+      "Epoch 110/5000\n",
+      "919/919 - 3s - loss: 2.7412 - accuracy: 0.3299 - val_loss: 2.8192 - val_accuracy: 0.0648\n",
+      "Epoch 111/5000\n",
+      "919/919 - 3s - loss: 2.6254 - accuracy: 0.3315 - val_loss: 2.8208 - val_accuracy: 0.0648\n",
+      "Epoch 112/5000\n",
+      "919/919 - 3s - loss: 2.6638 - accuracy: 0.3340 - val_loss: 2.8212 - val_accuracy: 0.0648\n",
+      "Epoch 113/5000\n",
+      "919/919 - 3s - loss: 2.5852 - accuracy: 0.3329 - val_loss: 2.8232 - val_accuracy: 0.0646\n",
+      "Epoch 114/5000\n",
+      "919/919 - 3s - loss: 2.8237 - accuracy: 0.3352 - val_loss: 2.8233 - val_accuracy: 0.0646\n",
+      "Epoch 115/5000\n",
+      "919/919 - 3s - loss: 2.6184 - accuracy: 0.3319 - val_loss: 2.8237 - val_accuracy: 0.0646\n",
+      "Epoch 116/5000\n",
+      "919/919 - 3s - loss: 2.7341 - accuracy: 0.3395 - val_loss: 2.8258 - val_accuracy: 0.0644\n",
+      "Epoch 117/5000\n",
+      "919/919 - 3s - loss: 2.8301 - accuracy: 0.3296 - val_loss: 2.8274 - val_accuracy: 0.0635\n",
+      "Epoch 118/5000\n",
+      "919/919 - 3s - loss: 2.6465 - accuracy: 0.3358 - val_loss: 2.8292 - val_accuracy: 0.0628\n",
+      "Epoch 119/5000\n",
+      "919/919 - 3s - loss: 2.4833 - accuracy: 0.3326 - val_loss: 2.8295 - val_accuracy: 0.0626\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 120/5000\n",
+      "919/919 - 3s - loss: 2.5426 - accuracy: 0.3337 - val_loss: 2.8289 - val_accuracy: 0.0626\n",
+      "Epoch 121/5000\n",
+      "919/919 - 3s - loss: 2.5234 - accuracy: 0.3349 - val_loss: 2.8283 - val_accuracy: 0.0625\n",
+      "Epoch 122/5000\n",
+      "919/919 - 3s - loss: 2.5861 - accuracy: 0.3327 - val_loss: 2.8279 - val_accuracy: 0.0626\n",
+      "Epoch 123/5000\n",
+      "919/919 - 3s - loss: 2.4892 - accuracy: 0.3378 - val_loss: 2.8298 - val_accuracy: 0.0638\n",
+      "Epoch 124/5000\n",
+      "919/919 - 3s - loss: 2.5579 - accuracy: 0.3339 - val_loss: 2.8326 - val_accuracy: 0.0637\n",
+      "Epoch 125/5000\n",
+      "919/919 - 3s - loss: 2.4498 - accuracy: 0.3386 - val_loss: 2.8324 - val_accuracy: 0.0644\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 126/5000\n",
+      "919/919 - 3s - loss: 2.4274 - accuracy: 0.3393 - val_loss: 2.8335 - val_accuracy: 0.0638\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 127/5000\n",
+      "919/919 - 3s - loss: 2.4585 - accuracy: 0.3339 - val_loss: 2.8331 - val_accuracy: 0.0642\n",
+      "Epoch 128/5000\n",
+      "919/919 - 3s - loss: 2.4875 - accuracy: 0.3357 - val_loss: 2.8334 - val_accuracy: 0.0643\n",
+      "Epoch 129/5000\n",
+      "919/919 - 3s - loss: 2.4999 - accuracy: 0.3344 - val_loss: 2.8348 - val_accuracy: 0.0643\n",
+      "Epoch 130/5000\n",
+      "919/919 - 3s - loss: 2.4078 - accuracy: 0.3426 - val_loss: 2.8321 - val_accuracy: 0.0644\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 131/5000\n",
+      "919/919 - 3s - loss: 2.4465 - accuracy: 0.3359 - val_loss: 2.8329 - val_accuracy: 0.0643\n",
+      "Epoch 132/5000\n",
+      "919/919 - 3s - loss: 2.6081 - accuracy: 0.3391 - val_loss: 2.8341 - val_accuracy: 0.0644\n",
+      "Epoch 133/5000\n",
+      "919/919 - 3s - loss: 2.4392 - accuracy: 0.3373 - val_loss: 2.8369 - val_accuracy: 0.0640\n",
+      "Epoch 134/5000\n",
+      "919/919 - 3s - loss: 2.4442 - accuracy: 0.3397 - val_loss: 2.8356 - val_accuracy: 0.0642\n",
+      "Epoch 135/5000\n",
+      "919/919 - 3s - loss: 2.3875 - accuracy: 0.3434 - val_loss: 2.8381 - val_accuracy: 0.0633\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 136/5000\n",
+      "919/919 - 3s - loss: 2.3957 - accuracy: 0.3415 - val_loss: 2.8419 - val_accuracy: 0.0629\n",
+      "Epoch 137/5000\n",
+      "919/919 - 3s - loss: 2.3820 - accuracy: 0.3401 - val_loss: 2.8437 - val_accuracy: 0.0627\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 138/5000\n",
+      "919/919 - 3s - loss: 2.5211 - accuracy: 0.3397 - val_loss: 2.8448 - val_accuracy: 0.0627\n",
+      "Epoch 139/5000\n",
+      "919/919 - 3s - loss: 2.4213 - accuracy: 0.3391 - val_loss: 2.8465 - val_accuracy: 0.0627\n",
+      "Epoch 140/5000\n",
+      "919/919 - 3s - loss: 2.4169 - accuracy: 0.3433 - val_loss: 2.8430 - val_accuracy: 0.0627\n",
+      "Epoch 141/5000\n",
+      "919/919 - 3s - loss: 2.7834 - accuracy: 0.3405 - val_loss: 2.8433 - val_accuracy: 0.0629\n",
+      "Epoch 142/5000\n",
+      "919/919 - 3s - loss: 2.6597 - accuracy: 0.3412 - val_loss: 2.8428 - val_accuracy: 0.0629\n",
+      "Epoch 143/5000\n",
+      "919/919 - 3s - loss: 2.6679 - accuracy: 0.3459 - val_loss: 2.8418 - val_accuracy: 0.0639\n",
+      "Epoch 144/5000\n",
+      "919/919 - 3s - loss: 2.3631 - accuracy: 0.3433 - val_loss: 2.8429 - val_accuracy: 0.0639\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 145/5000\n",
+      "919/919 - 3s - loss: 2.4199 - accuracy: 0.3407 - val_loss: 2.8425 - val_accuracy: 0.0639\n",
+      "Epoch 146/5000\n",
+      "919/919 - 3s - loss: 2.3703 - accuracy: 0.3425 - val_loss: 2.8426 - val_accuracy: 0.0629\n",
+      "Epoch 147/5000\n",
+      "919/919 - 3s - loss: 2.3832 - accuracy: 0.3437 - val_loss: 2.8467 - val_accuracy: 0.0629\n",
+      "Epoch 148/5000\n",
+      "919/919 - 3s - loss: 2.3836 - accuracy: 0.3421 - val_loss: 2.8463 - val_accuracy: 0.0629\n",
+      "Epoch 149/5000\n",
+      "919/919 - 3s - loss: 2.4379 - accuracy: 0.3433 - val_loss: 2.8495 - val_accuracy: 0.0626\n",
+      "Epoch 150/5000\n",
+      "919/919 - 3s - loss: 2.3438 - accuracy: 0.3425 - val_loss: 2.8501 - val_accuracy: 0.0628\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 151/5000\n",
+      "919/919 - 3s - loss: 2.3460 - accuracy: 0.3415 - val_loss: 2.8506 - val_accuracy: 0.0626\n",
+      "Epoch 152/5000\n",
+      "919/919 - 3s - loss: 2.4267 - accuracy: 0.3419 - val_loss: 2.8522 - val_accuracy: 0.0626\n",
+      "Epoch 153/5000\n",
+      "919/919 - 3s - loss: 2.4123 - accuracy: 0.3448 - val_loss: 2.8529 - val_accuracy: 0.0626\n",
+      "Epoch 154/5000\n",
+      "919/919 - 3s - loss: 2.3302 - accuracy: 0.3451 - val_loss: 2.8553 - val_accuracy: 0.0626\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 155/5000\n",
+      "919/919 - 3s - loss: 2.3921 - accuracy: 0.3470 - val_loss: 2.8546 - val_accuracy: 0.0626\n",
+      "Epoch 156/5000\n",
+      "919/919 - 3s - loss: 2.3823 - accuracy: 0.3448 - val_loss: 2.8555 - val_accuracy: 0.0626\n",
+      "Epoch 157/5000\n",
+      "919/919 - 3s - loss: 2.4439 - accuracy: 0.3463 - val_loss: 2.8573 - val_accuracy: 0.0626\n",
+      "Epoch 158/5000\n",
+      "919/919 - 3s - loss: 2.4478 - accuracy: 0.3462 - val_loss: 2.8602 - val_accuracy: 0.0626\n",
+      "Epoch 159/5000\n",
+      "919/919 - 3s - loss: 2.3972 - accuracy: 0.3448 - val_loss: 2.8621 - val_accuracy: 0.0626\n",
+      "Epoch 160/5000\n",
+      "919/919 - 3s - loss: 2.3096 - accuracy: 0.3460 - val_loss: 2.8616 - val_accuracy: 0.0626\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 161/5000\n",
+      "919/919 - 4s - loss: 2.3132 - accuracy: 0.3474 - val_loss: 2.8636 - val_accuracy: 0.0626\n",
+      "Epoch 162/5000\n",
+      "919/919 - 5s - loss: 2.3925 - accuracy: 0.3452 - val_loss: 2.8636 - val_accuracy: 0.0625\n",
+      "Epoch 163/5000\n",
+      "919/919 - 6s - loss: 2.4496 - accuracy: 0.3490 - val_loss: 2.8656 - val_accuracy: 0.0625\n",
+      "Epoch 164/5000\n",
+      "919/919 - 6s - loss: 2.3958 - accuracy: 0.3488 - val_loss: 2.8649 - val_accuracy: 0.0625\n",
+      "Epoch 165/5000\n",
+      "919/919 - 5s - loss: 2.3270 - accuracy: 0.3456 - val_loss: 2.8681 - val_accuracy: 0.0625\n",
+      "Epoch 166/5000\n",
+      "919/919 - 5s - loss: 2.3242 - accuracy: 0.3424 - val_loss: 2.8692 - val_accuracy: 0.0625\n",
+      "Epoch 167/5000\n",
+      "919/919 - 5s - loss: 2.3024 - accuracy: 0.3452 - val_loss: 2.8720 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 168/5000\n",
+      "919/919 - 3s - loss: 2.2865 - accuracy: 0.3478 - val_loss: 2.8699 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 169/5000\n",
+      "919/919 - 3s - loss: 2.3413 - accuracy: 0.3497 - val_loss: 2.8680 - val_accuracy: 0.0625\n",
+      "Epoch 170/5000\n",
+      "919/919 - 3s - loss: 2.3776 - accuracy: 0.3507 - val_loss: 2.8672 - val_accuracy: 0.0625\n",
+      "Epoch 171/5000\n",
+      "919/919 - 3s - loss: 2.2974 - accuracy: 0.3465 - val_loss: 2.8710 - val_accuracy: 0.0625\n",
+      "Epoch 172/5000\n",
+      "919/919 - 3s - loss: 2.2956 - accuracy: 0.3491 - val_loss: 2.8744 - val_accuracy: 0.0625\n",
+      "Epoch 173/5000\n",
+      "919/919 - 3s - loss: 2.2875 - accuracy: 0.3509 - val_loss: 2.8754 - val_accuracy: 0.0626\n",
+      "Epoch 174/5000\n",
+      "919/919 - 3s - loss: 2.2870 - accuracy: 0.3488 - val_loss: 2.8767 - val_accuracy: 0.0625\n",
+      "Epoch 175/5000\n",
+      "919/919 - 3s - loss: 2.2924 - accuracy: 0.3493 - val_loss: 2.8794 - val_accuracy: 0.0625\n",
+      "Epoch 176/5000\n",
+      "919/919 - 3s - loss: 2.2845 - accuracy: 0.3512 - val_loss: 2.8801 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 177/5000\n",
+      "919/919 - 3s - loss: 2.3069 - accuracy: 0.3533 - val_loss: 2.8803 - val_accuracy: 0.0625\n",
+      "Epoch 178/5000\n",
+      "919/919 - 3s - loss: 2.3213 - accuracy: 0.3522 - val_loss: 2.8807 - val_accuracy: 0.0623\n",
+      "Epoch 179/5000\n",
+      "919/919 - 3s - loss: 2.3220 - accuracy: 0.3484 - val_loss: 2.8816 - val_accuracy: 0.0623\n",
+      "Epoch 180/5000\n",
+      "919/919 - 3s - loss: 2.3157 - accuracy: 0.3520 - val_loss: 2.8847 - val_accuracy: 0.0624\n",
+      "Epoch 181/5000\n",
+      "919/919 - 3s - loss: 2.2516 - accuracy: 0.3541 - val_loss: 2.8829 - val_accuracy: 0.0623\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 182/5000\n",
+      "919/919 - 3s - loss: 2.3289 - accuracy: 0.3468 - val_loss: 2.8850 - val_accuracy: 0.0624\n",
+      "Epoch 183/5000\n",
+      "919/919 - 3s - loss: 2.2914 - accuracy: 0.3507 - val_loss: 2.8855 - val_accuracy: 0.0623\n",
+      "Epoch 184/5000\n",
+      "919/919 - 3s - loss: 2.3266 - accuracy: 0.3531 - val_loss: 2.8863 - val_accuracy: 0.0623\n",
+      "Epoch 185/5000\n",
+      "919/919 - 3s - loss: 2.2714 - accuracy: 0.3496 - val_loss: 2.8867 - val_accuracy: 0.0623\n",
+      "Epoch 186/5000\n",
+      "919/919 - 3s - loss: 2.3573 - accuracy: 0.3512 - val_loss: 2.8865 - val_accuracy: 0.0623\n",
+      "Epoch 187/5000\n",
+      "919/919 - 3s - loss: 2.2626 - accuracy: 0.3518 - val_loss: 2.8873 - val_accuracy: 0.0625\n",
+      "Epoch 188/5000\n",
+      "919/919 - 3s - loss: 2.2456 - accuracy: 0.3534 - val_loss: 2.8886 - val_accuracy: 0.0623\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 189/5000\n",
+      "919/919 - 3s - loss: 2.2515 - accuracy: 0.3522 - val_loss: 2.8937 - val_accuracy: 0.0623\n",
+      "Epoch 190/5000\n",
+      "919/919 - 3s - loss: 2.3863 - accuracy: 0.3539 - val_loss: 2.8990 - val_accuracy: 0.0623\n",
+      "Epoch 191/5000\n",
+      "919/919 - 3s - loss: 2.3045 - accuracy: 0.3548 - val_loss: 2.9092 - val_accuracy: 0.0624\n",
+      "Epoch 192/5000\n",
+      "919/919 - 3s - loss: 2.2054 - accuracy: 0.3563 - val_loss: 2.9080 - val_accuracy: 0.0623\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 193/5000\n",
+      "919/919 - 3s - loss: 2.2138 - accuracy: 0.3555 - val_loss: 2.9075 - val_accuracy: 0.0623\n",
+      "Epoch 194/5000\n",
+      "919/919 - 3s - loss: 2.2291 - accuracy: 0.3572 - val_loss: 2.9126 - val_accuracy: 0.0624\n",
+      "Epoch 195/5000\n",
+      "919/919 - 3s - loss: 2.2383 - accuracy: 0.3552 - val_loss: 2.9155 - val_accuracy: 0.0623\n",
+      "Epoch 196/5000\n",
+      "919/919 - 3s - loss: 2.4005 - accuracy: 0.3525 - val_loss: 2.9226 - val_accuracy: 0.0623\n",
+      "Epoch 197/5000\n",
+      "919/919 - 3s - loss: 2.2910 - accuracy: 0.3553 - val_loss: 2.9247 - val_accuracy: 0.0624\n",
+      "Epoch 198/5000\n",
+      "919/919 - 3s - loss: 2.2352 - accuracy: 0.3561 - val_loss: 2.9230 - val_accuracy: 0.0625\n",
+      "Epoch 199/5000\n",
+      "919/919 - 3s - loss: 2.2236 - accuracy: 0.3575 - val_loss: 2.9243 - val_accuracy: 0.0625\n",
+      "Epoch 200/5000\n",
+      "919/919 - 3s - loss: 2.2283 - accuracy: 0.3574 - val_loss: 2.9223 - val_accuracy: 0.0624\n",
+      "Epoch 201/5000\n",
+      "919/919 - 3s - loss: 2.2457 - accuracy: 0.3552 - val_loss: 2.9267 - val_accuracy: 0.0625\n",
+      "Epoch 202/5000\n",
+      "919/919 - 3s - loss: 2.2946 - accuracy: 0.3559 - val_loss: 2.9311 - val_accuracy: 0.0625\n",
+      "Epoch 203/5000\n",
+      "919/919 - 3s - loss: 2.2127 - accuracy: 0.3562 - val_loss: 2.9363 - val_accuracy: 0.0625\n",
+      "Epoch 204/5000\n",
+      "919/919 - 3s - loss: 2.2325 - accuracy: 0.3570 - val_loss: 2.9336 - val_accuracy: 0.0625\n",
+      "Epoch 205/5000\n",
+      "919/919 - 3s - loss: 2.2153 - accuracy: 0.3561 - val_loss: 2.9335 - val_accuracy: 0.0625\n",
+      "Epoch 206/5000\n",
+      "919/919 - 3s - loss: 2.2093 - accuracy: 0.3568 - val_loss: 2.9365 - val_accuracy: 0.0625\n",
+      "Epoch 207/5000\n",
+      "919/919 - 3s - loss: 2.5075 - accuracy: 0.3556 - val_loss: 2.9303 - val_accuracy: 0.0625\n",
+      "Epoch 208/5000\n",
+      "919/919 - 3s - loss: 2.2692 - accuracy: 0.3536 - val_loss: 2.9274 - val_accuracy: 0.0625\n",
+      "Epoch 209/5000\n",
+      "919/919 - 3s - loss: 2.3100 - accuracy: 0.3540 - val_loss: 2.9273 - val_accuracy: 0.0625\n",
+      "Epoch 210/5000\n",
+      "919/919 - 3s - loss: 2.2290 - accuracy: 0.3609 - val_loss: 2.9290 - val_accuracy: 0.0625\n",
+      "Epoch 211/5000\n",
+      "919/919 - 3s - loss: 2.2229 - accuracy: 0.3590 - val_loss: 2.9287 - val_accuracy: 0.0625\n",
+      "Epoch 212/5000\n",
+      "919/919 - 3s - loss: 2.2107 - accuracy: 0.3567 - val_loss: 2.9287 - val_accuracy: 0.0624\n",
+      "Epoch 213/5000\n",
+      "919/919 - 3s - loss: 2.2098 - accuracy: 0.3584 - val_loss: 2.9329 - val_accuracy: 0.0623\n",
+      "Epoch 214/5000\n",
+      "919/919 - 3s - loss: 2.2960 - accuracy: 0.3537 - val_loss: 2.9394 - val_accuracy: 0.0625\n",
+      "Epoch 215/5000\n",
+      "919/919 - 3s - loss: 2.2105 - accuracy: 0.3592 - val_loss: 2.9406 - val_accuracy: 0.0624\n",
+      "Epoch 216/5000\n",
+      "919/919 - 3s - loss: 2.2145 - accuracy: 0.3573 - val_loss: 2.9455 - val_accuracy: 0.0624\n",
+      "Epoch 217/5000\n",
+      "919/919 - 3s - loss: 2.2188 - accuracy: 0.3553 - val_loss: 2.9475 - val_accuracy: 0.0624\n",
+      "Epoch 218/5000\n",
+      "919/919 - 3s - loss: 2.4651 - accuracy: 0.3582 - val_loss: 2.9532 - val_accuracy: 0.0625\n",
+      "Epoch 219/5000\n",
+      "919/919 - 3s - loss: 2.2017 - accuracy: 0.3610 - val_loss: 2.9531 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 220/5000\n",
+      "919/919 - 3s - loss: 2.2811 - accuracy: 0.3604 - val_loss: 2.9571 - val_accuracy: 0.0624\n",
+      "Epoch 221/5000\n",
+      "919/919 - 3s - loss: 2.2548 - accuracy: 0.3590 - val_loss: 2.9602 - val_accuracy: 0.0625\n",
+      "Epoch 222/5000\n",
+      "919/919 - 3s - loss: 2.2244 - accuracy: 0.3578 - val_loss: 2.9632 - val_accuracy: 0.0625\n",
+      "Epoch 223/5000\n",
+      "919/919 - 3s - loss: 2.2826 - accuracy: 0.3582 - val_loss: 2.9586 - val_accuracy: 0.0625\n",
+      "Epoch 224/5000\n",
+      "919/919 - 3s - loss: 2.2683 - accuracy: 0.3577 - val_loss: 2.9549 - val_accuracy: 0.0625\n",
+      "Epoch 225/5000\n",
+      "919/919 - 3s - loss: 2.2097 - accuracy: 0.3627 - val_loss: 2.9515 - val_accuracy: 0.0625\n",
+      "Epoch 226/5000\n",
+      "919/919 - 3s - loss: 2.2531 - accuracy: 0.3563 - val_loss: 2.9581 - val_accuracy: 0.0625\n",
+      "Epoch 227/5000\n",
+      "919/919 - 3s - loss: 2.3547 - accuracy: 0.3618 - val_loss: 2.9588 - val_accuracy: 0.0625\n",
+      "Epoch 228/5000\n",
+      "919/919 - 3s - loss: 2.1762 - accuracy: 0.3607 - val_loss: 2.9623 - val_accuracy: 0.0624\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 229/5000\n",
+      "919/919 - 3s - loss: 2.2848 - accuracy: 0.3611 - val_loss: 2.9615 - val_accuracy: 0.0625\n",
+      "Epoch 230/5000\n",
+      "919/919 - 3s - loss: 2.1685 - accuracy: 0.3613 - val_loss: 2.9702 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 231/5000\n",
+      "919/919 - 3s - loss: 2.1768 - accuracy: 0.3606 - val_loss: 2.9694 - val_accuracy: 0.0625\n",
+      "Epoch 232/5000\n",
+      "919/919 - 3s - loss: 2.2052 - accuracy: 0.3602 - val_loss: 2.9764 - val_accuracy: 0.0624\n",
+      "Epoch 233/5000\n",
+      "919/919 - 3s - loss: 2.1743 - accuracy: 0.3612 - val_loss: 2.9752 - val_accuracy: 0.0624\n",
+      "Epoch 234/5000\n",
+      "919/919 - 3s - loss: 2.1772 - accuracy: 0.3643 - val_loss: 2.9809 - val_accuracy: 0.0625\n",
+      "Epoch 235/5000\n",
+      "919/919 - 3s - loss: 2.1642 - accuracy: 0.3633 - val_loss: 2.9819 - val_accuracy: 0.0623\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 236/5000\n",
+      "919/919 - 3s - loss: 2.1507 - accuracy: 0.3614 - val_loss: 2.9885 - val_accuracy: 0.0624\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 237/5000\n",
+      "919/919 - 3s - loss: 2.1580 - accuracy: 0.3615 - val_loss: 2.9926 - val_accuracy: 0.0625\n",
+      "Epoch 238/5000\n",
+      "919/919 - 3s - loss: 2.2115 - accuracy: 0.3610 - val_loss: 2.9908 - val_accuracy: 0.0625\n",
+      "Epoch 239/5000\n",
+      "919/919 - 3s - loss: 2.2118 - accuracy: 0.3639 - val_loss: 2.9933 - val_accuracy: 0.0624\n",
+      "Epoch 240/5000\n",
+      "919/919 - 3s - loss: 2.1668 - accuracy: 0.3609 - val_loss: 2.9915 - val_accuracy: 0.0624\n",
+      "Epoch 241/5000\n",
+      "919/919 - 3s - loss: 2.1556 - accuracy: 0.3667 - val_loss: 2.9985 - val_accuracy: 0.0624\n",
+      "Epoch 242/5000\n",
+      "919/919 - 3s - loss: 2.1646 - accuracy: 0.3629 - val_loss: 3.0027 - val_accuracy: 0.0624\n",
+      "Epoch 243/5000\n",
+      "919/919 - 3s - loss: 2.1594 - accuracy: 0.3661 - val_loss: 3.0043 - val_accuracy: 0.0623\n",
+      "Epoch 244/5000\n",
+      "919/919 - 3s - loss: 2.1583 - accuracy: 0.3652 - val_loss: 3.0065 - val_accuracy: 0.0624\n",
+      "Epoch 245/5000\n",
+      "919/919 - 3s - loss: 2.1776 - accuracy: 0.3621 - val_loss: 3.0127 - val_accuracy: 0.0623\n",
+      "Epoch 246/5000\n",
+      "919/919 - 3s - loss: 2.2240 - accuracy: 0.3635 - val_loss: 3.0087 - val_accuracy: 0.0623\n",
+      "Epoch 247/5000\n",
+      "919/919 - 3s - loss: 2.1491 - accuracy: 0.3641 - val_loss: 3.0126 - val_accuracy: 0.0623\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 248/5000\n",
+      "919/919 - 3s - loss: 2.1894 - accuracy: 0.3618 - val_loss: 3.0091 - val_accuracy: 0.0623\n",
+      "Epoch 249/5000\n",
+      "919/919 - 3s - loss: 2.3593 - accuracy: 0.3653 - val_loss: 3.0132 - val_accuracy: 0.0623\n",
+      "Epoch 250/5000\n",
+      "919/919 - 3s - loss: 2.1565 - accuracy: 0.3662 - val_loss: 3.0144 - val_accuracy: 0.0623\n",
+      "Epoch 251/5000\n",
+      "919/919 - 3s - loss: 2.1541 - accuracy: 0.3640 - val_loss: 3.0145 - val_accuracy: 0.0623\n",
+      "Epoch 252/5000\n",
+      "919/919 - 3s - loss: 2.1740 - accuracy: 0.3622 - val_loss: 3.0060 - val_accuracy: 0.0623\n",
+      "Epoch 253/5000\n",
+      "919/919 - 3s - loss: 2.1460 - accuracy: 0.3674 - val_loss: 3.0054 - val_accuracy: 0.0623\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 254/5000\n",
+      "919/919 - 3s - loss: 2.1873 - accuracy: 0.3636 - val_loss: 3.0084 - val_accuracy: 0.0623\n",
+      "Epoch 255/5000\n",
+      "919/919 - 3s - loss: 2.1277 - accuracy: 0.3722 - val_loss: 3.0165 - val_accuracy: 0.0623\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 256/5000\n",
+      "919/919 - 3s - loss: 2.2145 - accuracy: 0.3670 - val_loss: 3.0223 - val_accuracy: 0.0624\n",
+      "Epoch 257/5000\n",
+      "919/919 - 3s - loss: 2.1708 - accuracy: 0.3651 - val_loss: 3.0183 - val_accuracy: 0.0623\n",
+      "Epoch 258/5000\n",
+      "919/919 - 3s - loss: 2.1397 - accuracy: 0.3663 - val_loss: 3.0024 - val_accuracy: 0.0623\n",
+      "Epoch 259/5000\n",
+      "919/919 - 3s - loss: 2.2266 - accuracy: 0.3693 - val_loss: 3.0024 - val_accuracy: 0.0624\n",
+      "Epoch 260/5000\n",
+      "919/919 - 3s - loss: 2.1920 - accuracy: 0.3657 - val_loss: 2.9992 - val_accuracy: 0.0625\n",
+      "Epoch 261/5000\n",
+      "919/919 - 3s - loss: 2.1250 - accuracy: 0.3706 - val_loss: 2.9972 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 262/5000\n",
+      "919/919 - 3s - loss: 2.1865 - accuracy: 0.3659 - val_loss: 2.9952 - val_accuracy: 0.0624\n",
+      "Epoch 263/5000\n",
+      "919/919 - 3s - loss: 2.1381 - accuracy: 0.3688 - val_loss: 2.9969 - val_accuracy: 0.0624\n",
+      "Epoch 264/5000\n",
+      "919/919 - 3s - loss: 2.1856 - accuracy: 0.3699 - val_loss: 2.9976 - val_accuracy: 0.0624\n",
+      "Epoch 265/5000\n",
+      "919/919 - 3s - loss: 2.1446 - accuracy: 0.3695 - val_loss: 3.0052 - val_accuracy: 0.0624\n",
+      "Epoch 266/5000\n",
+      "919/919 - 3s - loss: 2.1232 - accuracy: 0.3659 - val_loss: 3.0167 - val_accuracy: 0.0624\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 267/5000\n",
+      "919/919 - 3s - loss: 2.1394 - accuracy: 0.3673 - val_loss: 3.0278 - val_accuracy: 0.0624\n",
+      "Epoch 268/5000\n",
+      "919/919 - 3s - loss: 2.2422 - accuracy: 0.3690 - val_loss: 3.0343 - val_accuracy: 0.0624\n",
+      "Epoch 269/5000\n",
+      "919/919 - 3s - loss: 2.1221 - accuracy: 0.3699 - val_loss: 3.0351 - val_accuracy: 0.0623\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 270/5000\n",
+      "919/919 - 3s - loss: 2.1266 - accuracy: 0.3686 - val_loss: 3.0495 - val_accuracy: 0.0625\n",
+      "Epoch 271/5000\n",
+      "919/919 - 3s - loss: 2.1331 - accuracy: 0.3677 - val_loss: 3.0482 - val_accuracy: 0.0624\n",
+      "Epoch 272/5000\n",
+      "919/919 - 3s - loss: 2.1253 - accuracy: 0.3684 - val_loss: 3.0644 - val_accuracy: 0.0624\n",
+      "Epoch 273/5000\n",
+      "919/919 - 3s - loss: 2.1377 - accuracy: 0.3668 - val_loss: 3.0630 - val_accuracy: 0.0625\n",
+      "Epoch 274/5000\n",
+      "919/919 - 4s - loss: 2.1248 - accuracy: 0.3677 - val_loss: 3.0623 - val_accuracy: 0.0625\n",
+      "Epoch 275/5000\n",
+      "919/919 - 5s - loss: 2.1161 - accuracy: 0.3733 - val_loss: 3.0555 - val_accuracy: 0.0624\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 276/5000\n",
+      "919/919 - 6s - loss: 2.1090 - accuracy: 0.3735 - val_loss: 3.0560 - val_accuracy: 0.0624\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 277/5000\n",
+      "919/919 - 5s - loss: 2.1316 - accuracy: 0.3718 - val_loss: 3.0604 - val_accuracy: 0.0624\n",
+      "Epoch 278/5000\n",
+      "919/919 - 5s - loss: 2.2255 - accuracy: 0.3729 - val_loss: 3.0502 - val_accuracy: 0.0625\n",
+      "Epoch 279/5000\n",
+      "919/919 - 5s - loss: 2.1171 - accuracy: 0.3700 - val_loss: 3.0525 - val_accuracy: 0.0626\n",
+      "Epoch 280/5000\n",
+      "919/919 - 3s - loss: 2.1358 - accuracy: 0.3671 - val_loss: 3.0522 - val_accuracy: 0.0626\n",
+      "Epoch 281/5000\n",
+      "919/919 - 3s - loss: 2.1017 - accuracy: 0.3719 - val_loss: 3.0447 - val_accuracy: 0.0626\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 282/5000\n",
+      "919/919 - 3s - loss: 2.1065 - accuracy: 0.3737 - val_loss: 3.0500 - val_accuracy: 0.0626\n",
+      "Epoch 283/5000\n",
+      "919/919 - 3s - loss: 2.1114 - accuracy: 0.3736 - val_loss: 3.0552 - val_accuracy: 0.0626\n",
+      "Epoch 284/5000\n",
+      "919/919 - 3s - loss: 2.2134 - accuracy: 0.3697 - val_loss: 3.0434 - val_accuracy: 0.0626\n",
+      "Epoch 285/5000\n",
+      "919/919 - 3s - loss: 2.1332 - accuracy: 0.3716 - val_loss: 3.0420 - val_accuracy: 0.0626\n",
+      "Epoch 286/5000\n",
+      "919/919 - 3s - loss: 2.1074 - accuracy: 0.3674 - val_loss: 3.0457 - val_accuracy: 0.0626\n",
+      "Epoch 287/5000\n",
+      "919/919 - 3s - loss: 2.1400 - accuracy: 0.3731 - val_loss: 3.0335 - val_accuracy: 0.0626\n",
+      "Epoch 288/5000\n",
+      "919/919 - 3s - loss: 2.2044 - accuracy: 0.3731 - val_loss: 3.0347 - val_accuracy: 0.0626\n",
+      "Epoch 289/5000\n",
+      "919/919 - 3s - loss: 2.1154 - accuracy: 0.3709 - val_loss: 3.0312 - val_accuracy: 0.0626\n",
+      "Epoch 290/5000\n",
+      "919/919 - 3s - loss: 2.1031 - accuracy: 0.3724 - val_loss: 3.0382 - val_accuracy: 0.0626\n",
+      "Epoch 291/5000\n",
+      "919/919 - 3s - loss: 2.1023 - accuracy: 0.3716 - val_loss: 3.0385 - val_accuracy: 0.0626\n",
+      "Epoch 292/5000\n",
+      "919/919 - 3s - loss: 2.1240 - accuracy: 0.3699 - val_loss: 3.0479 - val_accuracy: 0.0626\n",
+      "Epoch 293/5000\n",
+      "919/919 - 3s - loss: 2.1863 - accuracy: 0.3707 - val_loss: 3.0592 - val_accuracy: 0.0626\n",
+      "Epoch 294/5000\n",
+      "919/919 - 3s - loss: 2.0876 - accuracy: 0.3748 - val_loss: 3.0571 - val_accuracy: 0.0626\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 295/5000\n",
+      "919/919 - 3s - loss: 2.0932 - accuracy: 0.3713 - val_loss: 3.0595 - val_accuracy: 0.0626\n",
+      "Epoch 296/5000\n",
+      "919/919 - 3s - loss: 2.0908 - accuracy: 0.3756 - val_loss: 3.0577 - val_accuracy: 0.0626\n",
+      "Epoch 297/5000\n",
+      "919/919 - 3s - loss: 2.1006 - accuracy: 0.3720 - val_loss: 3.0622 - val_accuracy: 0.0626\n",
+      "Epoch 298/5000\n",
+      "919/919 - 3s - loss: 2.2515 - accuracy: 0.3726 - val_loss: 3.0673 - val_accuracy: 0.0626\n",
+      "Epoch 299/5000\n",
+      "919/919 - 3s - loss: 2.0895 - accuracy: 0.3750 - val_loss: 3.0678 - val_accuracy: 0.0626\n",
+      "Epoch 300/5000\n",
+      "919/919 - 3s - loss: 2.0809 - accuracy: 0.3730 - val_loss: 3.0545 - val_accuracy: 0.0626\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 301/5000\n",
+      "919/919 - 3s - loss: 2.2437 - accuracy: 0.3759 - val_loss: 3.0529 - val_accuracy: 0.0625\n",
+      "Epoch 302/5000\n",
+      "919/919 - 3s - loss: 2.1330 - accuracy: 0.3709 - val_loss: 3.0518 - val_accuracy: 0.0625\n",
+      "Epoch 303/5000\n",
+      "919/919 - 3s - loss: 2.0800 - accuracy: 0.3739 - val_loss: 3.0576 - val_accuracy: 0.0626\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 304/5000\n",
+      "919/919 - 3s - loss: 2.0792 - accuracy: 0.3752 - val_loss: 3.0510 - val_accuracy: 0.0626\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 305/5000\n",
+      "919/919 - 3s - loss: 2.1017 - accuracy: 0.3746 - val_loss: 3.0558 - val_accuracy: 0.0626\n",
+      "Epoch 306/5000\n",
+      "919/919 - 3s - loss: 2.0851 - accuracy: 0.3763 - val_loss: 3.0559 - val_accuracy: 0.0626\n",
+      "Epoch 307/5000\n",
+      "919/919 - 3s - loss: 2.0845 - accuracy: 0.3772 - val_loss: 3.0610 - val_accuracy: 0.0626\n",
+      "Epoch 308/5000\n",
+      "919/919 - 3s - loss: 2.0852 - accuracy: 0.3758 - val_loss: 3.0650 - val_accuracy: 0.0626\n",
+      "Epoch 309/5000\n",
+      "919/919 - 3s - loss: 2.1179 - accuracy: 0.3753 - val_loss: 3.0693 - val_accuracy: 0.0626\n",
+      "Epoch 310/5000\n",
+      "919/919 - 3s - loss: 2.0907 - accuracy: 0.3773 - val_loss: 3.0678 - val_accuracy: 0.0626\n",
+      "Epoch 311/5000\n",
+      "919/919 - 3s - loss: 2.1037 - accuracy: 0.3733 - val_loss: 3.0895 - val_accuracy: 0.0626\n",
+      "Epoch 312/5000\n",
+      "919/919 - 3s - loss: 2.0878 - accuracy: 0.3772 - val_loss: 3.0976 - val_accuracy: 0.0626\n",
+      "Epoch 313/5000\n",
+      "919/919 - 3s - loss: 2.0868 - accuracy: 0.3761 - val_loss: 3.0902 - val_accuracy: 0.0627\n",
+      "Epoch 314/5000\n",
+      "919/919 - 3s - loss: 2.1281 - accuracy: 0.3723 - val_loss: 3.0915 - val_accuracy: 0.0625\n",
+      "Epoch 315/5000\n",
+      "919/919 - 3s - loss: 2.0651 - accuracy: 0.3756 - val_loss: 3.0918 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 316/5000\n",
+      "919/919 - 3s - loss: 2.1142 - accuracy: 0.3771 - val_loss: 3.1058 - val_accuracy: 0.0624\n",
+      "Epoch 317/5000\n",
+      "919/919 - 3s - loss: 2.0720 - accuracy: 0.3789 - val_loss: 3.1018 - val_accuracy: 0.0624\n",
+      "Epoch 318/5000\n",
+      "919/919 - 3s - loss: 2.0571 - accuracy: 0.3759 - val_loss: 3.0902 - val_accuracy: 0.0624\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 319/5000\n",
+      "919/919 - 3s - loss: 2.0778 - accuracy: 0.3744 - val_loss: 3.0999 - val_accuracy: 0.0624\n",
+      "Epoch 320/5000\n",
+      "919/919 - 3s - loss: 2.0895 - accuracy: 0.3759 - val_loss: 3.0889 - val_accuracy: 0.0624\n",
+      "Epoch 321/5000\n",
+      "919/919 - 3s - loss: 2.0613 - accuracy: 0.3754 - val_loss: 3.0924 - val_accuracy: 0.0624\n",
+      "Epoch 322/5000\n",
+      "919/919 - 3s - loss: 2.1687 - accuracy: 0.3765 - val_loss: 3.0909 - val_accuracy: 0.0625\n",
+      "Epoch 323/5000\n",
+      "919/919 - 3s - loss: 2.0746 - accuracy: 0.3763 - val_loss: 3.0839 - val_accuracy: 0.0625\n",
+      "Epoch 324/5000\n",
+      "919/919 - 3s - loss: 2.0594 - accuracy: 0.3781 - val_loss: 3.0794 - val_accuracy: 0.0625\n",
+      "Epoch 325/5000\n",
+      "919/919 - 3s - loss: 2.1627 - accuracy: 0.3752 - val_loss: 3.0873 - val_accuracy: 0.0624\n",
+      "Epoch 326/5000\n",
+      "919/919 - 3s - loss: 2.0957 - accuracy: 0.3763 - val_loss: 3.0911 - val_accuracy: 0.0624\n",
+      "Epoch 327/5000\n",
+      "919/919 - 3s - loss: 2.0803 - accuracy: 0.3727 - val_loss: 3.0773 - val_accuracy: 0.0624\n",
+      "Epoch 328/5000\n",
+      "919/919 - 3s - loss: 2.0819 - accuracy: 0.3748 - val_loss: 3.0690 - val_accuracy: 0.0624\n",
+      "Epoch 329/5000\n",
+      "919/919 - 3s - loss: 2.0611 - accuracy: 0.3796 - val_loss: 3.0679 - val_accuracy: 0.0625\n",
+      "Epoch 330/5000\n",
+      "919/919 - 3s - loss: 2.0741 - accuracy: 0.3761 - val_loss: 3.0772 - val_accuracy: 0.0625\n",
+      "Epoch 331/5000\n",
+      "919/919 - 3s - loss: 2.0622 - accuracy: 0.3782 - val_loss: 3.0680 - val_accuracy: 0.0625\n",
+      "Epoch 332/5000\n",
+      "919/919 - 3s - loss: 2.0588 - accuracy: 0.3765 - val_loss: 3.0726 - val_accuracy: 0.0625\n",
+      "Epoch 333/5000\n",
+      "919/919 - 3s - loss: 2.2275 - accuracy: 0.3767 - val_loss: 3.0760 - val_accuracy: 0.0625\n",
+      "Epoch 334/5000\n",
+      "919/919 - 3s - loss: 2.0702 - accuracy: 0.3763 - val_loss: 3.0760 - val_accuracy: 0.0624\n",
+      "Epoch 335/5000\n",
+      "919/919 - 3s - loss: 2.0723 - accuracy: 0.3769 - val_loss: 3.0823 - val_accuracy: 0.0624\n",
+      "Epoch 336/5000\n",
+      "919/919 - 3s - loss: 2.0622 - accuracy: 0.3795 - val_loss: 3.0881 - val_accuracy: 0.0625\n",
+      "Epoch 337/5000\n",
+      "919/919 - 3s - loss: 2.0723 - accuracy: 0.3800 - val_loss: 3.0942 - val_accuracy: 0.0625\n",
+      "Epoch 338/5000\n",
+      "919/919 - 3s - loss: 2.0757 - accuracy: 0.3764 - val_loss: 3.0988 - val_accuracy: 0.0625\n",
+      "Epoch 339/5000\n",
+      "919/919 - 3s - loss: 2.0629 - accuracy: 0.3795 - val_loss: 3.1083 - val_accuracy: 0.0625\n",
+      "Epoch 340/5000\n",
+      "919/919 - 3s - loss: 2.0573 - accuracy: 0.3786 - val_loss: 3.1139 - val_accuracy: 0.0625\n",
+      "Epoch 341/5000\n",
+      "919/919 - 3s - loss: 2.0274 - accuracy: 0.3824 - val_loss: 3.1242 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 342/5000\n",
+      "919/919 - 3s - loss: 2.0547 - accuracy: 0.3815 - val_loss: 3.1141 - val_accuracy: 0.0625\n",
+      "Epoch 343/5000\n",
+      "919/919 - 3s - loss: 2.1828 - accuracy: 0.3780 - val_loss: 3.1119 - val_accuracy: 0.0625\n",
+      "Epoch 344/5000\n",
+      "919/919 - 3s - loss: 2.0786 - accuracy: 0.3769 - val_loss: 3.0944 - val_accuracy: 0.0624\n",
+      "Epoch 345/5000\n",
+      "919/919 - 3s - loss: 2.0675 - accuracy: 0.3800 - val_loss: 3.0967 - val_accuracy: 0.0625\n",
+      "Epoch 346/5000\n",
+      "919/919 - 3s - loss: 2.1197 - accuracy: 0.3802 - val_loss: 3.1003 - val_accuracy: 0.0625\n",
+      "Epoch 347/5000\n",
+      "919/919 - 3s - loss: 2.0441 - accuracy: 0.3781 - val_loss: 3.1147 - val_accuracy: 0.0625\n",
+      "Epoch 348/5000\n",
+      "919/919 - 3s - loss: 2.0371 - accuracy: 0.3808 - val_loss: 3.1137 - val_accuracy: 0.0625\n",
+      "Epoch 349/5000\n",
+      "919/919 - 3s - loss: 2.0649 - accuracy: 0.3797 - val_loss: 3.1093 - val_accuracy: 0.0625\n",
+      "Epoch 350/5000\n",
+      "919/919 - 3s - loss: 2.0406 - accuracy: 0.3784 - val_loss: 3.1092 - val_accuracy: 0.0625\n",
+      "Epoch 351/5000\n",
+      "919/919 - 3s - loss: 2.0510 - accuracy: 0.3788 - val_loss: 3.1048 - val_accuracy: 0.0625\n",
+      "Epoch 352/5000\n",
+      "919/919 - 3s - loss: 2.0470 - accuracy: 0.3794 - val_loss: 3.1019 - val_accuracy: 0.0625\n",
+      "Epoch 353/5000\n",
+      "919/919 - 3s - loss: 2.0676 - accuracy: 0.3781 - val_loss: 3.1038 - val_accuracy: 0.0625\n",
+      "Epoch 354/5000\n",
+      "919/919 - 3s - loss: 2.0240 - accuracy: 0.3846 - val_loss: 3.0996 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 355/5000\n",
+      "919/919 - 3s - loss: 2.0548 - accuracy: 0.3820 - val_loss: 3.0970 - val_accuracy: 0.0625\n",
+      "Epoch 356/5000\n",
+      "919/919 - 3s - loss: 2.0755 - accuracy: 0.3812 - val_loss: 3.1004 - val_accuracy: 0.0625\n",
+      "Epoch 357/5000\n",
+      "919/919 - 3s - loss: 2.0435 - accuracy: 0.3839 - val_loss: 3.1059 - val_accuracy: 0.0625\n",
+      "Epoch 358/5000\n",
+      "919/919 - 3s - loss: 2.0541 - accuracy: 0.3784 - val_loss: 3.1020 - val_accuracy: 0.0624\n",
+      "Epoch 359/5000\n",
+      "919/919 - 3s - loss: 2.1057 - accuracy: 0.3800 - val_loss: 3.1163 - val_accuracy: 0.0624\n",
+      "Epoch 360/5000\n",
+      "919/919 - 3s - loss: 2.1082 - accuracy: 0.3791 - val_loss: 3.1161 - val_accuracy: 0.0624\n",
+      "Epoch 361/5000\n",
+      "919/919 - 3s - loss: 2.1095 - accuracy: 0.3804 - val_loss: 3.1155 - val_accuracy: 0.0624\n",
+      "Epoch 362/5000\n",
+      "919/919 - 3s - loss: 2.0486 - accuracy: 0.3822 - val_loss: 3.1182 - val_accuracy: 0.0624\n",
+      "Epoch 363/5000\n",
+      "919/919 - 3s - loss: 2.0343 - accuracy: 0.3843 - val_loss: 3.1126 - val_accuracy: 0.0624\n",
+      "Epoch 364/5000\n",
+      "919/919 - 3s - loss: 2.0349 - accuracy: 0.3831 - val_loss: 3.1200 - val_accuracy: 0.0624\n",
+      "Epoch 365/5000\n",
+      "919/919 - 3s - loss: 2.0328 - accuracy: 0.3822 - val_loss: 3.1178 - val_accuracy: 0.0624\n",
+      "Epoch 366/5000\n",
+      "919/919 - 3s - loss: 2.0294 - accuracy: 0.3829 - val_loss: 3.1157 - val_accuracy: 0.0624\n",
+      "Epoch 367/5000\n",
+      "919/919 - 3s - loss: 2.0469 - accuracy: 0.3814 - val_loss: 3.1295 - val_accuracy: 0.0624\n",
+      "Epoch 368/5000\n",
+      "919/919 - 3s - loss: 2.1148 - accuracy: 0.3839 - val_loss: 3.1277 - val_accuracy: 0.0624\n",
+      "Epoch 369/5000\n",
+      "919/919 - 3s - loss: 2.0370 - accuracy: 0.3803 - val_loss: 3.1319 - val_accuracy: 0.0624\n",
+      "Epoch 370/5000\n",
+      "919/919 - 3s - loss: 2.0239 - accuracy: 0.3850 - val_loss: 3.1367 - val_accuracy: 0.0624\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 371/5000\n",
+      "919/919 - 3s - loss: 2.0237 - accuracy: 0.3815 - val_loss: 3.1314 - val_accuracy: 0.0624\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 372/5000\n",
+      "919/919 - 3s - loss: 2.0820 - accuracy: 0.3830 - val_loss: 3.1152 - val_accuracy: 0.0624\n",
+      "Epoch 373/5000\n",
+      "919/919 - 3s - loss: 2.0816 - accuracy: 0.3802 - val_loss: 3.1062 - val_accuracy: 0.0624\n",
+      "Epoch 374/5000\n",
+      "919/919 - 3s - loss: 2.0345 - accuracy: 0.3825 - val_loss: 3.0941 - val_accuracy: 0.0624\n",
+      "Epoch 375/5000\n",
+      "919/919 - 3s - loss: 2.0327 - accuracy: 0.3834 - val_loss: 3.0930 - val_accuracy: 0.0624\n",
+      "Epoch 376/5000\n",
+      "919/919 - 3s - loss: 2.0731 - accuracy: 0.3850 - val_loss: 3.0949 - val_accuracy: 0.0625\n",
+      "Epoch 377/5000\n",
+      "919/919 - 3s - loss: 2.0247 - accuracy: 0.3847 - val_loss: 3.1012 - val_accuracy: 0.0625\n",
+      "Epoch 378/5000\n",
+      "919/919 - 3s - loss: 2.0412 - accuracy: 0.3863 - val_loss: 3.0983 - val_accuracy: 0.0625\n",
+      "Epoch 379/5000\n",
+      "919/919 - 3s - loss: 2.0337 - accuracy: 0.3838 - val_loss: 3.1086 - val_accuracy: 0.0624\n",
+      "Epoch 380/5000\n",
+      "919/919 - 3s - loss: 2.0148 - accuracy: 0.3839 - val_loss: 3.1001 - val_accuracy: 0.0624\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 381/5000\n",
+      "919/919 - 3s - loss: 2.0687 - accuracy: 0.3820 - val_loss: 3.1096 - val_accuracy: 0.0625\n",
+      "Epoch 382/5000\n",
+      "919/919 - 3s - loss: 2.0238 - accuracy: 0.3871 - val_loss: 3.1078 - val_accuracy: 0.0625\n",
+      "Epoch 383/5000\n",
+      "919/919 - 3s - loss: 2.0210 - accuracy: 0.3879 - val_loss: 3.1083 - val_accuracy: 0.0625\n",
+      "Epoch 384/5000\n",
+      "919/919 - 3s - loss: 2.0234 - accuracy: 0.3852 - val_loss: 3.1083 - val_accuracy: 0.0625\n",
+      "Epoch 385/5000\n",
+      "919/919 - 3s - loss: 2.0354 - accuracy: 0.3853 - val_loss: 3.1031 - val_accuracy: 0.0624\n",
+      "Epoch 386/5000\n",
+      "919/919 - 3s - loss: 2.0002 - accuracy: 0.3878 - val_loss: 3.1014 - val_accuracy: 0.0624\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 387/5000\n",
+      "919/919 - 3s - loss: 1.9974 - accuracy: 0.3876 - val_loss: 3.1112 - val_accuracy: 0.0624\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 388/5000\n",
+      "919/919 - 3s - loss: 2.0037 - accuracy: 0.3867 - val_loss: 3.1082 - val_accuracy: 0.0624\n",
+      "Epoch 389/5000\n",
+      "919/919 - 3s - loss: 2.0056 - accuracy: 0.3857 - val_loss: 3.1132 - val_accuracy: 0.0624\n",
+      "Epoch 390/5000\n",
+      "919/919 - 3s - loss: 2.0168 - accuracy: 0.3879 - val_loss: 3.1119 - val_accuracy: 0.0624\n",
+      "Epoch 391/5000\n",
+      "919/919 - 3s - loss: 2.0025 - accuracy: 0.3858 - val_loss: 3.1293 - val_accuracy: 0.0624\n",
+      "Epoch 392/5000\n",
+      "919/919 - 3s - loss: 2.0100 - accuracy: 0.3893 - val_loss: 3.1264 - val_accuracy: 0.0624\n",
+      "Epoch 393/5000\n",
+      "919/919 - 3s - loss: 2.0066 - accuracy: 0.3869 - val_loss: 3.1197 - val_accuracy: 0.0624\n",
+      "Epoch 394/5000\n",
+      "919/919 - 3s - loss: 2.0739 - accuracy: 0.3870 - val_loss: 3.1247 - val_accuracy: 0.0624\n",
+      "Epoch 395/5000\n",
+      "919/919 - 3s - loss: 2.1263 - accuracy: 0.3878 - val_loss: 3.1325 - val_accuracy: 0.0624\n",
+      "Epoch 396/5000\n",
+      "919/919 - 3s - loss: 2.0276 - accuracy: 0.3897 - val_loss: 3.1629 - val_accuracy: 0.0624\n",
+      "Epoch 397/5000\n",
+      "919/919 - 3s - loss: 2.1484 - accuracy: 0.3869 - val_loss: 3.1559 - val_accuracy: 0.0624\n",
+      "Epoch 398/5000\n",
+      "919/919 - 3s - loss: 2.0393 - accuracy: 0.3908 - val_loss: 3.1748 - val_accuracy: 0.0624\n",
+      "Epoch 399/5000\n",
+      "919/919 - 3s - loss: 2.0312 - accuracy: 0.3873 - val_loss: 3.1884 - val_accuracy: 0.0624\n",
+      "Epoch 400/5000\n",
+      "919/919 - 3s - loss: 1.9970 - accuracy: 0.3888 - val_loss: 3.1934 - val_accuracy: 0.0624\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 401/5000\n",
+      "919/919 - 3s - loss: 2.0206 - accuracy: 0.3882 - val_loss: 3.1953 - val_accuracy: 0.0624\n",
+      "Epoch 402/5000\n",
+      "919/919 - 3s - loss: 2.0270 - accuracy: 0.3898 - val_loss: 3.1824 - val_accuracy: 0.0625\n",
+      "Epoch 403/5000\n",
+      "919/919 - 3s - loss: 1.9897 - accuracy: 0.3895 - val_loss: 3.1872 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 404/5000\n",
+      "919/919 - 3s - loss: 2.0968 - accuracy: 0.3880 - val_loss: 3.1954 - val_accuracy: 0.0625\n",
+      "Epoch 405/5000\n",
+      "919/919 - 3s - loss: 1.9946 - accuracy: 0.3895 - val_loss: 3.1980 - val_accuracy: 0.0625\n",
+      "Epoch 406/5000\n",
+      "919/919 - 3s - loss: 2.0211 - accuracy: 0.3889 - val_loss: 3.2016 - val_accuracy: 0.0625\n",
+      "Epoch 407/5000\n",
+      "919/919 - 3s - loss: 2.0032 - accuracy: 0.3884 - val_loss: 3.2106 - val_accuracy: 0.0625\n",
+      "Epoch 408/5000\n",
+      "919/919 - 3s - loss: 1.9966 - accuracy: 0.3910 - val_loss: 3.2082 - val_accuracy: 0.0625\n",
+      "Epoch 409/5000\n",
+      "919/919 - 3s - loss: 1.9990 - accuracy: 0.3914 - val_loss: 3.1996 - val_accuracy: 0.0625\n",
+      "Epoch 410/5000\n",
+      "919/919 - 3s - loss: 2.0473 - accuracy: 0.3889 - val_loss: 3.1995 - val_accuracy: 0.0625\n",
+      "Epoch 411/5000\n",
+      "919/919 - 3s - loss: 2.0586 - accuracy: 0.3867 - val_loss: 3.2037 - val_accuracy: 0.0625\n",
+      "Epoch 412/5000\n",
+      "919/919 - 3s - loss: 2.0219 - accuracy: 0.3890 - val_loss: 3.1914 - val_accuracy: 0.0625\n",
+      "Epoch 413/5000\n",
+      "919/919 - 3s - loss: 2.0046 - accuracy: 0.3911 - val_loss: 3.1666 - val_accuracy: 0.0625\n",
+      "Epoch 414/5000\n",
+      "919/919 - 3s - loss: 2.0162 - accuracy: 0.3886 - val_loss: 3.1893 - val_accuracy: 0.0626\n",
+      "Epoch 415/5000\n",
+      "919/919 - 3s - loss: 1.9845 - accuracy: 0.3948 - val_loss: 3.2102 - val_accuracy: 0.0627\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 416/5000\n",
+      "919/919 - 3s - loss: 2.0060 - accuracy: 0.3884 - val_loss: 3.1991 - val_accuracy: 0.0628\n",
+      "Epoch 417/5000\n",
+      "919/919 - 3s - loss: 1.9917 - accuracy: 0.3853 - val_loss: 3.1910 - val_accuracy: 0.0628\n",
+      "Epoch 418/5000\n",
+      "919/919 - 3s - loss: 2.0573 - accuracy: 0.3871 - val_loss: 3.1994 - val_accuracy: 0.0628\n",
+      "Epoch 419/5000\n",
+      "919/919 - 3s - loss: 1.9834 - accuracy: 0.3922 - val_loss: 3.2033 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 420/5000\n",
+      "919/919 - 3s - loss: 1.9780 - accuracy: 0.3921 - val_loss: 3.2064 - val_accuracy: 0.0628\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 421/5000\n",
+      "919/919 - 3s - loss: 1.9896 - accuracy: 0.3906 - val_loss: 3.1933 - val_accuracy: 0.0628\n",
+      "Epoch 422/5000\n",
+      "919/919 - 3s - loss: 1.9776 - accuracy: 0.3922 - val_loss: 3.2008 - val_accuracy: 0.0627\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 423/5000\n",
+      "919/919 - 3s - loss: 1.9958 - accuracy: 0.3894 - val_loss: 3.1970 - val_accuracy: 0.0629\n",
+      "Epoch 424/5000\n",
+      "919/919 - 3s - loss: 1.9926 - accuracy: 0.3894 - val_loss: 3.1926 - val_accuracy: 0.0626\n",
+      "Epoch 425/5000\n",
+      "919/919 - 3s - loss: 2.0181 - accuracy: 0.3901 - val_loss: 3.1875 - val_accuracy: 0.0626\n",
+      "Epoch 426/5000\n",
+      "919/919 - 3s - loss: 1.9863 - accuracy: 0.3939 - val_loss: 3.1727 - val_accuracy: 0.0626\n",
+      "Epoch 427/5000\n",
+      "919/919 - 3s - loss: 1.9892 - accuracy: 0.3902 - val_loss: 3.1857 - val_accuracy: 0.0626\n",
+      "Epoch 428/5000\n",
+      "919/919 - 3s - loss: 1.9715 - accuracy: 0.3926 - val_loss: 3.1794 - val_accuracy: 0.0628\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 429/5000\n",
+      "919/919 - 3s - loss: 1.9995 - accuracy: 0.3902 - val_loss: 3.1695 - val_accuracy: 0.0629\n",
+      "Epoch 430/5000\n",
+      "919/919 - 3s - loss: 2.0389 - accuracy: 0.3888 - val_loss: 3.1729 - val_accuracy: 0.0628\n",
+      "Epoch 431/5000\n",
+      "919/919 - 3s - loss: 1.9990 - accuracy: 0.3938 - val_loss: 3.1693 - val_accuracy: 0.0628\n",
+      "Epoch 432/5000\n",
+      "919/919 - 3s - loss: 1.9856 - accuracy: 0.3932 - val_loss: 3.1679 - val_accuracy: 0.0628\n",
+      "Epoch 433/5000\n",
+      "919/919 - 3s - loss: 2.0475 - accuracy: 0.3937 - val_loss: 3.1645 - val_accuracy: 0.0629\n",
+      "Epoch 434/5000\n",
+      "919/919 - 3s - loss: 1.9725 - accuracy: 0.3925 - val_loss: 3.1588 - val_accuracy: 0.0628\n",
+      "Epoch 435/5000\n",
+      "919/919 - 3s - loss: 1.9922 - accuracy: 0.3910 - val_loss: 3.1573 - val_accuracy: 0.0628\n",
+      "Epoch 436/5000\n",
+      "919/919 - 3s - loss: 1.9668 - accuracy: 0.3944 - val_loss: 3.1577 - val_accuracy: 0.0628\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 437/5000\n",
+      "919/919 - 3s - loss: 2.0559 - accuracy: 0.3897 - val_loss: 3.1543 - val_accuracy: 0.0628\n",
+      "Epoch 438/5000\n",
+      "919/919 - 3s - loss: 1.9801 - accuracy: 0.3920 - val_loss: 3.1542 - val_accuracy: 0.0630\n",
+      "Epoch 439/5000\n",
+      "919/919 - 3s - loss: 1.9768 - accuracy: 0.3945 - val_loss: 3.1419 - val_accuracy: 0.0630\n",
+      "Epoch 440/5000\n",
+      "919/919 - 3s - loss: 1.9879 - accuracy: 0.3926 - val_loss: 3.1405 - val_accuracy: 0.0630\n",
+      "Epoch 441/5000\n",
+      "919/919 - 3s - loss: 2.0469 - accuracy: 0.3913 - val_loss: 3.1414 - val_accuracy: 0.0628\n",
+      "Epoch 442/5000\n",
+      "919/919 - 3s - loss: 2.0017 - accuracy: 0.3903 - val_loss: 3.1314 - val_accuracy: 0.0628\n",
+      "Epoch 443/5000\n",
+      "919/919 - 3s - loss: 2.0921 - accuracy: 0.3890 - val_loss: 3.1404 - val_accuracy: 0.0629\n",
+      "Epoch 444/5000\n",
+      "919/919 - 3s - loss: 1.9840 - accuracy: 0.3922 - val_loss: 3.1439 - val_accuracy: 0.0630\n",
+      "Epoch 445/5000\n",
+      "919/919 - 3s - loss: 1.9857 - accuracy: 0.3905 - val_loss: 3.1474 - val_accuracy: 0.0629\n",
+      "Epoch 446/5000\n",
+      "919/919 - 3s - loss: 1.9832 - accuracy: 0.3916 - val_loss: 3.1494 - val_accuracy: 0.0630\n",
+      "Epoch 447/5000\n",
+      "919/919 - 3s - loss: 2.0111 - accuracy: 0.3913 - val_loss: 3.1463 - val_accuracy: 0.0629\n",
+      "Epoch 448/5000\n",
+      "919/919 - 3s - loss: 1.9666 - accuracy: 0.3959 - val_loss: 3.1425 - val_accuracy: 0.0630\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 449/5000\n",
+      "919/919 - 3s - loss: 1.9637 - accuracy: 0.3936 - val_loss: 3.1480 - val_accuracy: 0.0629\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 450/5000\n",
+      "919/919 - 3s - loss: 1.9646 - accuracy: 0.3931 - val_loss: 3.1349 - val_accuracy: 0.0629\n",
+      "Epoch 451/5000\n",
+      "919/919 - 3s - loss: 1.9839 - accuracy: 0.3911 - val_loss: 3.1330 - val_accuracy: 0.0630\n",
+      "Epoch 452/5000\n",
+      "919/919 - 3s - loss: 2.1276 - accuracy: 0.3912 - val_loss: 3.1482 - val_accuracy: 0.0629\n",
+      "Epoch 453/5000\n",
+      "919/919 - 3s - loss: 1.9730 - accuracy: 0.3946 - val_loss: 3.1536 - val_accuracy: 0.0631\n",
+      "Epoch 454/5000\n",
+      "919/919 - 3s - loss: 1.9704 - accuracy: 0.3933 - val_loss: 3.1556 - val_accuracy: 0.0631\n",
+      "Epoch 455/5000\n",
+      "919/919 - 3s - loss: 2.0785 - accuracy: 0.3927 - val_loss: 3.1440 - val_accuracy: 0.0631\n",
+      "Epoch 456/5000\n",
+      "919/919 - 3s - loss: 2.1728 - accuracy: 0.3926 - val_loss: 3.1446 - val_accuracy: 0.0631\n",
+      "Epoch 457/5000\n",
+      "919/919 - 3s - loss: 1.9633 - accuracy: 0.3931 - val_loss: 3.1610 - val_accuracy: 0.0632\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 458/5000\n",
+      "919/919 - 3s - loss: 2.0754 - accuracy: 0.3927 - val_loss: 3.1569 - val_accuracy: 0.0631\n",
+      "Epoch 459/5000\n",
+      "919/919 - 3s - loss: 2.0453 - accuracy: 0.3903 - val_loss: 3.1599 - val_accuracy: 0.0630\n",
+      "Epoch 460/5000\n",
+      "919/919 - 3s - loss: 1.9694 - accuracy: 0.3924 - val_loss: 3.1467 - val_accuracy: 0.0630\n",
+      "Epoch 461/5000\n",
+      "919/919 - 3s - loss: 1.9772 - accuracy: 0.3890 - val_loss: 3.1396 - val_accuracy: 0.0627\n",
+      "Epoch 462/5000\n",
+      "919/919 - 3s - loss: 1.9616 - accuracy: 0.3914 - val_loss: 3.1328 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 463/5000\n",
+      "919/919 - 3s - loss: 1.9591 - accuracy: 0.3949 - val_loss: 3.1367 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 464/5000\n",
+      "919/919 - 3s - loss: 1.9622 - accuracy: 0.3933 - val_loss: 3.1425 - val_accuracy: 0.0625\n",
+      "Epoch 465/5000\n",
+      "919/919 - 3s - loss: 1.9584 - accuracy: 0.3980 - val_loss: 3.1399 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 466/5000\n",
+      "919/919 - 3s - loss: 1.9709 - accuracy: 0.3960 - val_loss: 3.1488 - val_accuracy: 0.0625\n",
+      "Epoch 467/5000\n",
+      "919/919 - 3s - loss: 2.0532 - accuracy: 0.3991 - val_loss: 3.1455 - val_accuracy: 0.0630\n",
+      "Epoch 468/5000\n",
+      "919/919 - 3s - loss: 1.9527 - accuracy: 0.3960 - val_loss: 3.1496 - val_accuracy: 0.0631\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 469/5000\n",
+      "919/919 - 3s - loss: 1.9621 - accuracy: 0.3994 - val_loss: 3.1540 - val_accuracy: 0.0630\n",
+      "Epoch 470/5000\n",
+      "919/919 - 3s - loss: 2.1235 - accuracy: 0.3973 - val_loss: 3.1584 - val_accuracy: 0.0631\n",
+      "Epoch 471/5000\n",
+      "919/919 - 3s - loss: 1.9760 - accuracy: 0.3931 - val_loss: 3.1709 - val_accuracy: 0.0630\n",
+      "Epoch 472/5000\n",
+      "919/919 - 3s - loss: 1.9578 - accuracy: 0.3973 - val_loss: 3.1630 - val_accuracy: 0.0631\n",
+      "Epoch 473/5000\n",
+      "919/919 - 3s - loss: 1.9505 - accuracy: 0.3941 - val_loss: 3.1652 - val_accuracy: 0.0632\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 474/5000\n",
+      "919/919 - 3s - loss: 1.9748 - accuracy: 0.3970 - val_loss: 3.1630 - val_accuracy: 0.0631\n",
+      "Epoch 475/5000\n",
+      "919/919 - 3s - loss: 1.9705 - accuracy: 0.3969 - val_loss: 3.1632 - val_accuracy: 0.0630\n",
+      "Epoch 476/5000\n",
+      "919/919 - 3s - loss: 1.9516 - accuracy: 0.3976 - val_loss: 3.1561 - val_accuracy: 0.0630\n",
+      "Epoch 477/5000\n",
+      "919/919 - 3s - loss: 2.0334 - accuracy: 0.3945 - val_loss: 3.1743 - val_accuracy: 0.0631\n",
+      "Epoch 478/5000\n",
+      "919/919 - 3s - loss: 1.9534 - accuracy: 0.3972 - val_loss: 3.1718 - val_accuracy: 0.0632\n",
+      "Epoch 479/5000\n",
+      "919/919 - 3s - loss: 2.0213 - accuracy: 0.3975 - val_loss: 3.1713 - val_accuracy: 0.0631\n",
+      "Epoch 480/5000\n",
+      "919/919 - 3s - loss: 1.9421 - accuracy: 0.3991 - val_loss: 3.1686 - val_accuracy: 0.0631\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 481/5000\n",
+      "919/919 - 3s - loss: 1.9387 - accuracy: 0.3985 - val_loss: 3.1864 - val_accuracy: 0.0631\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 482/5000\n",
+      "919/919 - 3s - loss: 1.9462 - accuracy: 0.3988 - val_loss: 3.1907 - val_accuracy: 0.0631\n",
+      "Epoch 483/5000\n",
+      "919/919 - 3s - loss: 2.0774 - accuracy: 0.3954 - val_loss: 3.1920 - val_accuracy: 0.0632\n",
+      "Epoch 484/5000\n",
+      "919/919 - 3s - loss: 1.9685 - accuracy: 0.3988 - val_loss: 3.1868 - val_accuracy: 0.0632\n",
+      "Epoch 485/5000\n",
+      "919/919 - 3s - loss: 1.9521 - accuracy: 0.3970 - val_loss: 3.1881 - val_accuracy: 0.0632\n",
+      "Epoch 486/5000\n",
+      "919/919 - 3s - loss: 1.9750 - accuracy: 0.3931 - val_loss: 3.1875 - val_accuracy: 0.0631\n",
+      "Epoch 487/5000\n",
+      "919/919 - 3s - loss: 2.0201 - accuracy: 0.3941 - val_loss: 3.1857 - val_accuracy: 0.0630\n",
+      "Epoch 488/5000\n",
+      "919/919 - 3s - loss: 1.9365 - accuracy: 0.3965 - val_loss: 3.1827 - val_accuracy: 0.0631\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 489/5000\n",
+      "919/919 - 3s - loss: 1.9449 - accuracy: 0.3976 - val_loss: 3.1832 - val_accuracy: 0.0632\n",
+      "Epoch 490/5000\n",
+      "919/919 - 3s - loss: 1.9386 - accuracy: 0.3986 - val_loss: 3.1737 - val_accuracy: 0.0632\n",
+      "Epoch 491/5000\n",
+      "919/919 - 3s - loss: 2.1000 - accuracy: 0.3980 - val_loss: 3.1717 - val_accuracy: 0.0632\n",
+      "Epoch 492/5000\n",
+      "919/919 - 3s - loss: 1.9343 - accuracy: 0.3989 - val_loss: 3.1810 - val_accuracy: 0.0632\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 493/5000\n",
+      "919/919 - 3s - loss: 1.9447 - accuracy: 0.3963 - val_loss: 3.1782 - val_accuracy: 0.0632\n",
+      "Epoch 494/5000\n",
+      "919/919 - 3s - loss: 1.9467 - accuracy: 0.3996 - val_loss: 3.1815 - val_accuracy: 0.0632\n",
+      "Epoch 495/5000\n",
+      "919/919 - 3s - loss: 1.9447 - accuracy: 0.3952 - val_loss: 3.1829 - val_accuracy: 0.0632\n",
+      "Epoch 496/5000\n",
+      "919/919 - 3s - loss: 1.9188 - accuracy: 0.4015 - val_loss: 3.1715 - val_accuracy: 0.0632\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 497/5000\n",
+      "919/919 - 3s - loss: 1.9489 - accuracy: 0.3948 - val_loss: 3.1711 - val_accuracy: 0.0632\n",
+      "Epoch 498/5000\n",
+      "919/919 - 3s - loss: 1.9420 - accuracy: 0.3972 - val_loss: 3.1782 - val_accuracy: 0.0632\n",
+      "Epoch 499/5000\n",
+      "919/919 - 3s - loss: 1.9511 - accuracy: 0.3984 - val_loss: 3.1624 - val_accuracy: 0.0632\n",
+      "Epoch 500/5000\n",
+      "919/919 - 3s - loss: 1.9438 - accuracy: 0.3977 - val_loss: 3.1506 - val_accuracy: 0.0632\n",
+      "Epoch 501/5000\n",
+      "919/919 - 3s - loss: 1.9484 - accuracy: 0.3986 - val_loss: 3.1519 - val_accuracy: 0.0632\n",
+      "Epoch 502/5000\n",
+      "919/919 - 3s - loss: 1.9290 - accuracy: 0.3999 - val_loss: 3.1519 - val_accuracy: 0.0632\n",
+      "Epoch 503/5000\n",
+      "919/919 - 3s - loss: 1.9462 - accuracy: 0.3963 - val_loss: 3.1488 - val_accuracy: 0.0632\n",
+      "Epoch 504/5000\n",
+      "919/919 - 3s - loss: 1.9375 - accuracy: 0.3986 - val_loss: 3.1514 - val_accuracy: 0.0632\n",
+      "Epoch 505/5000\n",
+      "919/919 - 3s - loss: 1.9333 - accuracy: 0.3990 - val_loss: 3.1428 - val_accuracy: 0.0632\n",
+      "Epoch 506/5000\n",
+      "919/919 - 3s - loss: 1.9459 - accuracy: 0.3989 - val_loss: 3.1538 - val_accuracy: 0.0632\n",
+      "Epoch 507/5000\n",
+      "919/919 - 3s - loss: 2.0855 - accuracy: 0.3976 - val_loss: 3.1657 - val_accuracy: 0.0632\n",
+      "Epoch 508/5000\n",
+      "919/919 - 3s - loss: 1.9420 - accuracy: 0.3968 - val_loss: 3.1688 - val_accuracy: 0.0632\n",
+      "Epoch 509/5000\n",
+      "919/919 - 3s - loss: 1.9417 - accuracy: 0.3981 - val_loss: 3.1764 - val_accuracy: 0.0632\n",
+      "Epoch 510/5000\n",
+      "919/919 - 3s - loss: 1.9295 - accuracy: 0.3994 - val_loss: 3.1740 - val_accuracy: 0.0632\n",
+      "Epoch 511/5000\n",
+      "919/919 - 3s - loss: 1.9460 - accuracy: 0.4024 - val_loss: 3.1772 - val_accuracy: 0.0632\n",
+      "Epoch 512/5000\n",
+      "919/919 - 3s - loss: 1.9443 - accuracy: 0.4002 - val_loss: 3.1718 - val_accuracy: 0.0632\n",
+      "Epoch 513/5000\n",
+      "919/919 - 3s - loss: 1.9644 - accuracy: 0.3992 - val_loss: 3.1679 - val_accuracy: 0.0632\n",
+      "Epoch 514/5000\n",
+      "919/919 - 3s - loss: 1.9492 - accuracy: 0.3948 - val_loss: 3.1591 - val_accuracy: 0.0632\n",
+      "Epoch 515/5000\n",
+      "919/919 - 3s - loss: 1.9305 - accuracy: 0.3996 - val_loss: 3.1640 - val_accuracy: 0.0632\n",
+      "Epoch 516/5000\n",
+      "919/919 - 3s - loss: 1.9827 - accuracy: 0.4040 - val_loss: 3.1601 - val_accuracy: 0.0632\n",
+      "Epoch 517/5000\n",
+      "919/919 - 3s - loss: 1.9312 - accuracy: 0.4001 - val_loss: 3.1696 - val_accuracy: 0.0632\n",
+      "Epoch 518/5000\n",
+      "919/919 - 3s - loss: 1.9333 - accuracy: 0.4002 - val_loss: 3.1661 - val_accuracy: 0.0632\n",
+      "Epoch 519/5000\n",
+      "919/919 - 3s - loss: 1.9965 - accuracy: 0.3990 - val_loss: 3.1604 - val_accuracy: 0.0632\n",
+      "Epoch 520/5000\n",
+      "919/919 - 3s - loss: 1.9215 - accuracy: 0.4043 - val_loss: 3.1596 - val_accuracy: 0.0632\n",
+      "Epoch 521/5000\n",
+      "919/919 - 3s - loss: 1.9209 - accuracy: 0.4021 - val_loss: 3.1665 - val_accuracy: 0.0632\n",
+      "Epoch 522/5000\n",
+      "919/919 - 3s - loss: 1.9132 - accuracy: 0.4035 - val_loss: 3.1721 - val_accuracy: 0.0632\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 523/5000\n",
+      "919/919 - 3s - loss: 1.9171 - accuracy: 0.4024 - val_loss: 3.1800 - val_accuracy: 0.0632\n",
+      "Epoch 524/5000\n",
+      "919/919 - 3s - loss: 1.9171 - accuracy: 0.4012 - val_loss: 3.1777 - val_accuracy: 0.0632\n",
+      "Epoch 525/5000\n",
+      "919/919 - 3s - loss: 1.9196 - accuracy: 0.4018 - val_loss: 3.1828 - val_accuracy: 0.0632\n",
+      "Epoch 526/5000\n",
+      "919/919 - 3s - loss: 1.9218 - accuracy: 0.3991 - val_loss: 3.1837 - val_accuracy: 0.0632\n",
+      "Epoch 527/5000\n",
+      "919/919 - 3s - loss: 1.9283 - accuracy: 0.4007 - val_loss: 3.1884 - val_accuracy: 0.0632\n",
+      "Epoch 528/5000\n",
+      "919/919 - 3s - loss: 1.9357 - accuracy: 0.4010 - val_loss: 3.1940 - val_accuracy: 0.0632\n",
+      "Epoch 529/5000\n",
+      "919/919 - 3s - loss: 1.9343 - accuracy: 0.4018 - val_loss: 3.1914 - val_accuracy: 0.0631\n",
+      "Epoch 530/5000\n",
+      "919/919 - 3s - loss: 1.9091 - accuracy: 0.4033 - val_loss: 3.2226 - val_accuracy: 0.0631\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 531/5000\n",
+      "919/919 - 3s - loss: 1.9363 - accuracy: 0.4006 - val_loss: 3.2302 - val_accuracy: 0.0632\n",
+      "Epoch 532/5000\n",
+      "919/919 - 3s - loss: 1.9365 - accuracy: 0.4016 - val_loss: 3.2319 - val_accuracy: 0.0632\n",
+      "Epoch 533/5000\n",
+      "919/919 - 3s - loss: 1.9026 - accuracy: 0.4022 - val_loss: 3.2328 - val_accuracy: 0.0632\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 534/5000\n",
+      "919/919 - 3s - loss: 1.9155 - accuracy: 0.4018 - val_loss: 3.2192 - val_accuracy: 0.0631\n",
+      "Epoch 535/5000\n",
+      "919/919 - 3s - loss: 1.9288 - accuracy: 0.4036 - val_loss: 3.2180 - val_accuracy: 0.0630\n",
+      "Epoch 536/5000\n",
+      "919/919 - 3s - loss: 1.8901 - accuracy: 0.4067 - val_loss: 3.2114 - val_accuracy: 0.0630\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 537/5000\n",
+      "919/919 - 3s - loss: 1.9325 - accuracy: 0.4003 - val_loss: 3.2174 - val_accuracy: 0.0630\n",
+      "Epoch 538/5000\n",
+      "919/919 - 3s - loss: 2.0157 - accuracy: 0.4022 - val_loss: 3.2136 - val_accuracy: 0.0630\n",
+      "Epoch 539/5000\n",
+      "919/919 - 3s - loss: 1.9232 - accuracy: 0.4032 - val_loss: 3.2043 - val_accuracy: 0.0631\n",
+      "Epoch 540/5000\n",
+      "919/919 - 3s - loss: 1.9239 - accuracy: 0.4046 - val_loss: 3.2004 - val_accuracy: 0.0631\n",
+      "Epoch 541/5000\n",
+      "919/919 - 3s - loss: 1.9152 - accuracy: 0.4003 - val_loss: 3.2137 - val_accuracy: 0.0632\n",
+      "Epoch 542/5000\n",
+      "919/919 - 3s - loss: 1.9231 - accuracy: 0.4019 - val_loss: 3.2065 - val_accuracy: 0.0631\n",
+      "Epoch 543/5000\n",
+      "919/919 - 3s - loss: 1.9514 - accuracy: 0.4008 - val_loss: 3.2048 - val_accuracy: 0.0630\n",
+      "Epoch 544/5000\n",
+      "919/919 - 3s - loss: 1.9489 - accuracy: 0.4022 - val_loss: 3.2113 - val_accuracy: 0.0631\n",
+      "Epoch 545/5000\n",
+      "919/919 - 3s - loss: 1.9173 - accuracy: 0.4027 - val_loss: 3.2064 - val_accuracy: 0.0632\n",
+      "Epoch 546/5000\n",
+      "919/919 - 3s - loss: 1.9008 - accuracy: 0.4044 - val_loss: 3.2003 - val_accuracy: 0.0631\n",
+      "Epoch 547/5000\n",
+      "919/919 - 3s - loss: 1.9213 - accuracy: 0.4000 - val_loss: 3.2080 - val_accuracy: 0.0632\n",
+      "Epoch 548/5000\n",
+      "919/919 - 3s - loss: 1.8886 - accuracy: 0.4078 - val_loss: 3.2110 - val_accuracy: 0.0632\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 549/5000\n",
+      "919/919 - 3s - loss: 1.9188 - accuracy: 0.4075 - val_loss: 3.2099 - val_accuracy: 0.0631\n",
+      "Epoch 550/5000\n",
+      "919/919 - 3s - loss: 1.9151 - accuracy: 0.4041 - val_loss: 3.2018 - val_accuracy: 0.0632\n",
+      "Epoch 551/5000\n",
+      "919/919 - 3s - loss: 1.8934 - accuracy: 0.4059 - val_loss: 3.2023 - val_accuracy: 0.0631\n",
+      "Epoch 552/5000\n",
+      "919/919 - 3s - loss: 1.9222 - accuracy: 0.4031 - val_loss: 3.1947 - val_accuracy: 0.0632\n",
+      "Epoch 553/5000\n",
+      "919/919 - 3s - loss: 1.9104 - accuracy: 0.4025 - val_loss: 3.1929 - val_accuracy: 0.0633\n",
+      "Epoch 554/5000\n",
+      "919/919 - 3s - loss: 1.9237 - accuracy: 0.4018 - val_loss: 3.1999 - val_accuracy: 0.0632\n",
+      "Epoch 555/5000\n",
+      "919/919 - 3s - loss: 1.9103 - accuracy: 0.4024 - val_loss: 3.1995 - val_accuracy: 0.0632\n",
+      "Epoch 556/5000\n",
+      "919/919 - 3s - loss: 1.9045 - accuracy: 0.4050 - val_loss: 3.2021 - val_accuracy: 0.0632\n",
+      "Epoch 557/5000\n",
+      "919/919 - 3s - loss: 1.9149 - accuracy: 0.4005 - val_loss: 3.1969 - val_accuracy: 0.0632\n",
+      "Epoch 558/5000\n",
+      "919/919 - 3s - loss: 1.9124 - accuracy: 0.4040 - val_loss: 3.1917 - val_accuracy: 0.0632\n",
+      "Epoch 559/5000\n",
+      "919/919 - 3s - loss: 1.9064 - accuracy: 0.4039 - val_loss: 3.1973 - val_accuracy: 0.0632\n",
+      "Epoch 560/5000\n",
+      "919/919 - 3s - loss: 1.9134 - accuracy: 0.4063 - val_loss: 3.1912 - val_accuracy: 0.0632\n",
+      "Epoch 561/5000\n",
+      "919/919 - 3s - loss: 1.9039 - accuracy: 0.4045 - val_loss: 3.2020 - val_accuracy: 0.0632\n",
+      "Epoch 562/5000\n",
+      "919/919 - 3s - loss: 1.9009 - accuracy: 0.4016 - val_loss: 3.1934 - val_accuracy: 0.0632\n",
+      "Epoch 563/5000\n",
+      "919/919 - 3s - loss: 2.0129 - accuracy: 0.4041 - val_loss: 3.1834 - val_accuracy: 0.0632\n",
+      "Epoch 564/5000\n",
+      "919/919 - 3s - loss: 1.8975 - accuracy: 0.4050 - val_loss: 3.1876 - val_accuracy: 0.0632\n",
+      "Epoch 565/5000\n",
+      "919/919 - 3s - loss: 1.9124 - accuracy: 0.4058 - val_loss: 3.2047 - val_accuracy: 0.0632\n",
+      "Epoch 566/5000\n",
+      "919/919 - 3s - loss: 1.8941 - accuracy: 0.4039 - val_loss: 3.2067 - val_accuracy: 0.0632\n",
+      "Epoch 567/5000\n",
+      "919/919 - 3s - loss: 1.9140 - accuracy: 0.4078 - val_loss: 3.2154 - val_accuracy: 0.0632\n",
+      "Epoch 568/5000\n",
+      "919/919 - 3s - loss: 1.8995 - accuracy: 0.4040 - val_loss: 3.2262 - val_accuracy: 0.0632\n",
+      "Epoch 569/5000\n",
+      "919/919 - 3s - loss: 1.8981 - accuracy: 0.4041 - val_loss: 3.2070 - val_accuracy: 0.0632\n",
+      "Epoch 570/5000\n",
+      "919/919 - 3s - loss: 1.9029 - accuracy: 0.4046 - val_loss: 3.1939 - val_accuracy: 0.0632\n",
+      "Epoch 571/5000\n",
+      "919/919 - 3s - loss: 1.8815 - accuracy: 0.4097 - val_loss: 3.2092 - val_accuracy: 0.0632\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 572/5000\n",
+      "919/919 - 3s - loss: 1.9058 - accuracy: 0.4066 - val_loss: 3.2147 - val_accuracy: 0.0632\n",
+      "Epoch 573/5000\n",
+      "919/919 - 3s - loss: 1.9125 - accuracy: 0.4059 - val_loss: 3.2145 - val_accuracy: 0.0632\n",
+      "Epoch 574/5000\n",
+      "919/919 - 3s - loss: 1.9063 - accuracy: 0.4060 - val_loss: 3.2136 - val_accuracy: 0.0633\n",
+      "Epoch 575/5000\n",
+      "919/919 - 3s - loss: 1.8833 - accuracy: 0.4052 - val_loss: 3.2195 - val_accuracy: 0.0632\n",
+      "Epoch 576/5000\n",
+      "919/919 - 3s - loss: 1.9715 - accuracy: 0.4067 - val_loss: 3.1977 - val_accuracy: 0.0632\n",
+      "Epoch 577/5000\n",
+      "919/919 - 3s - loss: 1.9082 - accuracy: 0.4074 - val_loss: 3.1945 - val_accuracy: 0.0632\n",
+      "Epoch 578/5000\n",
+      "919/919 - 3s - loss: 1.8893 - accuracy: 0.4073 - val_loss: 3.1948 - val_accuracy: 0.0632\n",
+      "Epoch 579/5000\n",
+      "919/919 - 3s - loss: 1.8968 - accuracy: 0.4068 - val_loss: 3.1937 - val_accuracy: 0.0632\n",
+      "Epoch 580/5000\n",
+      "919/919 - 3s - loss: 1.8888 - accuracy: 0.4069 - val_loss: 3.1980 - val_accuracy: 0.0632\n",
+      "Epoch 581/5000\n",
+      "919/919 - 3s - loss: 1.9023 - accuracy: 0.4054 - val_loss: 3.1953 - val_accuracy: 0.0632\n",
+      "Epoch 582/5000\n",
+      "919/919 - 3s - loss: 1.8903 - accuracy: 0.4073 - val_loss: 3.2094 - val_accuracy: 0.0632\n",
+      "Epoch 583/5000\n",
+      "919/919 - 3s - loss: 1.8854 - accuracy: 0.4074 - val_loss: 3.2158 - val_accuracy: 0.0632\n",
+      "Epoch 584/5000\n",
+      "919/919 - 3s - loss: 1.9041 - accuracy: 0.4056 - val_loss: 3.2145 - val_accuracy: 0.0633\n",
+      "Epoch 585/5000\n",
+      "919/919 - 3s - loss: 1.9703 - accuracy: 0.4041 - val_loss: 3.2033 - val_accuracy: 0.0633\n",
+      "Epoch 586/5000\n",
+      "919/919 - 3s - loss: 1.8806 - accuracy: 0.4059 - val_loss: 3.2054 - val_accuracy: 0.0633\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 587/5000\n",
+      "919/919 - 3s - loss: 2.0911 - accuracy: 0.4063 - val_loss: 3.1959 - val_accuracy: 0.0632\n",
+      "Epoch 588/5000\n",
+      "919/919 - 3s - loss: 1.9102 - accuracy: 0.4069 - val_loss: 3.1880 - val_accuracy: 0.0632\n",
+      "Epoch 589/5000\n",
+      "919/919 - 3s - loss: 1.8741 - accuracy: 0.4080 - val_loss: 3.1915 - val_accuracy: 0.0632\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 590/5000\n",
+      "919/919 - 3s - loss: 1.8886 - accuracy: 0.4069 - val_loss: 3.2178 - val_accuracy: 0.0632\n",
+      "Epoch 591/5000\n",
+      "919/919 - 3s - loss: 1.8856 - accuracy: 0.4082 - val_loss: 3.2280 - val_accuracy: 0.0632\n",
+      "Epoch 592/5000\n",
+      "919/919 - 3s - loss: 1.9022 - accuracy: 0.4025 - val_loss: 3.2184 - val_accuracy: 0.0632\n",
+      "Epoch 593/5000\n",
+      "919/919 - 3s - loss: 1.8828 - accuracy: 0.4020 - val_loss: 3.2210 - val_accuracy: 0.0632\n",
+      "Epoch 594/5000\n",
+      "919/919 - 3s - loss: 1.8952 - accuracy: 0.4083 - val_loss: 3.2136 - val_accuracy: 0.0632\n",
+      "Epoch 595/5000\n",
+      "919/919 - 3s - loss: 1.8940 - accuracy: 0.4090 - val_loss: 3.2060 - val_accuracy: 0.0632\n",
+      "Epoch 596/5000\n",
+      "919/919 - 3s - loss: 1.8846 - accuracy: 0.4099 - val_loss: 3.2040 - val_accuracy: 0.0632\n",
+      "Epoch 597/5000\n",
+      "919/919 - 3s - loss: 1.9151 - accuracy: 0.4103 - val_loss: 3.2007 - val_accuracy: 0.0632\n",
+      "Epoch 598/5000\n",
+      "919/919 - 3s - loss: 1.8745 - accuracy: 0.4069 - val_loss: 3.1887 - val_accuracy: 0.0631\n",
+      "Epoch 599/5000\n",
+      "919/919 - 3s - loss: 1.9091 - accuracy: 0.4046 - val_loss: 3.1910 - val_accuracy: 0.0631\n",
+      "Epoch 600/5000\n",
+      "919/919 - 3s - loss: 1.8885 - accuracy: 0.4050 - val_loss: 3.1860 - val_accuracy: 0.0632\n",
+      "Epoch 601/5000\n",
+      "919/919 - 3s - loss: 1.9107 - accuracy: 0.4056 - val_loss: 3.1815 - val_accuracy: 0.0631\n",
+      "Epoch 602/5000\n",
+      "919/919 - 3s - loss: 1.8833 - accuracy: 0.4072 - val_loss: 3.1822 - val_accuracy: 0.0631\n",
+      "Epoch 603/5000\n",
+      "919/919 - 3s - loss: 1.8853 - accuracy: 0.4078 - val_loss: 3.1883 - val_accuracy: 0.0631\n",
+      "Epoch 604/5000\n",
+      "919/919 - 3s - loss: 1.8774 - accuracy: 0.4118 - val_loss: 3.1885 - val_accuracy: 0.0631\n",
+      "Epoch 605/5000\n",
+      "919/919 - 3s - loss: 1.8613 - accuracy: 0.4118 - val_loss: 3.1883 - val_accuracy: 0.0632\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 606/5000\n",
+      "919/919 - 3s - loss: 1.8653 - accuracy: 0.4109 - val_loss: 3.1887 - val_accuracy: 0.0632\n",
+      "Epoch 607/5000\n",
+      "919/919 - 3s - loss: 1.8776 - accuracy: 0.4071 - val_loss: 3.1883 - val_accuracy: 0.0632\n",
+      "Epoch 608/5000\n",
+      "919/919 - 3s - loss: 1.8744 - accuracy: 0.4041 - val_loss: 3.1853 - val_accuracy: 0.0632\n",
+      "Epoch 609/5000\n",
+      "919/919 - 3s - loss: 1.9586 - accuracy: 0.4109 - val_loss: 3.1865 - val_accuracy: 0.0632\n",
+      "Epoch 610/5000\n",
+      "919/919 - 3s - loss: 1.8836 - accuracy: 0.4067 - val_loss: 3.1876 - val_accuracy: 0.0632\n",
+      "Epoch 611/5000\n",
+      "919/919 - 3s - loss: 2.0705 - accuracy: 0.4068 - val_loss: 3.1859 - val_accuracy: 0.0632\n",
+      "Epoch 612/5000\n",
+      "919/919 - 3s - loss: 1.8861 - accuracy: 0.4054 - val_loss: 3.1916 - val_accuracy: 0.0632\n",
+      "Epoch 613/5000\n",
+      "919/919 - 3s - loss: 1.9305 - accuracy: 0.4073 - val_loss: 3.2028 - val_accuracy: 0.0632\n",
+      "Epoch 614/5000\n",
+      "919/919 - 3s - loss: 1.8578 - accuracy: 0.4101 - val_loss: 3.1971 - val_accuracy: 0.0632\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 615/5000\n",
+      "919/919 - 3s - loss: 2.0401 - accuracy: 0.4069 - val_loss: 3.1994 - val_accuracy: 0.0632\n",
+      "Epoch 616/5000\n",
+      "919/919 - 3s - loss: 1.9159 - accuracy: 0.4088 - val_loss: 3.2142 - val_accuracy: 0.0632\n",
+      "Epoch 617/5000\n",
+      "919/919 - 3s - loss: 1.8844 - accuracy: 0.4070 - val_loss: 3.2263 - val_accuracy: 0.0632\n",
+      "Epoch 618/5000\n",
+      "919/919 - 3s - loss: 2.0889 - accuracy: 0.4114 - val_loss: 3.2453 - val_accuracy: 0.0632\n",
+      "Epoch 619/5000\n",
+      "919/919 - 3s - loss: 1.8791 - accuracy: 0.4088 - val_loss: 3.2522 - val_accuracy: 0.0632\n",
+      "Epoch 620/5000\n",
+      "919/919 - 3s - loss: 1.8867 - accuracy: 0.4038 - val_loss: 3.2373 - val_accuracy: 0.0631\n",
+      "Epoch 621/5000\n",
+      "919/919 - 3s - loss: 1.8730 - accuracy: 0.4085 - val_loss: 3.2354 - val_accuracy: 0.0632\n",
+      "Epoch 622/5000\n",
+      "919/919 - 3s - loss: 1.8676 - accuracy: 0.4047 - val_loss: 3.2261 - val_accuracy: 0.0631\n",
+      "Epoch 623/5000\n",
+      "919/919 - 3s - loss: 1.8958 - accuracy: 0.4087 - val_loss: 3.2247 - val_accuracy: 0.0631\n",
+      "Epoch 624/5000\n",
+      "919/919 - 3s - loss: 1.8628 - accuracy: 0.4121 - val_loss: 3.2123 - val_accuracy: 0.0632\n",
+      "Epoch 625/5000\n",
+      "919/919 - 3s - loss: 1.8811 - accuracy: 0.4101 - val_loss: 3.2156 - val_accuracy: 0.0632\n",
+      "Epoch 626/5000\n",
+      "919/919 - 3s - loss: 1.8883 - accuracy: 0.4078 - val_loss: 3.2092 - val_accuracy: 0.0631\n",
+      "Epoch 627/5000\n",
+      "919/919 - 3s - loss: 1.8610 - accuracy: 0.4080 - val_loss: 3.2266 - val_accuracy: 0.0632\n",
+      "Epoch 628/5000\n",
+      "919/919 - 3s - loss: 1.8664 - accuracy: 0.4076 - val_loss: 3.2287 - val_accuracy: 0.0631\n",
+      "Epoch 629/5000\n",
+      "919/919 - 3s - loss: 1.8726 - accuracy: 0.4107 - val_loss: 3.2158 - val_accuracy: 0.0631\n",
+      "Epoch 630/5000\n",
+      "919/919 - 3s - loss: 1.8792 - accuracy: 0.4070 - val_loss: 3.2124 - val_accuracy: 0.0631\n",
+      "Epoch 631/5000\n",
+      "919/919 - 3s - loss: 1.8649 - accuracy: 0.4071 - val_loss: 3.2145 - val_accuracy: 0.0631\n",
+      "Epoch 632/5000\n",
+      "919/919 - 3s - loss: 1.8854 - accuracy: 0.4065 - val_loss: 3.2220 - val_accuracy: 0.0631\n",
+      "Epoch 633/5000\n",
+      "919/919 - 3s - loss: 1.8744 - accuracy: 0.4135 - val_loss: 3.2286 - val_accuracy: 0.0631\n",
+      "Epoch 634/5000\n",
+      "919/919 - 3s - loss: 1.8719 - accuracy: 0.4117 - val_loss: 3.2283 - val_accuracy: 0.0632\n",
+      "Epoch 635/5000\n",
+      "919/919 - 3s - loss: 1.8751 - accuracy: 0.4131 - val_loss: 3.2314 - val_accuracy: 0.0631\n",
+      "Epoch 636/5000\n",
+      "919/919 - 3s - loss: 1.8631 - accuracy: 0.4093 - val_loss: 3.2232 - val_accuracy: 0.0633\n",
+      "Epoch 637/5000\n",
+      "919/919 - 3s - loss: 1.8830 - accuracy: 0.4110 - val_loss: 3.2266 - val_accuracy: 0.0632\n",
+      "Epoch 638/5000\n",
+      "919/919 - 3s - loss: 1.8868 - accuracy: 0.4099 - val_loss: 3.2247 - val_accuracy: 0.0632\n",
+      "Epoch 639/5000\n",
+      "919/919 - 3s - loss: 1.9397 - accuracy: 0.4110 - val_loss: 3.2237 - val_accuracy: 0.0632\n",
+      "Epoch 640/5000\n",
+      "919/919 - 3s - loss: 1.8772 - accuracy: 0.4095 - val_loss: 3.2339 - val_accuracy: 0.0632\n",
+      "Epoch 641/5000\n",
+      "919/919 - 3s - loss: 2.1821 - accuracy: 0.4090 - val_loss: 3.2277 - val_accuracy: 0.0632\n",
+      "Epoch 642/5000\n",
+      "919/919 - 3s - loss: 1.8708 - accuracy: 0.4088 - val_loss: 3.2242 - val_accuracy: 0.0632\n",
+      "Epoch 643/5000\n",
+      "919/919 - 3s - loss: 1.8635 - accuracy: 0.4093 - val_loss: 3.2184 - val_accuracy: 0.0633\n",
+      "Epoch 644/5000\n",
+      "919/919 - 3s - loss: 1.8527 - accuracy: 0.4135 - val_loss: 3.2138 - val_accuracy: 0.0633\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 645/5000\n",
+      "919/919 - 3s - loss: 1.8653 - accuracy: 0.4114 - val_loss: 3.2208 - val_accuracy: 0.0633\n",
+      "Epoch 646/5000\n",
+      "919/919 - 3s - loss: 1.8768 - accuracy: 0.4133 - val_loss: 3.2131 - val_accuracy: 0.0632\n",
+      "Epoch 647/5000\n",
+      "919/919 - 3s - loss: 1.8511 - accuracy: 0.4135 - val_loss: 3.2134 - val_accuracy: 0.0632\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 648/5000\n",
+      "919/919 - 3s - loss: 1.8635 - accuracy: 0.4093 - val_loss: 3.2066 - val_accuracy: 0.0633\n",
+      "Epoch 649/5000\n",
+      "919/919 - 3s - loss: 1.8704 - accuracy: 0.4093 - val_loss: 3.1914 - val_accuracy: 0.0632\n",
+      "Epoch 650/5000\n",
+      "919/919 - 3s - loss: 1.8565 - accuracy: 0.4111 - val_loss: 3.2023 - val_accuracy: 0.0633\n",
+      "Epoch 651/5000\n",
+      "919/919 - 3s - loss: 1.8607 - accuracy: 0.4110 - val_loss: 3.2042 - val_accuracy: 0.0633\n",
+      "Epoch 652/5000\n",
+      "919/919 - 3s - loss: 2.0295 - accuracy: 0.4111 - val_loss: 3.2137 - val_accuracy: 0.0633\n",
+      "Epoch 653/5000\n",
+      "919/919 - 3s - loss: 1.9057 - accuracy: 0.4098 - val_loss: 3.2270 - val_accuracy: 0.0632\n",
+      "Epoch 654/5000\n",
+      "919/919 - 3s - loss: 1.8555 - accuracy: 0.4152 - val_loss: 3.2374 - val_accuracy: 0.0633\n",
+      "Epoch 655/5000\n",
+      "919/919 - 3s - loss: 1.8682 - accuracy: 0.4144 - val_loss: 3.2494 - val_accuracy: 0.0632\n",
+      "Epoch 656/5000\n",
+      "919/919 - 3s - loss: 1.8926 - accuracy: 0.4114 - val_loss: 3.2610 - val_accuracy: 0.0633\n",
+      "Epoch 657/5000\n",
+      "919/919 - 3s - loss: 1.8355 - accuracy: 0.4132 - val_loss: 3.2558 - val_accuracy: 0.0634\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 658/5000\n",
+      "919/919 - 3s - loss: 1.8459 - accuracy: 0.4139 - val_loss: 3.2522 - val_accuracy: 0.0632\n",
+      "Epoch 659/5000\n",
+      "919/919 - 3s - loss: 1.8606 - accuracy: 0.4118 - val_loss: 3.2431 - val_accuracy: 0.0633\n",
+      "Epoch 660/5000\n",
+      "919/919 - 3s - loss: 1.8530 - accuracy: 0.4095 - val_loss: 3.2388 - val_accuracy: 0.0634\n",
+      "Epoch 661/5000\n",
+      "919/919 - 3s - loss: 1.8590 - accuracy: 0.4105 - val_loss: 3.2315 - val_accuracy: 0.0634\n",
+      "Epoch 662/5000\n",
+      "919/919 - 3s - loss: 1.8478 - accuracy: 0.4119 - val_loss: 3.2346 - val_accuracy: 0.0634\n",
+      "Epoch 663/5000\n",
+      "919/919 - 3s - loss: 1.9047 - accuracy: 0.4133 - val_loss: 3.2235 - val_accuracy: 0.0634\n",
+      "Epoch 664/5000\n",
+      "919/919 - 3s - loss: 1.8550 - accuracy: 0.4135 - val_loss: 3.2257 - val_accuracy: 0.0633\n",
+      "Epoch 665/5000\n",
+      "919/919 - 3s - loss: 1.8547 - accuracy: 0.4126 - val_loss: 3.2298 - val_accuracy: 0.0633\n",
+      "Epoch 666/5000\n",
+      "919/919 - 3s - loss: 2.0666 - accuracy: 0.4137 - val_loss: 3.2293 - val_accuracy: 0.0633\n",
+      "Epoch 667/5000\n",
+      "919/919 - 3s - loss: 1.8507 - accuracy: 0.4148 - val_loss: 3.2242 - val_accuracy: 0.0632\n",
+      "Epoch 668/5000\n",
+      "919/919 - 3s - loss: 1.8602 - accuracy: 0.4124 - val_loss: 3.2284 - val_accuracy: 0.0632\n",
+      "Epoch 669/5000\n",
+      "919/919 - 3s - loss: 1.8478 - accuracy: 0.4122 - val_loss: 3.2197 - val_accuracy: 0.0632\n",
+      "Epoch 670/5000\n",
+      "919/919 - 3s - loss: 1.8786 - accuracy: 0.4159 - val_loss: 3.2279 - val_accuracy: 0.0633\n",
+      "Epoch 671/5000\n",
+      "919/919 - 3s - loss: 1.8488 - accuracy: 0.4124 - val_loss: 3.2244 - val_accuracy: 0.0633\n",
+      "Epoch 672/5000\n",
+      "919/919 - 3s - loss: 1.8372 - accuracy: 0.4159 - val_loss: 3.2292 - val_accuracy: 0.0633\n",
+      "Epoch 673/5000\n",
+      "919/919 - 3s - loss: 1.8480 - accuracy: 0.4093 - val_loss: 3.2472 - val_accuracy: 0.0633\n",
+      "Epoch 674/5000\n",
+      "919/919 - 3s - loss: 1.8614 - accuracy: 0.4117 - val_loss: 3.2446 - val_accuracy: 0.0633\n",
+      "Epoch 675/5000\n",
+      "919/919 - 3s - loss: 1.8604 - accuracy: 0.4110 - val_loss: 3.2453 - val_accuracy: 0.0633\n",
+      "Epoch 676/5000\n",
+      "919/919 - 3s - loss: 1.8502 - accuracy: 0.4169 - val_loss: 3.2316 - val_accuracy: 0.0634\n",
+      "Epoch 677/5000\n",
+      "919/919 - 3s - loss: 1.8446 - accuracy: 0.4161 - val_loss: 3.2183 - val_accuracy: 0.0634\n",
+      "Epoch 678/5000\n",
+      "919/919 - 3s - loss: 1.8581 - accuracy: 0.4154 - val_loss: 3.2363 - val_accuracy: 0.0633\n",
+      "Epoch 679/5000\n",
+      "919/919 - 3s - loss: 1.8411 - accuracy: 0.4170 - val_loss: 3.2337 - val_accuracy: 0.0633\n",
+      "Epoch 680/5000\n",
+      "919/919 - 3s - loss: 1.8493 - accuracy: 0.4137 - val_loss: 3.2279 - val_accuracy: 0.0633\n",
+      "Epoch 681/5000\n",
+      "919/919 - 3s - loss: 1.8664 - accuracy: 0.4132 - val_loss: 3.2166 - val_accuracy: 0.0632\n",
+      "Epoch 682/5000\n",
+      "919/919 - 3s - loss: 1.8448 - accuracy: 0.4167 - val_loss: 3.2183 - val_accuracy: 0.0633\n",
+      "Epoch 683/5000\n",
+      "919/919 - 3s - loss: 1.8414 - accuracy: 0.4146 - val_loss: 3.2103 - val_accuracy: 0.0632\n",
+      "Epoch 684/5000\n",
+      "919/919 - 3s - loss: 1.8422 - accuracy: 0.4154 - val_loss: 3.2101 - val_accuracy: 0.0632\n",
+      "Epoch 685/5000\n",
+      "919/919 - 3s - loss: 1.8409 - accuracy: 0.4188 - val_loss: 3.2067 - val_accuracy: 0.0632\n",
+      "Epoch 686/5000\n",
+      "919/919 - 3s - loss: 2.0113 - accuracy: 0.4156 - val_loss: 3.2018 - val_accuracy: 0.0632\n",
+      "Epoch 687/5000\n",
+      "919/919 - 3s - loss: 1.8436 - accuracy: 0.4159 - val_loss: 3.1968 - val_accuracy: 0.0633\n",
+      "Epoch 688/5000\n",
+      "919/919 - 3s - loss: 1.8395 - accuracy: 0.4138 - val_loss: 3.2148 - val_accuracy: 0.0633\n",
+      "Epoch 689/5000\n",
+      "919/919 - 3s - loss: 1.8406 - accuracy: 0.4114 - val_loss: 3.2044 - val_accuracy: 0.0633\n",
+      "Epoch 690/5000\n",
+      "919/919 - 3s - loss: 1.8515 - accuracy: 0.4106 - val_loss: 3.2076 - val_accuracy: 0.0633\n",
+      "Epoch 691/5000\n",
+      "919/919 - 3s - loss: 1.8424 - accuracy: 0.4188 - val_loss: 3.2118 - val_accuracy: 0.0633\n",
+      "Epoch 692/5000\n",
+      "919/919 - 3s - loss: 1.8696 - accuracy: 0.4130 - val_loss: 3.2146 - val_accuracy: 0.0633\n",
+      "Epoch 693/5000\n",
+      "919/919 - 3s - loss: 1.8389 - accuracy: 0.4135 - val_loss: 3.2111 - val_accuracy: 0.0633\n",
+      "Epoch 694/5000\n",
+      "919/919 - 3s - loss: 1.9319 - accuracy: 0.4172 - val_loss: 3.2053 - val_accuracy: 0.0633\n",
+      "Epoch 695/5000\n",
+      "919/919 - 3s - loss: 1.8480 - accuracy: 0.4128 - val_loss: 3.2043 - val_accuracy: 0.0633\n",
+      "Epoch 696/5000\n",
+      "919/919 - 3s - loss: 1.8547 - accuracy: 0.4167 - val_loss: 3.2160 - val_accuracy: 0.0633\n",
+      "Epoch 697/5000\n",
+      "919/919 - 3s - loss: 1.8289 - accuracy: 0.4169 - val_loss: 3.2160 - val_accuracy: 0.0633\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 698/5000\n",
+      "919/919 - 3s - loss: 1.8501 - accuracy: 0.4141 - val_loss: 3.2100 - val_accuracy: 0.0633\n",
+      "Epoch 699/5000\n",
+      "919/919 - 3s - loss: 1.8419 - accuracy: 0.4178 - val_loss: 3.2086 - val_accuracy: 0.0633\n",
+      "Epoch 700/5000\n",
+      "919/919 - 3s - loss: 1.8490 - accuracy: 0.4176 - val_loss: 3.2048 - val_accuracy: 0.0633\n",
+      "Epoch 701/5000\n",
+      "919/919 - 3s - loss: 1.8692 - accuracy: 0.4144 - val_loss: 3.1959 - val_accuracy: 0.0633\n",
+      "Epoch 702/5000\n",
+      "919/919 - 3s - loss: 1.8396 - accuracy: 0.4128 - val_loss: 3.2091 - val_accuracy: 0.0633\n",
+      "Epoch 703/5000\n",
+      "919/919 - 3s - loss: 1.8334 - accuracy: 0.4127 - val_loss: 3.2250 - val_accuracy: 0.0633\n",
+      "Epoch 704/5000\n",
+      "919/919 - 3s - loss: 1.9103 - accuracy: 0.4150 - val_loss: 3.2249 - val_accuracy: 0.0633\n",
+      "Epoch 705/5000\n",
+      "919/919 - 3s - loss: 1.8237 - accuracy: 0.4161 - val_loss: 3.2201 - val_accuracy: 0.0633\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 706/5000\n",
+      "919/919 - 3s - loss: 1.8774 - accuracy: 0.4178 - val_loss: 3.2211 - val_accuracy: 0.0633\n",
+      "Epoch 707/5000\n",
+      "919/919 - 3s - loss: 1.8207 - accuracy: 0.4184 - val_loss: 3.2195 - val_accuracy: 0.0633\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 708/5000\n",
+      "919/919 - 3s - loss: 1.8383 - accuracy: 0.4180 - val_loss: 3.2246 - val_accuracy: 0.0633\n",
+      "Epoch 709/5000\n",
+      "919/919 - 3s - loss: 1.8199 - accuracy: 0.4154 - val_loss: 3.2367 - val_accuracy: 0.0633\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 710/5000\n",
+      "919/919 - 3s - loss: 1.8259 - accuracy: 0.4198 - val_loss: 3.2483 - val_accuracy: 0.0633\n",
+      "Epoch 711/5000\n",
+      "919/919 - 3s - loss: 1.8403 - accuracy: 0.4147 - val_loss: 3.2526 - val_accuracy: 0.0633\n",
+      "Epoch 712/5000\n",
+      "919/919 - 3s - loss: 1.8412 - accuracy: 0.4173 - val_loss: 3.2391 - val_accuracy: 0.0633\n",
+      "Epoch 713/5000\n",
+      "919/919 - 3s - loss: 1.8415 - accuracy: 0.4154 - val_loss: 3.2306 - val_accuracy: 0.0633\n",
+      "Epoch 714/5000\n",
+      "919/919 - 3s - loss: 1.8359 - accuracy: 0.4173 - val_loss: 3.2327 - val_accuracy: 0.0632\n",
+      "Epoch 715/5000\n",
+      "919/919 - 3s - loss: 1.8343 - accuracy: 0.4152 - val_loss: 3.2463 - val_accuracy: 0.0632\n",
+      "Epoch 716/5000\n",
+      "919/919 - 3s - loss: 1.8358 - accuracy: 0.4171 - val_loss: 3.2547 - val_accuracy: 0.0632\n",
+      "Epoch 717/5000\n",
+      "919/919 - 3s - loss: 1.8408 - accuracy: 0.4211 - val_loss: 3.2573 - val_accuracy: 0.0632\n",
+      "Epoch 718/5000\n",
+      "919/919 - 3s - loss: 1.8265 - accuracy: 0.4203 - val_loss: 3.2639 - val_accuracy: 0.0632\n",
+      "Epoch 719/5000\n",
+      "919/919 - 3s - loss: 1.8342 - accuracy: 0.4182 - val_loss: 3.2743 - val_accuracy: 0.0632\n",
+      "Epoch 720/5000\n",
+      "919/919 - 3s - loss: 1.8186 - accuracy: 0.4184 - val_loss: 3.2619 - val_accuracy: 0.0632\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 721/5000\n",
+      "919/919 - 3s - loss: 1.8362 - accuracy: 0.4197 - val_loss: 3.2738 - val_accuracy: 0.0632\n",
+      "Epoch 722/5000\n",
+      "919/919 - 3s - loss: 1.8375 - accuracy: 0.4146 - val_loss: 3.2593 - val_accuracy: 0.0632\n",
+      "Epoch 723/5000\n",
+      "919/919 - 3s - loss: 1.8153 - accuracy: 0.4190 - val_loss: 3.2596 - val_accuracy: 0.0632\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 724/5000\n",
+      "919/919 - 3s - loss: 1.8270 - accuracy: 0.4168 - val_loss: 3.2488 - val_accuracy: 0.0630\n",
+      "Epoch 725/5000\n",
+      "919/919 - 3s - loss: 1.8259 - accuracy: 0.4194 - val_loss: 3.2636 - val_accuracy: 0.0629\n",
+      "Epoch 726/5000\n",
+      "919/919 - 3s - loss: 1.8133 - accuracy: 0.4208 - val_loss: 3.2622 - val_accuracy: 0.0629\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 727/5000\n",
+      "919/919 - 3s - loss: 1.8232 - accuracy: 0.4127 - val_loss: 3.2645 - val_accuracy: 0.0629\n",
+      "Epoch 728/5000\n",
+      "919/919 - 3s - loss: 1.9214 - accuracy: 0.4206 - val_loss: 3.2435 - val_accuracy: 0.0628\n",
+      "Epoch 729/5000\n",
+      "919/919 - 3s - loss: 1.8170 - accuracy: 0.4158 - val_loss: 3.2380 - val_accuracy: 0.0626\n",
+      "Epoch 730/5000\n",
+      "919/919 - 3s - loss: 1.8147 - accuracy: 0.4173 - val_loss: 3.2387 - val_accuracy: 0.0626\n",
+      "Epoch 731/5000\n",
+      "919/919 - 3s - loss: 1.8146 - accuracy: 0.4219 - val_loss: 3.2656 - val_accuracy: 0.0626\n",
+      "Epoch 732/5000\n",
+      "919/919 - 3s - loss: 1.8211 - accuracy: 0.4179 - val_loss: 3.2733 - val_accuracy: 0.0627\n",
+      "Epoch 733/5000\n",
+      "919/919 - 3s - loss: 1.9343 - accuracy: 0.4186 - val_loss: 3.2726 - val_accuracy: 0.0627\n",
+      "Epoch 734/5000\n",
+      "919/919 - 3s - loss: 1.8282 - accuracy: 0.4146 - val_loss: 3.2830 - val_accuracy: 0.0628\n",
+      "Epoch 735/5000\n",
+      "919/919 - 3s - loss: 1.8398 - accuracy: 0.4152 - val_loss: 3.2602 - val_accuracy: 0.0628\n",
+      "Epoch 736/5000\n",
+      "919/919 - 3s - loss: 1.8135 - accuracy: 0.4207 - val_loss: 3.2582 - val_accuracy: 0.0626\n",
+      "Epoch 737/5000\n",
+      "919/919 - 3s - loss: 1.8454 - accuracy: 0.4205 - val_loss: 3.2583 - val_accuracy: 0.0626\n",
+      "Epoch 738/5000\n",
+      "919/919 - 3s - loss: 1.8229 - accuracy: 0.4195 - val_loss: 3.2571 - val_accuracy: 0.0628\n",
+      "Epoch 739/5000\n",
+      "919/919 - 3s - loss: 1.9241 - accuracy: 0.4174 - val_loss: 3.2606 - val_accuracy: 0.0630\n",
+      "Epoch 740/5000\n",
+      "919/919 - 3s - loss: 1.8501 - accuracy: 0.4198 - val_loss: 3.2590 - val_accuracy: 0.0630\n",
+      "Epoch 741/5000\n",
+      "919/919 - 3s - loss: 1.8505 - accuracy: 0.4150 - val_loss: 3.2616 - val_accuracy: 0.0627\n",
+      "Epoch 742/5000\n",
+      "919/919 - 3s - loss: 1.8931 - accuracy: 0.4204 - val_loss: 3.2857 - val_accuracy: 0.0628\n",
+      "Epoch 743/5000\n",
+      "919/919 - 3s - loss: 1.8142 - accuracy: 0.4200 - val_loss: 3.2882 - val_accuracy: 0.0629\n",
+      "Epoch 744/5000\n",
+      "919/919 - 3s - loss: 1.8164 - accuracy: 0.4224 - val_loss: 3.2806 - val_accuracy: 0.0629\n",
+      "Epoch 745/5000\n",
+      "919/919 - 3s - loss: 1.8318 - accuracy: 0.4232 - val_loss: 3.2741 - val_accuracy: 0.0628\n",
+      "Epoch 746/5000\n",
+      "919/919 - 3s - loss: 1.8173 - accuracy: 0.4254 - val_loss: 3.2889 - val_accuracy: 0.0627\n",
+      "Epoch 747/5000\n",
+      "919/919 - 3s - loss: 1.8698 - accuracy: 0.4218 - val_loss: 3.2935 - val_accuracy: 0.0627\n",
+      "Epoch 748/5000\n",
+      "919/919 - 3s - loss: 1.8160 - accuracy: 0.4202 - val_loss: 3.2924 - val_accuracy: 0.0627\n",
+      "Epoch 749/5000\n",
+      "919/919 - 3s - loss: 1.8108 - accuracy: 0.4203 - val_loss: 3.2913 - val_accuracy: 0.0627\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 750/5000\n",
+      "919/919 - 3s - loss: 1.8120 - accuracy: 0.4176 - val_loss: 3.3034 - val_accuracy: 0.0628\n",
+      "Epoch 751/5000\n",
+      "919/919 - 3s - loss: 1.8362 - accuracy: 0.4227 - val_loss: 3.3126 - val_accuracy: 0.0628\n",
+      "Epoch 752/5000\n",
+      "919/919 - 3s - loss: 1.8209 - accuracy: 0.4192 - val_loss: 3.3019 - val_accuracy: 0.0627\n",
+      "Epoch 753/5000\n",
+      "919/919 - 3s - loss: 1.8170 - accuracy: 0.4214 - val_loss: 3.2987 - val_accuracy: 0.0627\n",
+      "Epoch 754/5000\n",
+      "919/919 - 3s - loss: 1.8746 - accuracy: 0.4233 - val_loss: 3.2918 - val_accuracy: 0.0626\n",
+      "Epoch 755/5000\n",
+      "919/919 - 3s - loss: 1.8033 - accuracy: 0.4235 - val_loss: 3.2806 - val_accuracy: 0.0627\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 756/5000\n",
+      "919/919 - 3s - loss: 1.8191 - accuracy: 0.4183 - val_loss: 3.2820 - val_accuracy: 0.0627\n",
+      "Epoch 757/5000\n",
+      "919/919 - 3s - loss: 1.8215 - accuracy: 0.4192 - val_loss: 3.2815 - val_accuracy: 0.0627\n",
+      "Epoch 758/5000\n",
+      "919/919 - 3s - loss: 1.8007 - accuracy: 0.4233 - val_loss: 3.2839 - val_accuracy: 0.0627\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 759/5000\n",
+      "919/919 - 3s - loss: 1.8106 - accuracy: 0.4213 - val_loss: 3.2819 - val_accuracy: 0.0628\n",
+      "Epoch 760/5000\n",
+      "919/919 - 3s - loss: 1.8174 - accuracy: 0.4219 - val_loss: 3.2494 - val_accuracy: 0.0628\n",
+      "Epoch 761/5000\n",
+      "919/919 - 3s - loss: 1.8189 - accuracy: 0.4196 - val_loss: 3.2435 - val_accuracy: 0.0629\n",
+      "Epoch 762/5000\n",
+      "919/919 - 3s - loss: 1.8105 - accuracy: 0.4224 - val_loss: 3.2332 - val_accuracy: 0.0629\n",
+      "Epoch 763/5000\n",
+      "919/919 - 3s - loss: 1.8240 - accuracy: 0.4226 - val_loss: 3.2314 - val_accuracy: 0.0626\n",
+      "Epoch 764/5000\n",
+      "919/919 - 3s - loss: 1.8581 - accuracy: 0.4173 - val_loss: 3.2308 - val_accuracy: 0.0627\n",
+      "Epoch 765/5000\n",
+      "919/919 - 3s - loss: 1.8008 - accuracy: 0.4223 - val_loss: 3.2378 - val_accuracy: 0.0624\n",
+      "Epoch 766/5000\n",
+      "919/919 - 3s - loss: 1.8096 - accuracy: 0.4194 - val_loss: 3.2453 - val_accuracy: 0.0622\n",
+      "Epoch 767/5000\n",
+      "919/919 - 3s - loss: 1.8865 - accuracy: 0.4221 - val_loss: 3.2427 - val_accuracy: 0.0623\n",
+      "Epoch 768/5000\n",
+      "919/919 - 3s - loss: 1.8109 - accuracy: 0.4205 - val_loss: 3.2518 - val_accuracy: 0.0624\n",
+      "Epoch 769/5000\n",
+      "919/919 - 3s - loss: 1.8167 - accuracy: 0.4226 - val_loss: 3.2417 - val_accuracy: 0.0625\n",
+      "Epoch 770/5000\n",
+      "919/919 - 3s - loss: 1.8069 - accuracy: 0.4227 - val_loss: 3.2435 - val_accuracy: 0.0625\n",
+      "Epoch 771/5000\n",
+      "919/919 - 3s - loss: 1.8376 - accuracy: 0.4173 - val_loss: 3.2382 - val_accuracy: 0.0626\n",
+      "Epoch 772/5000\n",
+      "919/919 - 3s - loss: 1.8266 - accuracy: 0.4202 - val_loss: 3.2422 - val_accuracy: 0.0625\n",
+      "Epoch 773/5000\n",
+      "919/919 - 3s - loss: 1.8068 - accuracy: 0.4215 - val_loss: 3.2436 - val_accuracy: 0.0625\n",
+      "Epoch 774/5000\n",
+      "919/919 - 3s - loss: 1.8001 - accuracy: 0.4214 - val_loss: 3.2468 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 775/5000\n",
+      "919/919 - 3s - loss: 1.8025 - accuracy: 0.4267 - val_loss: 3.2445 - val_accuracy: 0.0626\n",
+      "Epoch 776/5000\n",
+      "919/919 - 3s - loss: 1.8060 - accuracy: 0.4202 - val_loss: 3.2369 - val_accuracy: 0.0625\n",
+      "Epoch 777/5000\n",
+      "919/919 - 3s - loss: 1.8552 - accuracy: 0.4216 - val_loss: 3.2514 - val_accuracy: 0.0626\n",
+      "Epoch 778/5000\n",
+      "919/919 - 3s - loss: 1.8128 - accuracy: 0.4197 - val_loss: 3.2435 - val_accuracy: 0.0627\n",
+      "Epoch 779/5000\n",
+      "919/919 - 3s - loss: 1.8070 - accuracy: 0.4227 - val_loss: 3.2405 - val_accuracy: 0.0628\n",
+      "Epoch 780/5000\n",
+      "919/919 - 3s - loss: 1.8177 - accuracy: 0.4259 - val_loss: 3.2519 - val_accuracy: 0.0625\n",
+      "Epoch 781/5000\n",
+      "919/919 - 3s - loss: 1.8174 - accuracy: 0.4218 - val_loss: 3.2645 - val_accuracy: 0.0623\n",
+      "Epoch 782/5000\n",
+      "919/919 - 3s - loss: 1.8331 - accuracy: 0.4278 - val_loss: 3.2669 - val_accuracy: 0.0625\n",
+      "Epoch 783/5000\n",
+      "919/919 - 3s - loss: 1.7979 - accuracy: 0.4215 - val_loss: 3.2715 - val_accuracy: 0.0625\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 784/5000\n",
+      "919/919 - 3s - loss: 1.8027 - accuracy: 0.4200 - val_loss: 3.2802 - val_accuracy: 0.0627\n",
+      "Epoch 785/5000\n",
+      "919/919 - 3s - loss: 1.8114 - accuracy: 0.4235 - val_loss: 3.2685 - val_accuracy: 0.0628\n",
+      "Epoch 786/5000\n",
+      "919/919 - 3s - loss: 1.7957 - accuracy: 0.4248 - val_loss: 3.2670 - val_accuracy: 0.0628\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 787/5000\n",
+      "919/919 - 3s - loss: 1.8014 - accuracy: 0.4217 - val_loss: 3.2730 - val_accuracy: 0.0627\n",
+      "Epoch 788/5000\n",
+      "919/919 - 3s - loss: 1.8097 - accuracy: 0.4239 - val_loss: 3.2578 - val_accuracy: 0.0620\n",
+      "Epoch 789/5000\n",
+      "919/919 - 3s - loss: 1.8031 - accuracy: 0.4201 - val_loss: 3.2666 - val_accuracy: 0.0618\n",
+      "Epoch 790/5000\n",
+      "919/919 - 3s - loss: 1.7924 - accuracy: 0.4265 - val_loss: 3.2812 - val_accuracy: 0.0621\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 791/5000\n",
+      "919/919 - 3s - loss: 1.8035 - accuracy: 0.4221 - val_loss: 3.2831 - val_accuracy: 0.0619\n",
+      "Epoch 792/5000\n",
+      "919/919 - 3s - loss: 1.7790 - accuracy: 0.4273 - val_loss: 3.2828 - val_accuracy: 0.0619\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 793/5000\n",
+      "919/919 - 3s - loss: 1.8363 - accuracy: 0.4245 - val_loss: 3.2714 - val_accuracy: 0.0618\n",
+      "Epoch 794/5000\n",
+      "919/919 - 3s - loss: 1.8024 - accuracy: 0.4244 - val_loss: 3.2687 - val_accuracy: 0.0619\n",
+      "Epoch 795/5000\n",
+      "919/919 - 3s - loss: 1.7945 - accuracy: 0.4220 - val_loss: 3.2747 - val_accuracy: 0.0620\n",
+      "Epoch 796/5000\n",
+      "919/919 - 3s - loss: 1.7870 - accuracy: 0.4266 - val_loss: 3.2781 - val_accuracy: 0.0620\n",
+      "Epoch 797/5000\n",
+      "919/919 - 3s - loss: 1.8003 - accuracy: 0.4234 - val_loss: 3.2893 - val_accuracy: 0.0620\n",
+      "Epoch 798/5000\n",
+      "919/919 - 3s - loss: 1.7927 - accuracy: 0.4254 - val_loss: 3.2854 - val_accuracy: 0.0621\n",
+      "Epoch 799/5000\n",
+      "919/919 - 3s - loss: 1.8107 - accuracy: 0.4216 - val_loss: 3.2898 - val_accuracy: 0.0622\n",
+      "Epoch 800/5000\n",
+      "919/919 - 3s - loss: 1.8114 - accuracy: 0.4232 - val_loss: 3.3079 - val_accuracy: 0.0623\n",
+      "Epoch 801/5000\n",
+      "919/919 - 3s - loss: 1.7975 - accuracy: 0.4254 - val_loss: 3.3031 - val_accuracy: 0.0622\n",
+      "Epoch 802/5000\n",
+      "919/919 - 3s - loss: 1.9100 - accuracy: 0.4250 - val_loss: 3.2958 - val_accuracy: 0.0622\n",
+      "Epoch 803/5000\n",
+      "919/919 - 3s - loss: 1.9156 - accuracy: 0.4180 - val_loss: 3.2908 - val_accuracy: 0.0621\n",
+      "Epoch 804/5000\n",
+      "919/919 - 3s - loss: 1.8238 - accuracy: 0.4190 - val_loss: 3.2835 - val_accuracy: 0.0621\n",
+      "Epoch 805/5000\n",
+      "919/919 - 3s - loss: 1.7835 - accuracy: 0.4266 - val_loss: 3.2833 - val_accuracy: 0.0621\n",
+      "Epoch 806/5000\n",
+      "919/919 - 3s - loss: 1.7945 - accuracy: 0.4250 - val_loss: 3.2929 - val_accuracy: 0.0622\n",
+      "Epoch 807/5000\n",
+      "919/919 - 3s - loss: 1.8776 - accuracy: 0.4287 - val_loss: 3.2817 - val_accuracy: 0.0622\n",
+      "Epoch 808/5000\n",
+      "919/919 - 3s - loss: 1.7950 - accuracy: 0.4272 - val_loss: 3.2779 - val_accuracy: 0.0620\n",
+      "Epoch 809/5000\n",
+      "919/919 - 3s - loss: 1.7822 - accuracy: 0.4248 - val_loss: 3.2725 - val_accuracy: 0.0620\n",
+      "Epoch 810/5000\n",
+      "919/919 - 3s - loss: 1.7885 - accuracy: 0.4236 - val_loss: 3.2770 - val_accuracy: 0.0620\n",
+      "Epoch 811/5000\n",
+      "919/919 - 3s - loss: 1.7837 - accuracy: 0.4281 - val_loss: 3.2679 - val_accuracy: 0.0621\n",
+      "Epoch 812/5000\n",
+      "919/919 - 3s - loss: 1.8536 - accuracy: 0.4233 - val_loss: 3.2606 - val_accuracy: 0.0621\n",
+      "Epoch 813/5000\n",
+      "919/919 - 3s - loss: 1.8267 - accuracy: 0.4271 - val_loss: 3.2740 - val_accuracy: 0.0621\n",
+      "Epoch 814/5000\n",
+      "919/919 - 3s - loss: 1.8139 - accuracy: 0.4201 - val_loss: 3.2814 - val_accuracy: 0.0620\n",
+      "Epoch 815/5000\n",
+      "919/919 - 3s - loss: 1.8043 - accuracy: 0.4276 - val_loss: 3.2746 - val_accuracy: 0.0620\n",
+      "Epoch 816/5000\n",
+      "919/919 - 3s - loss: 1.7806 - accuracy: 0.4268 - val_loss: 3.2746 - val_accuracy: 0.0621\n",
+      "Epoch 817/5000\n",
+      "919/919 - 3s - loss: 1.8012 - accuracy: 0.4223 - val_loss: 3.2873 - val_accuracy: 0.0621\n",
+      "Epoch 818/5000\n",
+      "919/919 - 3s - loss: 1.7863 - accuracy: 0.4295 - val_loss: 3.2779 - val_accuracy: 0.0619\n",
+      "Epoch 819/5000\n",
+      "919/919 - 3s - loss: 1.7807 - accuracy: 0.4280 - val_loss: 3.2696 - val_accuracy: 0.0619\n",
+      "Epoch 820/5000\n",
+      "919/919 - 3s - loss: 1.8622 - accuracy: 0.4255 - val_loss: 3.2793 - val_accuracy: 0.0618\n",
+      "Epoch 821/5000\n",
+      "919/919 - 3s - loss: 1.8094 - accuracy: 0.4278 - val_loss: 3.2786 - val_accuracy: 0.0618\n",
+      "Epoch 822/5000\n",
+      "919/919 - 3s - loss: 1.8245 - accuracy: 0.4250 - val_loss: 3.2742 - val_accuracy: 0.0620\n",
+      "Epoch 823/5000\n",
+      "919/919 - 3s - loss: 1.7763 - accuracy: 0.4307 - val_loss: 3.2750 - val_accuracy: 0.0617\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 824/5000\n",
+      "919/919 - 3s - loss: 1.7796 - accuracy: 0.4300 - val_loss: 3.2863 - val_accuracy: 0.0617\n",
+      "Epoch 825/5000\n",
+      "919/919 - 3s - loss: 1.7883 - accuracy: 0.4282 - val_loss: 3.2817 - val_accuracy: 0.0619\n",
+      "Epoch 826/5000\n",
+      "919/919 - 3s - loss: 1.7814 - accuracy: 0.4280 - val_loss: 3.2933 - val_accuracy: 0.0619\n",
+      "Epoch 827/5000\n",
+      "919/919 - 3s - loss: 1.7916 - accuracy: 0.4257 - val_loss: 3.2908 - val_accuracy: 0.0620\n",
+      "Epoch 828/5000\n",
+      "919/919 - 3s - loss: 1.7755 - accuracy: 0.4253 - val_loss: 3.3002 - val_accuracy: 0.0620\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 829/5000\n",
+      "919/919 - 3s - loss: 1.7667 - accuracy: 0.4301 - val_loss: 3.3142 - val_accuracy: 0.0620\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 830/5000\n",
+      "919/919 - 3s - loss: 1.7808 - accuracy: 0.4312 - val_loss: 3.3251 - val_accuracy: 0.0620\n",
+      "Epoch 831/5000\n",
+      "919/919 - 3s - loss: 1.7775 - accuracy: 0.4301 - val_loss: 3.3230 - val_accuracy: 0.0621\n",
+      "Epoch 832/5000\n",
+      "919/919 - 3s - loss: 1.7979 - accuracy: 0.4250 - val_loss: 3.3259 - val_accuracy: 0.0621\n",
+      "Epoch 833/5000\n",
+      "919/919 - 3s - loss: 1.7801 - accuracy: 0.4294 - val_loss: 3.3232 - val_accuracy: 0.0622\n",
+      "Epoch 834/5000\n",
+      "919/919 - 3s - loss: 1.7631 - accuracy: 0.4299 - val_loss: 3.3242 - val_accuracy: 0.0621\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 835/5000\n",
+      "919/919 - 3s - loss: 1.8439 - accuracy: 0.4280 - val_loss: 3.3209 - val_accuracy: 0.0370\n",
+      "Epoch 836/5000\n",
+      "919/919 - 3s - loss: 1.7729 - accuracy: 0.4278 - val_loss: 3.3229 - val_accuracy: 0.0368\n",
+      "Epoch 837/5000\n",
+      "919/919 - 3s - loss: 1.7863 - accuracy: 0.4271 - val_loss: 3.3019 - val_accuracy: 0.0373\n",
+      "Epoch 838/5000\n",
+      "919/919 - 3s - loss: 1.7846 - accuracy: 0.4293 - val_loss: 3.3100 - val_accuracy: 0.0372\n",
+      "Epoch 839/5000\n",
+      "919/919 - 3s - loss: 1.7917 - accuracy: 0.4273 - val_loss: 3.3017 - val_accuracy: 0.0372\n",
+      "Epoch 840/5000\n",
+      "919/919 - 3s - loss: 1.7787 - accuracy: 0.4291 - val_loss: 3.3058 - val_accuracy: 0.0369\n",
+      "Epoch 841/5000\n",
+      "919/919 - 3s - loss: 1.7879 - accuracy: 0.4289 - val_loss: 3.3125 - val_accuracy: 0.0373\n",
+      "Epoch 842/5000\n",
+      "919/919 - 3s - loss: 1.7763 - accuracy: 0.4330 - val_loss: 3.3033 - val_accuracy: 0.0374\n",
+      "Epoch 843/5000\n",
+      "919/919 - 3s - loss: 1.8207 - accuracy: 0.4348 - val_loss: 3.2963 - val_accuracy: 0.0372\n",
+      "Epoch 844/5000\n",
+      "919/919 - 3s - loss: 1.7718 - accuracy: 0.4301 - val_loss: 3.2907 - val_accuracy: 0.0367\n",
+      "Epoch 845/5000\n",
+      "919/919 - 3s - loss: 1.7850 - accuracy: 0.4361 - val_loss: 3.2742 - val_accuracy: 0.0371\n",
+      "Epoch 846/5000\n",
+      "919/919 - 3s - loss: 1.7915 - accuracy: 0.4313 - val_loss: 3.2719 - val_accuracy: 0.0367\n",
+      "Epoch 847/5000\n",
+      "919/919 - 3s - loss: 1.7709 - accuracy: 0.4297 - val_loss: 3.2853 - val_accuracy: 0.0365\n",
+      "Epoch 848/5000\n",
+      "919/919 - 3s - loss: 1.7735 - accuracy: 0.4285 - val_loss: 3.2878 - val_accuracy: 0.0365\n",
+      "Epoch 849/5000\n",
+      "919/919 - 3s - loss: 1.7804 - accuracy: 0.4299 - val_loss: 3.2853 - val_accuracy: 0.0366\n",
+      "Epoch 850/5000\n",
+      "919/919 - 3s - loss: 1.7702 - accuracy: 0.4272 - val_loss: 3.2807 - val_accuracy: 0.0366\n",
+      "Epoch 851/5000\n",
+      "919/919 - 3s - loss: 1.7609 - accuracy: 0.4306 - val_loss: 3.2946 - val_accuracy: 0.0365\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 852/5000\n",
+      "919/919 - 3s - loss: 1.7779 - accuracy: 0.4284 - val_loss: 3.2942 - val_accuracy: 0.0365\n",
+      "Epoch 853/5000\n",
+      "919/919 - 3s - loss: 1.8171 - accuracy: 0.4271 - val_loss: 3.2782 - val_accuracy: 0.0363\n",
+      "Epoch 854/5000\n",
+      "919/919 - 3s - loss: 1.7683 - accuracy: 0.4323 - val_loss: 3.2832 - val_accuracy: 0.0366\n",
+      "Epoch 855/5000\n",
+      "919/919 - 3s - loss: 1.8974 - accuracy: 0.4232 - val_loss: 3.2890 - val_accuracy: 0.0368\n",
+      "Epoch 856/5000\n",
+      "919/919 - 3s - loss: 1.7565 - accuracy: 0.4320 - val_loss: 3.2897 - val_accuracy: 0.0368\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 857/5000\n",
+      "919/919 - 3s - loss: 1.8084 - accuracy: 0.4297 - val_loss: 3.3063 - val_accuracy: 0.0370\n",
+      "Epoch 858/5000\n",
+      "919/919 - 3s - loss: 1.7907 - accuracy: 0.4345 - val_loss: 3.2913 - val_accuracy: 0.0365\n",
+      "Epoch 859/5000\n",
+      "919/919 - 3s - loss: 1.7678 - accuracy: 0.4299 - val_loss: 3.2875 - val_accuracy: 0.0366\n",
+      "Epoch 860/5000\n",
+      "919/919 - 3s - loss: 1.7695 - accuracy: 0.4305 - val_loss: 3.2774 - val_accuracy: 0.0363\n",
+      "Epoch 861/5000\n",
+      "919/919 - 3s - loss: 1.7710 - accuracy: 0.4282 - val_loss: 3.2769 - val_accuracy: 0.0365\n",
+      "Epoch 862/5000\n",
+      "919/919 - 3s - loss: 1.7973 - accuracy: 0.4307 - val_loss: 3.2980 - val_accuracy: 0.0364\n",
+      "Epoch 863/5000\n",
+      "919/919 - 3s - loss: 1.7778 - accuracy: 0.4312 - val_loss: 3.2927 - val_accuracy: 0.0363\n",
+      "Epoch 864/5000\n",
+      "919/919 - 3s - loss: 1.7746 - accuracy: 0.4293 - val_loss: 3.2936 - val_accuracy: 0.0366\n",
+      "Epoch 865/5000\n",
+      "919/919 - 3s - loss: 1.7761 - accuracy: 0.4353 - val_loss: 3.3012 - val_accuracy: 0.0365\n",
+      "Epoch 866/5000\n",
+      "919/919 - 3s - loss: 1.7679 - accuracy: 0.4294 - val_loss: 3.3202 - val_accuracy: 0.0372\n",
+      "Epoch 867/5000\n",
+      "919/919 - 3s - loss: 1.7643 - accuracy: 0.4335 - val_loss: 3.3089 - val_accuracy: 0.0372\n",
+      "Epoch 868/5000\n",
+      "919/919 - 3s - loss: 1.7792 - accuracy: 0.4294 - val_loss: 3.2974 - val_accuracy: 0.0370\n",
+      "Epoch 869/5000\n",
+      "919/919 - 3s - loss: 1.7642 - accuracy: 0.4308 - val_loss: 3.3052 - val_accuracy: 0.0369\n",
+      "Epoch 870/5000\n",
+      "919/919 - 3s - loss: 1.7731 - accuracy: 0.4300 - val_loss: 3.2843 - val_accuracy: 0.0367\n",
+      "Epoch 871/5000\n",
+      "919/919 - 3s - loss: 1.7716 - accuracy: 0.4333 - val_loss: 3.2952 - val_accuracy: 0.0367\n",
+      "Epoch 872/5000\n",
+      "919/919 - 3s - loss: 1.8109 - accuracy: 0.4305 - val_loss: 3.2881 - val_accuracy: 0.0371\n",
+      "Epoch 873/5000\n",
+      "919/919 - 3s - loss: 1.7591 - accuracy: 0.4254 - val_loss: 3.2766 - val_accuracy: 0.0368\n",
+      "Epoch 874/5000\n",
+      "919/919 - 3s - loss: 1.7635 - accuracy: 0.4316 - val_loss: 3.2811 - val_accuracy: 0.0368\n",
+      "Epoch 875/5000\n",
+      "919/919 - 3s - loss: 1.7773 - accuracy: 0.4286 - val_loss: 3.2823 - val_accuracy: 0.0370\n",
+      "Epoch 876/5000\n",
+      "919/919 - 3s - loss: 1.7880 - accuracy: 0.4365 - val_loss: 3.2699 - val_accuracy: 0.0366\n",
+      "Epoch 877/5000\n",
+      "919/919 - 3s - loss: 1.7705 - accuracy: 0.4329 - val_loss: 3.2719 - val_accuracy: 0.0367\n",
+      "Epoch 878/5000\n",
+      "919/919 - 3s - loss: 1.7661 - accuracy: 0.4326 - val_loss: 3.2838 - val_accuracy: 0.0369\n",
+      "Epoch 879/5000\n",
+      "919/919 - 3s - loss: 1.8217 - accuracy: 0.4322 - val_loss: 3.2874 - val_accuracy: 0.0369\n",
+      "Epoch 880/5000\n",
+      "919/919 - 3s - loss: 1.7676 - accuracy: 0.4345 - val_loss: 3.2812 - val_accuracy: 0.0371\n",
+      "Epoch 881/5000\n",
+      "919/919 - 3s - loss: 1.7714 - accuracy: 0.4307 - val_loss: 3.2909 - val_accuracy: 0.0366\n",
+      "Epoch 882/5000\n",
+      "919/919 - 3s - loss: 1.8045 - accuracy: 0.4337 - val_loss: 3.2900 - val_accuracy: 0.0369\n",
+      "Epoch 883/5000\n",
+      "919/919 - 3s - loss: 1.7669 - accuracy: 0.4305 - val_loss: 3.2963 - val_accuracy: 0.0363\n",
+      "Epoch 884/5000\n",
+      "919/919 - 3s - loss: 1.7823 - accuracy: 0.4351 - val_loss: 3.3019 - val_accuracy: 0.0363\n",
+      "Epoch 885/5000\n",
+      "919/919 - 3s - loss: 1.8420 - accuracy: 0.4314 - val_loss: 3.2992 - val_accuracy: 0.0360\n",
+      "Epoch 886/5000\n",
+      "919/919 - 3s - loss: 1.7840 - accuracy: 0.4350 - val_loss: 3.2976 - val_accuracy: 0.0362\n",
+      "Epoch 887/5000\n",
+      "919/919 - 3s - loss: 1.7696 - accuracy: 0.4318 - val_loss: 3.2894 - val_accuracy: 0.0363\n",
+      "Epoch 888/5000\n",
+      "919/919 - 3s - loss: 1.9693 - accuracy: 0.4318 - val_loss: 3.2865 - val_accuracy: 0.0365\n",
+      "Epoch 889/5000\n",
+      "919/919 - 3s - loss: 1.7789 - accuracy: 0.4296 - val_loss: 3.2803 - val_accuracy: 0.0364\n",
+      "Epoch 890/5000\n",
+      "919/919 - 3s - loss: 1.7581 - accuracy: 0.4318 - val_loss: 3.2902 - val_accuracy: 0.0366\n",
+      "Epoch 891/5000\n",
+      "919/919 - 3s - loss: 1.7662 - accuracy: 0.4294 - val_loss: 3.2881 - val_accuracy: 0.0366\n",
+      "Epoch 892/5000\n",
+      "919/919 - 3s - loss: 1.7589 - accuracy: 0.4316 - val_loss: 3.2951 - val_accuracy: 0.0371\n",
+      "Epoch 893/5000\n",
+      "919/919 - 3s - loss: 1.7660 - accuracy: 0.4291 - val_loss: 3.2877 - val_accuracy: 0.0372\n",
+      "Epoch 894/5000\n",
+      "919/919 - 3s - loss: 1.7931 - accuracy: 0.4305 - val_loss: 3.2774 - val_accuracy: 0.0369\n",
+      "Epoch 895/5000\n",
+      "919/919 - 3s - loss: 1.7486 - accuracy: 0.4322 - val_loss: 3.2887 - val_accuracy: 0.0368\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 896/5000\n",
+      "919/919 - 3s - loss: 1.7668 - accuracy: 0.4310 - val_loss: 3.2924 - val_accuracy: 0.0369\n",
+      "Epoch 897/5000\n",
+      "919/919 - 3s - loss: 1.7770 - accuracy: 0.4303 - val_loss: 3.3015 - val_accuracy: 0.0376\n",
+      "Epoch 898/5000\n",
+      "919/919 - 3s - loss: 1.7736 - accuracy: 0.4339 - val_loss: 3.3067 - val_accuracy: 0.0372\n",
+      "Epoch 899/5000\n",
+      "919/919 - 3s - loss: 1.7702 - accuracy: 0.4365 - val_loss: 3.3158 - val_accuracy: 0.0369\n",
+      "Epoch 900/5000\n",
+      "919/919 - 3s - loss: 1.7565 - accuracy: 0.4312 - val_loss: 3.3151 - val_accuracy: 0.0372\n",
+      "Epoch 901/5000\n",
+      "919/919 - 3s - loss: 1.7656 - accuracy: 0.4350 - val_loss: 3.3250 - val_accuracy: 0.0371\n",
+      "Epoch 902/5000\n",
+      "919/919 - 3s - loss: 1.7776 - accuracy: 0.4352 - val_loss: 3.3091 - val_accuracy: 0.0375\n",
+      "Epoch 903/5000\n",
+      "919/919 - 3s - loss: 1.7696 - accuracy: 0.4360 - val_loss: 3.3177 - val_accuracy: 0.0375\n",
+      "Epoch 904/5000\n",
+      "919/919 - 3s - loss: 1.7491 - accuracy: 0.4310 - val_loss: 3.3180 - val_accuracy: 0.0377\n",
+      "Epoch 905/5000\n",
+      "919/919 - 3s - loss: 1.7651 - accuracy: 0.4366 - val_loss: 3.3117 - val_accuracy: 0.0375\n",
+      "Epoch 906/5000\n",
+      "919/919 - 3s - loss: 1.7477 - accuracy: 0.4390 - val_loss: 3.3079 - val_accuracy: 0.0377\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 907/5000\n",
+      "919/919 - 3s - loss: 1.7607 - accuracy: 0.4288 - val_loss: 3.3101 - val_accuracy: 0.0373\n",
+      "Epoch 908/5000\n",
+      "919/919 - 3s - loss: 1.7488 - accuracy: 0.4376 - val_loss: 3.3057 - val_accuracy: 0.0373\n",
+      "Epoch 909/5000\n",
+      "919/919 - 3s - loss: 1.7567 - accuracy: 0.4337 - val_loss: 3.3069 - val_accuracy: 0.0375\n",
+      "Epoch 910/5000\n",
+      "919/919 - 3s - loss: 1.8538 - accuracy: 0.4411 - val_loss: 3.3255 - val_accuracy: 0.0375\n",
+      "Epoch 911/5000\n",
+      "919/919 - 3s - loss: 1.7512 - accuracy: 0.4380 - val_loss: 3.3188 - val_accuracy: 0.0373\n",
+      "Epoch 912/5000\n",
+      "919/919 - 3s - loss: 1.7498 - accuracy: 0.4387 - val_loss: 3.3081 - val_accuracy: 0.0372\n",
+      "Epoch 913/5000\n",
+      "919/919 - 3s - loss: 1.7479 - accuracy: 0.4367 - val_loss: 3.3104 - val_accuracy: 0.0376\n",
+      "Epoch 914/5000\n",
+      "919/919 - 3s - loss: 1.7605 - accuracy: 0.4351 - val_loss: 3.3280 - val_accuracy: 0.0378\n",
+      "Epoch 915/5000\n",
+      "919/919 - 3s - loss: 1.7608 - accuracy: 0.4333 - val_loss: 3.3098 - val_accuracy: 0.0377\n",
+      "Epoch 916/5000\n",
+      "919/919 - 3s - loss: 1.7524 - accuracy: 0.4393 - val_loss: 3.3081 - val_accuracy: 0.0383\n",
+      "Epoch 917/5000\n",
+      "919/919 - 3s - loss: 1.7662 - accuracy: 0.4358 - val_loss: 3.3003 - val_accuracy: 0.0381\n",
+      "Epoch 918/5000\n",
+      "919/919 - 3s - loss: 1.7656 - accuracy: 0.4359 - val_loss: 3.3074 - val_accuracy: 0.0380\n",
+      "Epoch 919/5000\n",
+      "919/919 - 3s - loss: 1.8006 - accuracy: 0.4405 - val_loss: 3.3016 - val_accuracy: 0.0373\n",
+      "Epoch 920/5000\n",
+      "919/919 - 3s - loss: 1.8242 - accuracy: 0.4394 - val_loss: 3.3060 - val_accuracy: 0.0375\n",
+      "Epoch 921/5000\n",
+      "919/919 - 3s - loss: 1.7452 - accuracy: 0.4338 - val_loss: 3.3070 - val_accuracy: 0.0374\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 922/5000\n",
+      "919/919 - 3s - loss: 1.7527 - accuracy: 0.4374 - val_loss: 3.2965 - val_accuracy: 0.0372\n",
+      "Epoch 923/5000\n",
+      "919/919 - 3s - loss: 1.7602 - accuracy: 0.4367 - val_loss: 3.3041 - val_accuracy: 0.0373\n",
+      "Epoch 924/5000\n",
+      "919/919 - 3s - loss: 1.7459 - accuracy: 0.4339 - val_loss: 3.3040 - val_accuracy: 0.0376\n",
+      "Epoch 925/5000\n",
+      "919/919 - 3s - loss: 1.7673 - accuracy: 0.4329 - val_loss: 3.2992 - val_accuracy: 0.0374\n",
+      "Epoch 926/5000\n",
+      "919/919 - 3s - loss: 1.7410 - accuracy: 0.4372 - val_loss: 3.3050 - val_accuracy: 0.0373\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 927/5000\n",
+      "919/919 - 3s - loss: 1.7455 - accuracy: 0.4371 - val_loss: 3.3033 - val_accuracy: 0.0377\n",
+      "Epoch 928/5000\n",
+      "919/919 - 3s - loss: 1.7519 - accuracy: 0.4382 - val_loss: 3.3036 - val_accuracy: 0.0381\n",
+      "Epoch 929/5000\n",
+      "919/919 - 3s - loss: 1.7664 - accuracy: 0.4338 - val_loss: 3.3130 - val_accuracy: 0.0380\n",
+      "Epoch 930/5000\n",
+      "919/919 - 3s - loss: 1.8241 - accuracy: 0.4356 - val_loss: 3.2979 - val_accuracy: 0.0377\n",
+      "Epoch 931/5000\n",
+      "919/919 - 3s - loss: 1.7715 - accuracy: 0.4368 - val_loss: 3.2889 - val_accuracy: 0.0381\n",
+      "Epoch 932/5000\n",
+      "919/919 - 3s - loss: 1.7451 - accuracy: 0.4403 - val_loss: 3.2875 - val_accuracy: 0.0380\n",
+      "Epoch 933/5000\n",
+      "919/919 - 3s - loss: 1.7487 - accuracy: 0.4384 - val_loss: 3.2909 - val_accuracy: 0.0376\n",
+      "Epoch 934/5000\n",
+      "919/919 - 3s - loss: 1.7498 - accuracy: 0.4336 - val_loss: 3.2893 - val_accuracy: 0.0376\n",
+      "Epoch 935/5000\n",
+      "919/919 - 3s - loss: 1.7562 - accuracy: 0.4373 - val_loss: 3.2881 - val_accuracy: 0.0382\n",
+      "Epoch 936/5000\n",
+      "919/919 - 3s - loss: 1.7446 - accuracy: 0.4378 - val_loss: 3.2948 - val_accuracy: 0.0382\n",
+      "Epoch 937/5000\n",
+      "919/919 - 3s - loss: 1.7374 - accuracy: 0.4456 - val_loss: 3.2945 - val_accuracy: 0.0380\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 938/5000\n",
+      "919/919 - 3s - loss: 1.7490 - accuracy: 0.4367 - val_loss: 3.2863 - val_accuracy: 0.0372\n",
+      "Epoch 939/5000\n",
+      "919/919 - 3s - loss: 1.7831 - accuracy: 0.4376 - val_loss: 3.2873 - val_accuracy: 0.0374\n",
+      "Epoch 940/5000\n",
+      "919/919 - 3s - loss: 1.7913 - accuracy: 0.4371 - val_loss: 3.2962 - val_accuracy: 0.0372\n",
+      "Epoch 941/5000\n",
+      "919/919 - 3s - loss: 1.7484 - accuracy: 0.4351 - val_loss: 3.2939 - val_accuracy: 0.0380\n",
+      "Epoch 942/5000\n",
+      "919/919 - 3s - loss: 1.7395 - accuracy: 0.4437 - val_loss: 3.2909 - val_accuracy: 0.0376\n",
+      "Epoch 943/5000\n",
+      "919/919 - 3s - loss: 1.7445 - accuracy: 0.4390 - val_loss: 3.2867 - val_accuracy: 0.0376\n",
+      "Epoch 944/5000\n",
+      "919/919 - 3s - loss: 1.7486 - accuracy: 0.4409 - val_loss: 3.2827 - val_accuracy: 0.0374\n",
+      "Epoch 945/5000\n",
+      "919/919 - 3s - loss: 1.7497 - accuracy: 0.4356 - val_loss: 3.2786 - val_accuracy: 0.0375\n",
+      "Epoch 946/5000\n",
+      "919/919 - 3s - loss: 1.7386 - accuracy: 0.4397 - val_loss: 3.2900 - val_accuracy: 0.0375\n",
+      "Epoch 947/5000\n",
+      "919/919 - 3s - loss: 1.7533 - accuracy: 0.4377 - val_loss: 3.2993 - val_accuracy: 0.0377\n",
+      "Epoch 948/5000\n",
+      "919/919 - 3s - loss: 1.7602 - accuracy: 0.4404 - val_loss: 3.2979 - val_accuracy: 0.0381\n",
+      "Epoch 949/5000\n",
+      "919/919 - 3s - loss: 1.7550 - accuracy: 0.4436 - val_loss: 3.2864 - val_accuracy: 0.0376\n",
+      "Epoch 950/5000\n",
+      "919/919 - 3s - loss: 1.7407 - accuracy: 0.4405 - val_loss: 3.2950 - val_accuracy: 0.0377\n",
+      "Epoch 951/5000\n",
+      "919/919 - 3s - loss: 1.7756 - accuracy: 0.4393 - val_loss: 3.2988 - val_accuracy: 0.0375\n",
+      "Epoch 952/5000\n",
+      "919/919 - 3s - loss: 1.7432 - accuracy: 0.4386 - val_loss: 3.3008 - val_accuracy: 0.0377\n",
+      "Epoch 953/5000\n",
+      "919/919 - 3s - loss: 1.8209 - accuracy: 0.4415 - val_loss: 3.2942 - val_accuracy: 0.0382\n",
+      "Epoch 954/5000\n",
+      "919/919 - 3s - loss: 1.7581 - accuracy: 0.4361 - val_loss: 3.2974 - val_accuracy: 0.0377\n",
+      "Epoch 955/5000\n",
+      "919/919 - 3s - loss: 1.8453 - accuracy: 0.4390 - val_loss: 3.2978 - val_accuracy: 0.0379\n",
+      "Epoch 956/5000\n",
+      "919/919 - 3s - loss: 1.7496 - accuracy: 0.4361 - val_loss: 3.3147 - val_accuracy: 0.0381\n",
+      "Epoch 957/5000\n",
+      "919/919 - 3s - loss: 1.7523 - accuracy: 0.4377 - val_loss: 3.3119 - val_accuracy: 0.0380\n",
+      "Epoch 958/5000\n",
+      "919/919 - 3s - loss: 1.7373 - accuracy: 0.4385 - val_loss: 3.3166 - val_accuracy: 0.0382\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 959/5000\n",
+      "919/919 - 3s - loss: 1.7648 - accuracy: 0.4387 - val_loss: 3.3147 - val_accuracy: 0.0387\n",
+      "Epoch 960/5000\n",
+      "919/919 - 3s - loss: 1.7782 - accuracy: 0.4384 - val_loss: 3.3218 - val_accuracy: 0.0383\n",
+      "Epoch 961/5000\n",
+      "919/919 - 3s - loss: 1.7445 - accuracy: 0.4383 - val_loss: 3.3197 - val_accuracy: 0.0377\n",
+      "Epoch 962/5000\n",
+      "919/919 - 3s - loss: 1.8961 - accuracy: 0.4378 - val_loss: 3.3168 - val_accuracy: 0.0381\n",
+      "Epoch 963/5000\n",
+      "919/919 - 3s - loss: 1.8004 - accuracy: 0.4382 - val_loss: 3.3218 - val_accuracy: 0.0383\n",
+      "Epoch 964/5000\n",
+      "919/919 - 3s - loss: 1.7479 - accuracy: 0.4424 - val_loss: 3.3247 - val_accuracy: 0.0380\n",
+      "Epoch 965/5000\n",
+      "919/919 - 3s - loss: 1.7861 - accuracy: 0.4380 - val_loss: 3.3379 - val_accuracy: 0.0378\n",
+      "Epoch 966/5000\n",
+      "919/919 - 3s - loss: 1.7432 - accuracy: 0.4418 - val_loss: 3.3551 - val_accuracy: 0.0380\n",
+      "Epoch 967/5000\n",
+      "919/919 - 3s - loss: 1.7804 - accuracy: 0.4390 - val_loss: 3.3377 - val_accuracy: 0.0383\n",
+      "Epoch 968/5000\n",
+      "919/919 - 3s - loss: 1.7333 - accuracy: 0.4393 - val_loss: 3.3352 - val_accuracy: 0.0380\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 969/5000\n",
+      "919/919 - 3s - loss: 1.7362 - accuracy: 0.4420 - val_loss: 3.3354 - val_accuracy: 0.0380\n",
+      "Epoch 970/5000\n",
+      "919/919 - 3s - loss: 1.7329 - accuracy: 0.4420 - val_loss: 3.3564 - val_accuracy: 0.0386\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 971/5000\n",
+      "919/919 - 3s - loss: 1.7411 - accuracy: 0.4418 - val_loss: 3.3514 - val_accuracy: 0.0390\n",
+      "Epoch 972/5000\n",
+      "919/919 - 3s - loss: 1.7356 - accuracy: 0.4410 - val_loss: 3.3302 - val_accuracy: 0.0388\n",
+      "Epoch 973/5000\n",
+      "919/919 - 3s - loss: 1.8451 - accuracy: 0.4416 - val_loss: 3.3305 - val_accuracy: 0.0389\n",
+      "Epoch 974/5000\n",
+      "919/919 - 3s - loss: 1.7409 - accuracy: 0.4412 - val_loss: 3.3461 - val_accuracy: 0.0392\n",
+      "Epoch 975/5000\n",
+      "919/919 - 3s - loss: 1.7593 - accuracy: 0.4363 - val_loss: 3.3395 - val_accuracy: 0.0390\n",
+      "Epoch 976/5000\n",
+      "919/919 - 3s - loss: 1.7229 - accuracy: 0.4420 - val_loss: 3.3402 - val_accuracy: 0.0391\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 977/5000\n",
+      "919/919 - 3s - loss: 1.7433 - accuracy: 0.4441 - val_loss: 3.3369 - val_accuracy: 0.0388\n",
+      "Epoch 978/5000\n",
+      "919/919 - 3s - loss: 1.7404 - accuracy: 0.4401 - val_loss: 3.3433 - val_accuracy: 0.0389\n",
+      "Epoch 979/5000\n",
+      "919/919 - 3s - loss: 1.7357 - accuracy: 0.4414 - val_loss: 3.3386 - val_accuracy: 0.0392\n",
+      "Epoch 980/5000\n",
+      "919/919 - 3s - loss: 1.7329 - accuracy: 0.4404 - val_loss: 3.3300 - val_accuracy: 0.0392\n",
+      "Epoch 981/5000\n",
+      "919/919 - 3s - loss: 1.7288 - accuracy: 0.4450 - val_loss: 3.3311 - val_accuracy: 0.0391\n",
+      "Epoch 982/5000\n",
+      "919/919 - 3s - loss: 1.7282 - accuracy: 0.4414 - val_loss: 3.3350 - val_accuracy: 0.0390\n",
+      "Epoch 983/5000\n",
+      "919/919 - 3s - loss: 1.8495 - accuracy: 0.4398 - val_loss: 3.3232 - val_accuracy: 0.0389\n",
+      "Epoch 984/5000\n",
+      "919/919 - 3s - loss: 1.7303 - accuracy: 0.4444 - val_loss: 3.3394 - val_accuracy: 0.0390\n",
+      "Epoch 985/5000\n",
+      "919/919 - 3s - loss: 1.7391 - accuracy: 0.4414 - val_loss: 3.3536 - val_accuracy: 0.0391\n",
+      "Epoch 986/5000\n",
+      "919/919 - 3s - loss: 1.7660 - accuracy: 0.4432 - val_loss: 3.3426 - val_accuracy: 0.0390\n",
+      "Epoch 987/5000\n",
+      "919/919 - 3s - loss: 1.8761 - accuracy: 0.4392 - val_loss: 3.3306 - val_accuracy: 0.0390\n",
+      "Epoch 988/5000\n",
+      "919/919 - 3s - loss: 1.7537 - accuracy: 0.4427 - val_loss: 3.3278 - val_accuracy: 0.0389\n",
+      "Epoch 989/5000\n",
+      "919/919 - 3s - loss: 1.7287 - accuracy: 0.4445 - val_loss: 3.3250 - val_accuracy: 0.0393\n",
+      "Epoch 990/5000\n",
+      "919/919 - 3s - loss: 1.7272 - accuracy: 0.4418 - val_loss: 3.3178 - val_accuracy: 0.0392\n",
+      "Epoch 991/5000\n",
+      "919/919 - 3s - loss: 1.7258 - accuracy: 0.4433 - val_loss: 3.3161 - val_accuracy: 0.0390\n",
+      "Epoch 992/5000\n",
+      "919/919 - 3s - loss: 1.7321 - accuracy: 0.4403 - val_loss: 3.3265 - val_accuracy: 0.0390\n",
+      "Epoch 993/5000\n",
+      "919/919 - 3s - loss: 1.7522 - accuracy: 0.4437 - val_loss: 3.3113 - val_accuracy: 0.0392\n",
+      "Epoch 994/5000\n",
+      "919/919 - 3s - loss: 1.7280 - accuracy: 0.4454 - val_loss: 3.3155 - val_accuracy: 0.0390\n",
+      "Epoch 995/5000\n",
+      "919/919 - 3s - loss: 1.7127 - accuracy: 0.4490 - val_loss: 3.3227 - val_accuracy: 0.0390\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 996/5000\n",
+      "919/919 - 3s - loss: 1.7658 - accuracy: 0.4415 - val_loss: 3.3309 - val_accuracy: 0.0391\n",
+      "Epoch 997/5000\n",
+      "919/919 - 3s - loss: 1.7317 - accuracy: 0.4399 - val_loss: 3.3216 - val_accuracy: 0.0389\n",
+      "Epoch 998/5000\n",
+      "919/919 - 3s - loss: 1.7460 - accuracy: 0.4438 - val_loss: 3.3278 - val_accuracy: 0.0389\n",
+      "Epoch 999/5000\n",
+      "919/919 - 3s - loss: 1.7351 - accuracy: 0.4433 - val_loss: 3.3358 - val_accuracy: 0.0392\n",
+      "Epoch 1000/5000\n",
+      "919/919 - 3s - loss: 1.7307 - accuracy: 0.4435 - val_loss: 3.3339 - val_accuracy: 0.0390\n",
+      "Epoch 1001/5000\n",
+      "919/919 - 3s - loss: 1.7305 - accuracy: 0.4415 - val_loss: 3.3263 - val_accuracy: 0.0388\n",
+      "Epoch 1002/5000\n",
+      "919/919 - 3s - loss: 1.7647 - accuracy: 0.4481 - val_loss: 3.3365 - val_accuracy: 0.0393\n",
+      "Epoch 1003/5000\n",
+      "919/919 - 3s - loss: 1.7660 - accuracy: 0.4480 - val_loss: 3.3320 - val_accuracy: 0.0389\n",
+      "Epoch 1004/5000\n",
+      "919/919 - 3s - loss: 1.7228 - accuracy: 0.4403 - val_loss: 3.3339 - val_accuracy: 0.0390\n",
+      "Epoch 1005/5000\n",
+      "919/919 - 3s - loss: 1.7179 - accuracy: 0.4448 - val_loss: 3.3267 - val_accuracy: 0.0389\n",
+      "Epoch 1006/5000\n",
+      "919/919 - 3s - loss: 1.7187 - accuracy: 0.4422 - val_loss: 3.3267 - val_accuracy: 0.0388\n",
+      "Epoch 1007/5000\n",
+      "919/919 - 3s - loss: 1.7959 - accuracy: 0.4458 - val_loss: 3.3321 - val_accuracy: 0.0388\n",
+      "Epoch 1008/5000\n",
+      "919/919 - 3s - loss: 1.7169 - accuracy: 0.4495 - val_loss: 3.3354 - val_accuracy: 0.0391\n",
+      "Epoch 1009/5000\n",
+      "919/919 - 3s - loss: 1.7260 - accuracy: 0.4469 - val_loss: 3.3421 - val_accuracy: 0.0390\n",
+      "Epoch 1010/5000\n",
+      "919/919 - 3s - loss: 1.7236 - accuracy: 0.4460 - val_loss: 3.3422 - val_accuracy: 0.0389\n",
+      "Epoch 1011/5000\n",
+      "919/919 - 3s - loss: 1.7275 - accuracy: 0.4470 - val_loss: 3.3456 - val_accuracy: 0.0391\n",
+      "Epoch 1012/5000\n",
+      "919/919 - 3s - loss: 1.7465 - accuracy: 0.4460 - val_loss: 3.3374 - val_accuracy: 0.0390\n",
+      "Epoch 1013/5000\n",
+      "919/919 - 3s - loss: 1.7227 - accuracy: 0.4459 - val_loss: 3.3542 - val_accuracy: 0.0394\n",
+      "Epoch 1014/5000\n",
+      "919/919 - 3s - loss: 1.7297 - accuracy: 0.4431 - val_loss: 3.3591 - val_accuracy: 0.0394\n",
+      "Epoch 1015/5000\n",
+      "919/919 - 3s - loss: 1.7523 - accuracy: 0.4411 - val_loss: 3.3432 - val_accuracy: 0.0392\n",
+      "Epoch 1016/5000\n",
+      "919/919 - 3s - loss: 1.7120 - accuracy: 0.4455 - val_loss: 3.3511 - val_accuracy: 0.0394\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1017/5000\n",
+      "919/919 - 3s - loss: 1.7150 - accuracy: 0.4507 - val_loss: 3.3617 - val_accuracy: 0.0394\n",
+      "Epoch 1018/5000\n",
+      "919/919 - 3s - loss: 1.7285 - accuracy: 0.4440 - val_loss: 3.3494 - val_accuracy: 0.0389\n",
+      "Epoch 1019/5000\n",
+      "919/919 - 3s - loss: 1.7254 - accuracy: 0.4454 - val_loss: 3.3376 - val_accuracy: 0.0390\n",
+      "Epoch 1020/5000\n",
+      "919/919 - 3s - loss: 1.7251 - accuracy: 0.4499 - val_loss: 3.3479 - val_accuracy: 0.0388\n",
+      "Epoch 1021/5000\n",
+      "919/919 - 3s - loss: 1.7179 - accuracy: 0.4423 - val_loss: 3.3644 - val_accuracy: 0.0396\n",
+      "Epoch 1022/5000\n",
+      "919/919 - 3s - loss: 1.7687 - accuracy: 0.4465 - val_loss: 3.3586 - val_accuracy: 0.0400\n",
+      "Epoch 1023/5000\n",
+      "919/919 - 3s - loss: 1.7239 - accuracy: 0.4469 - val_loss: 3.3537 - val_accuracy: 0.0397\n",
+      "Epoch 1024/5000\n",
+      "919/919 - 3s - loss: 1.7116 - accuracy: 0.4472 - val_loss: 3.3709 - val_accuracy: 0.0401\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1025/5000\n",
+      "919/919 - 3s - loss: 1.7403 - accuracy: 0.4439 - val_loss: 3.3639 - val_accuracy: 0.0399\n",
+      "Epoch 1026/5000\n",
+      "919/919 - 3s - loss: 1.7129 - accuracy: 0.4479 - val_loss: 3.3626 - val_accuracy: 0.0397\n",
+      "Epoch 1027/5000\n",
+      "919/919 - 3s - loss: 1.7383 - accuracy: 0.4440 - val_loss: 3.3568 - val_accuracy: 0.0393\n",
+      "Epoch 1028/5000\n",
+      "919/919 - 3s - loss: 1.7593 - accuracy: 0.4439 - val_loss: 3.3503 - val_accuracy: 0.0392\n",
+      "Epoch 1029/5000\n",
+      "919/919 - 3s - loss: 1.7205 - accuracy: 0.4496 - val_loss: 3.3597 - val_accuracy: 0.0393\n",
+      "Epoch 1030/5000\n",
+      "919/919 - 3s - loss: 1.7169 - accuracy: 0.4535 - val_loss: 3.3668 - val_accuracy: 0.0396\n",
+      "Epoch 1031/5000\n",
+      "919/919 - 3s - loss: 1.7188 - accuracy: 0.4445 - val_loss: 3.3767 - val_accuracy: 0.0396\n",
+      "Epoch 1032/5000\n",
+      "919/919 - 3s - loss: 1.7234 - accuracy: 0.4481 - val_loss: 3.3762 - val_accuracy: 0.0400\n",
+      "Epoch 1033/5000\n",
+      "919/919 - 3s - loss: 1.7364 - accuracy: 0.4475 - val_loss: 3.3748 - val_accuracy: 0.0397\n",
+      "Epoch 1034/5000\n",
+      "919/919 - 3s - loss: 1.7264 - accuracy: 0.4443 - val_loss: 3.3643 - val_accuracy: 0.0399\n",
+      "Epoch 1035/5000\n",
+      "919/919 - 3s - loss: 1.7221 - accuracy: 0.4518 - val_loss: 3.3611 - val_accuracy: 0.0399\n",
+      "Epoch 1036/5000\n",
+      "919/919 - 3s - loss: 1.7133 - accuracy: 0.4471 - val_loss: 3.3553 - val_accuracy: 0.0394\n",
+      "Epoch 1037/5000\n",
+      "919/919 - 3s - loss: 1.6978 - accuracy: 0.4536 - val_loss: 3.3525 - val_accuracy: 0.0401\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1038/5000\n",
+      "919/919 - 3s - loss: 1.7420 - accuracy: 0.4474 - val_loss: 3.3435 - val_accuracy: 0.0399\n",
+      "Epoch 1039/5000\n",
+      "919/919 - 3s - loss: 1.7200 - accuracy: 0.4507 - val_loss: 3.3427 - val_accuracy: 0.0399\n",
+      "Epoch 1040/5000\n",
+      "919/919 - 3s - loss: 1.7265 - accuracy: 0.4454 - val_loss: 3.3494 - val_accuracy: 0.0399\n",
+      "Epoch 1041/5000\n",
+      "919/919 - 3s - loss: 1.7187 - accuracy: 0.4477 - val_loss: 3.3601 - val_accuracy: 0.0396\n",
+      "Epoch 1042/5000\n",
+      "919/919 - 3s - loss: 1.7182 - accuracy: 0.4491 - val_loss: 3.3668 - val_accuracy: 0.0393\n",
+      "Epoch 1043/5000\n",
+      "919/919 - 3s - loss: 1.7223 - accuracy: 0.4482 - val_loss: 3.3758 - val_accuracy: 0.0394\n",
+      "Epoch 1044/5000\n",
+      "919/919 - 3s - loss: 1.7107 - accuracy: 0.4484 - val_loss: 3.3782 - val_accuracy: 0.0394\n",
+      "Epoch 1045/5000\n",
+      "919/919 - 3s - loss: 1.7056 - accuracy: 0.4516 - val_loss: 3.3722 - val_accuracy: 0.0395\n",
+      "Epoch 1046/5000\n",
+      "919/919 - 3s - loss: 1.7038 - accuracy: 0.4527 - val_loss: 3.3691 - val_accuracy: 0.0400\n",
+      "Epoch 1047/5000\n",
+      "919/919 - 3s - loss: 1.7120 - accuracy: 0.4507 - val_loss: 3.3554 - val_accuracy: 0.0396\n",
+      "Epoch 1048/5000\n",
+      "919/919 - 3s - loss: 1.6985 - accuracy: 0.4544 - val_loss: 3.3608 - val_accuracy: 0.0394\n",
+      "Epoch 1049/5000\n",
+      "919/919 - 3s - loss: 1.7044 - accuracy: 0.4510 - val_loss: 3.3613 - val_accuracy: 0.0393\n",
+      "Epoch 1050/5000\n",
+      "919/919 - 3s - loss: 1.7171 - accuracy: 0.4482 - val_loss: 3.3701 - val_accuracy: 0.0397\n",
+      "Epoch 1051/5000\n",
+      "919/919 - 3s - loss: 1.7180 - accuracy: 0.4500 - val_loss: 3.3770 - val_accuracy: 0.0396\n",
+      "Epoch 1052/5000\n",
+      "919/919 - 3s - loss: 1.7098 - accuracy: 0.4484 - val_loss: 3.3743 - val_accuracy: 0.0398\n",
+      "Epoch 1053/5000\n",
+      "919/919 - 3s - loss: 1.6996 - accuracy: 0.4527 - val_loss: 3.3816 - val_accuracy: 0.0396\n",
+      "Epoch 1054/5000\n",
+      "919/919 - 3s - loss: 1.8030 - accuracy: 0.4501 - val_loss: 3.3776 - val_accuracy: 0.0398\n",
+      "Epoch 1055/5000\n",
+      "919/919 - 3s - loss: 1.7158 - accuracy: 0.4492 - val_loss: 3.3755 - val_accuracy: 0.0398\n",
+      "Epoch 1056/5000\n",
+      "919/919 - 3s - loss: 1.7000 - accuracy: 0.4531 - val_loss: 3.3641 - val_accuracy: 0.0399\n",
+      "Epoch 1057/5000\n",
+      "919/919 - 3s - loss: 1.7146 - accuracy: 0.4497 - val_loss: 3.3595 - val_accuracy: 0.0399\n",
+      "Epoch 1058/5000\n",
+      "919/919 - 3s - loss: 1.7261 - accuracy: 0.4482 - val_loss: 3.3735 - val_accuracy: 0.0403\n",
+      "Epoch 1059/5000\n",
+      "919/919 - 3s - loss: 1.8397 - accuracy: 0.4493 - val_loss: 3.3656 - val_accuracy: 0.0402\n",
+      "Epoch 1060/5000\n",
+      "919/919 - 3s - loss: 1.7182 - accuracy: 0.4454 - val_loss: 3.3657 - val_accuracy: 0.0404\n",
+      "Epoch 1061/5000\n",
+      "919/919 - 3s - loss: 1.7152 - accuracy: 0.4531 - val_loss: 3.3710 - val_accuracy: 0.0401\n",
+      "Epoch 1062/5000\n",
+      "919/919 - 3s - loss: 1.7143 - accuracy: 0.4510 - val_loss: 3.3606 - val_accuracy: 0.0402\n",
+      "Epoch 1063/5000\n",
+      "919/919 - 3s - loss: 1.7134 - accuracy: 0.4505 - val_loss: 3.3565 - val_accuracy: 0.0402\n",
+      "Epoch 1064/5000\n",
+      "919/919 - 3s - loss: 1.7086 - accuracy: 0.4527 - val_loss: 3.3609 - val_accuracy: 0.0398\n",
+      "Epoch 1065/5000\n",
+      "919/919 - 3s - loss: 1.7128 - accuracy: 0.4539 - val_loss: 3.3558 - val_accuracy: 0.0399\n",
+      "Epoch 1066/5000\n",
+      "919/919 - 3s - loss: 1.7098 - accuracy: 0.4496 - val_loss: 3.3597 - val_accuracy: 0.0399\n",
+      "Epoch 1067/5000\n",
+      "919/919 - 3s - loss: 1.7100 - accuracy: 0.4518 - val_loss: 3.3554 - val_accuracy: 0.0400\n",
+      "Epoch 1068/5000\n",
+      "919/919 - 3s - loss: 1.7023 - accuracy: 0.4507 - val_loss: 3.3502 - val_accuracy: 0.0398\n",
+      "Epoch 1069/5000\n",
+      "919/919 - 3s - loss: 1.7114 - accuracy: 0.4481 - val_loss: 3.3528 - val_accuracy: 0.0402\n",
+      "Epoch 1070/5000\n",
+      "919/919 - 3s - loss: 1.7023 - accuracy: 0.4544 - val_loss: 3.3486 - val_accuracy: 0.0404\n",
+      "Epoch 1071/5000\n",
+      "919/919 - 3s - loss: 1.6969 - accuracy: 0.4532 - val_loss: 3.3418 - val_accuracy: 0.0401\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1072/5000\n",
+      "919/919 - 3s - loss: 1.7142 - accuracy: 0.4556 - val_loss: 3.3248 - val_accuracy: 0.0401\n",
+      "Epoch 1073/5000\n",
+      "919/919 - 3s - loss: 1.7027 - accuracy: 0.4546 - val_loss: 3.3322 - val_accuracy: 0.0402\n",
+      "Epoch 1074/5000\n",
+      "919/919 - 3s - loss: 1.7003 - accuracy: 0.4574 - val_loss: 3.3312 - val_accuracy: 0.0401\n",
+      "Epoch 1075/5000\n",
+      "919/919 - 3s - loss: 1.7080 - accuracy: 0.4560 - val_loss: 3.3270 - val_accuracy: 0.0400\n",
+      "Epoch 1076/5000\n",
+      "919/919 - 3s - loss: 1.7047 - accuracy: 0.4558 - val_loss: 3.3347 - val_accuracy: 0.0399\n",
+      "Epoch 1077/5000\n",
+      "919/919 - 3s - loss: 1.7077 - accuracy: 0.4549 - val_loss: 3.3426 - val_accuracy: 0.0403\n",
+      "Epoch 1078/5000\n",
+      "919/919 - 3s - loss: 1.7982 - accuracy: 0.4548 - val_loss: 3.3384 - val_accuracy: 0.0405\n",
+      "Epoch 1079/5000\n",
+      "919/919 - 3s - loss: 1.7136 - accuracy: 0.4505 - val_loss: 3.3406 - val_accuracy: 0.0407\n",
+      "Epoch 1080/5000\n",
+      "919/919 - 3s - loss: 1.7186 - accuracy: 0.4521 - val_loss: 3.3436 - val_accuracy: 0.0404\n",
+      "Epoch 1081/5000\n",
+      "919/919 - 3s - loss: 1.7131 - accuracy: 0.4471 - val_loss: 3.3488 - val_accuracy: 0.0401\n",
+      "Epoch 1082/5000\n",
+      "919/919 - 3s - loss: 1.7073 - accuracy: 0.4530 - val_loss: 3.3469 - val_accuracy: 0.0399\n",
+      "Epoch 1083/5000\n",
+      "919/919 - 3s - loss: 1.7067 - accuracy: 0.4521 - val_loss: 3.3501 - val_accuracy: 0.0400\n",
+      "Epoch 1084/5000\n",
+      "919/919 - 3s - loss: 1.7010 - accuracy: 0.4557 - val_loss: 3.3479 - val_accuracy: 0.0402\n",
+      "Epoch 1085/5000\n",
+      "919/919 - 3s - loss: 1.7013 - accuracy: 0.4539 - val_loss: 3.3554 - val_accuracy: 0.0399\n",
+      "Epoch 1086/5000\n",
+      "919/919 - 3s - loss: 1.7065 - accuracy: 0.4569 - val_loss: 3.3720 - val_accuracy: 0.0399\n",
+      "Epoch 1087/5000\n",
+      "919/919 - 3s - loss: 1.7080 - accuracy: 0.4550 - val_loss: 3.3761 - val_accuracy: 0.0402\n",
+      "Epoch 1088/5000\n",
+      "919/919 - 3s - loss: 1.7138 - accuracy: 0.4534 - val_loss: 3.3839 - val_accuracy: 0.0401\n",
+      "Epoch 1089/5000\n",
+      "919/919 - 3s - loss: 1.7046 - accuracy: 0.4520 - val_loss: 3.3884 - val_accuracy: 0.0402\n",
+      "Epoch 1090/5000\n",
+      "919/919 - 3s - loss: 1.7025 - accuracy: 0.4554 - val_loss: 3.3712 - val_accuracy: 0.0400\n",
+      "Epoch 1091/5000\n",
+      "919/919 - 3s - loss: 1.7006 - accuracy: 0.4552 - val_loss: 3.3791 - val_accuracy: 0.0402\n",
+      "Epoch 1092/5000\n",
+      "919/919 - 3s - loss: 1.7087 - accuracy: 0.4549 - val_loss: 3.3772 - val_accuracy: 0.0403\n",
+      "Epoch 1093/5000\n",
+      "919/919 - 3s - loss: 1.7121 - accuracy: 0.4562 - val_loss: 3.3689 - val_accuracy: 0.0405\n",
+      "Epoch 1094/5000\n",
+      "919/919 - 3s - loss: 1.6915 - accuracy: 0.4531 - val_loss: 3.3555 - val_accuracy: 0.0406\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1095/5000\n",
+      "919/919 - 3s - loss: 1.7013 - accuracy: 0.4521 - val_loss: 3.3616 - val_accuracy: 0.0403\n",
+      "Epoch 1096/5000\n",
+      "919/919 - 3s - loss: 1.7102 - accuracy: 0.4517 - val_loss: 3.3579 - val_accuracy: 0.0399\n",
+      "Epoch 1097/5000\n",
+      "919/919 - 3s - loss: 1.7017 - accuracy: 0.4524 - val_loss: 3.3472 - val_accuracy: 0.0400\n",
+      "Epoch 1098/5000\n",
+      "919/919 - 3s - loss: 1.7106 - accuracy: 0.4561 - val_loss: 3.3462 - val_accuracy: 0.0402\n",
+      "Epoch 1099/5000\n",
+      "919/919 - 3s - loss: 1.7148 - accuracy: 0.4531 - val_loss: 3.3445 - val_accuracy: 0.0402\n",
+      "Epoch 1100/5000\n",
+      "919/919 - 3s - loss: 1.7059 - accuracy: 0.4546 - val_loss: 3.3427 - val_accuracy: 0.0403\n",
+      "Epoch 1101/5000\n",
+      "919/919 - 3s - loss: 1.7062 - accuracy: 0.4552 - val_loss: 3.3428 - val_accuracy: 0.0401\n",
+      "Epoch 1102/5000\n",
+      "919/919 - 3s - loss: 1.6995 - accuracy: 0.4556 - val_loss: 3.3361 - val_accuracy: 0.0401\n",
+      "Epoch 1103/5000\n",
+      "919/919 - 3s - loss: 1.6969 - accuracy: 0.4560 - val_loss: 3.3337 - val_accuracy: 0.0401\n",
+      "Epoch 1104/5000\n",
+      "919/919 - 3s - loss: 1.7584 - accuracy: 0.4553 - val_loss: 3.3440 - val_accuracy: 0.0404\n",
+      "Epoch 1105/5000\n",
+      "919/919 - 3s - loss: 1.6890 - accuracy: 0.4569 - val_loss: 3.3436 - val_accuracy: 0.0403\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1106/5000\n",
+      "919/919 - 3s - loss: 1.6979 - accuracy: 0.4535 - val_loss: 3.3496 - val_accuracy: 0.0404\n",
+      "Epoch 1107/5000\n",
+      "919/919 - 3s - loss: 1.7397 - accuracy: 0.4536 - val_loss: 3.3518 - val_accuracy: 0.0399\n",
+      "Epoch 1108/5000\n",
+      "919/919 - 3s - loss: 1.7228 - accuracy: 0.4565 - val_loss: 3.3664 - val_accuracy: 0.0399\n",
+      "Epoch 1109/5000\n",
+      "919/919 - 3s - loss: 1.7142 - accuracy: 0.4556 - val_loss: 3.3689 - val_accuracy: 0.0405\n",
+      "Epoch 1110/5000\n",
+      "919/919 - 3s - loss: 1.7041 - accuracy: 0.4535 - val_loss: 3.3693 - val_accuracy: 0.0405\n",
+      "Epoch 1111/5000\n",
+      "919/919 - 3s - loss: 1.6833 - accuracy: 0.4582 - val_loss: 3.3762 - val_accuracy: 0.0406\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1112/5000\n",
+      "919/919 - 3s - loss: 1.6948 - accuracy: 0.4551 - val_loss: 3.3810 - val_accuracy: 0.0408\n",
+      "Epoch 1113/5000\n",
+      "919/919 - 3s - loss: 1.6891 - accuracy: 0.4574 - val_loss: 3.3887 - val_accuracy: 0.0411\n",
+      "Epoch 1114/5000\n",
+      "919/919 - 3s - loss: 1.6923 - accuracy: 0.4570 - val_loss: 3.3831 - val_accuracy: 0.0410\n",
+      "Epoch 1115/5000\n",
+      "919/919 - 3s - loss: 1.8059 - accuracy: 0.4559 - val_loss: 3.3803 - val_accuracy: 0.0407\n",
+      "Epoch 1116/5000\n",
+      "919/919 - 3s - loss: 1.6976 - accuracy: 0.4571 - val_loss: 3.3776 - val_accuracy: 0.0405\n",
+      "Epoch 1117/5000\n",
+      "919/919 - 3s - loss: 1.7047 - accuracy: 0.4548 - val_loss: 3.3753 - val_accuracy: 0.0405\n",
+      "Epoch 1118/5000\n",
+      "919/919 - 3s - loss: 1.7123 - accuracy: 0.4518 - val_loss: 3.3637 - val_accuracy: 0.0405\n",
+      "Epoch 1119/5000\n",
+      "919/919 - 3s - loss: 1.6952 - accuracy: 0.4578 - val_loss: 3.3737 - val_accuracy: 0.0408\n",
+      "Epoch 1120/5000\n",
+      "919/919 - 3s - loss: 1.7092 - accuracy: 0.4583 - val_loss: 3.3791 - val_accuracy: 0.0415\n",
+      "Epoch 1121/5000\n",
+      "919/919 - 3s - loss: 1.7071 - accuracy: 0.4546 - val_loss: 3.3663 - val_accuracy: 0.0409\n",
+      "Epoch 1122/5000\n",
+      "919/919 - 3s - loss: 1.6821 - accuracy: 0.4611 - val_loss: 3.3782 - val_accuracy: 0.0408\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1123/5000\n",
+      "919/919 - 3s - loss: 1.6801 - accuracy: 0.4604 - val_loss: 3.3768 - val_accuracy: 0.0408\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1124/5000\n",
+      "919/919 - 3s - loss: 1.7441 - accuracy: 0.4601 - val_loss: 3.3790 - val_accuracy: 0.0412\n",
+      "Epoch 1125/5000\n",
+      "919/919 - 3s - loss: 1.6898 - accuracy: 0.4596 - val_loss: 3.3670 - val_accuracy: 0.0410\n",
+      "Epoch 1126/5000\n",
+      "919/919 - 3s - loss: 1.6885 - accuracy: 0.4547 - val_loss: 3.3644 - val_accuracy: 0.0408\n",
+      "Epoch 1127/5000\n",
+      "919/919 - 3s - loss: 1.6866 - accuracy: 0.4578 - val_loss: 3.3719 - val_accuracy: 0.0410\n",
+      "Epoch 1128/5000\n",
+      "919/919 - 3s - loss: 1.6961 - accuracy: 0.4572 - val_loss: 3.3729 - val_accuracy: 0.0413\n",
+      "Epoch 1129/5000\n",
+      "919/919 - 3s - loss: 1.6900 - accuracy: 0.4595 - val_loss: 3.3944 - val_accuracy: 0.0411\n",
+      "Epoch 1130/5000\n",
+      "919/919 - 3s - loss: 1.6962 - accuracy: 0.4639 - val_loss: 3.3861 - val_accuracy: 0.0410\n",
+      "Epoch 1131/5000\n",
+      "919/919 - 3s - loss: 1.6951 - accuracy: 0.4582 - val_loss: 3.3847 - val_accuracy: 0.0410\n",
+      "Epoch 1132/5000\n",
+      "919/919 - 3s - loss: 1.6892 - accuracy: 0.4609 - val_loss: 3.3946 - val_accuracy: 0.0413\n",
+      "Epoch 1133/5000\n",
+      "919/919 - 3s - loss: 1.7007 - accuracy: 0.4618 - val_loss: 3.3924 - val_accuracy: 0.0412\n",
+      "Epoch 1134/5000\n",
+      "919/919 - 3s - loss: 1.7007 - accuracy: 0.4571 - val_loss: 3.4016 - val_accuracy: 0.0416\n",
+      "Epoch 1135/5000\n",
+      "919/919 - 3s - loss: 1.6903 - accuracy: 0.4574 - val_loss: 3.4059 - val_accuracy: 0.0414\n",
+      "Epoch 1136/5000\n",
+      "919/919 - 3s - loss: 1.6913 - accuracy: 0.4598 - val_loss: 3.3954 - val_accuracy: 0.0415\n",
+      "Epoch 1137/5000\n",
+      "919/919 - 3s - loss: 1.6864 - accuracy: 0.4602 - val_loss: 3.4069 - val_accuracy: 0.0420\n",
+      "Epoch 1138/5000\n",
+      "919/919 - 3s - loss: 1.7047 - accuracy: 0.4586 - val_loss: 3.4145 - val_accuracy: 0.0412\n",
+      "Epoch 1139/5000\n",
+      "919/919 - 3s - loss: 1.6845 - accuracy: 0.4618 - val_loss: 3.4068 - val_accuracy: 0.0414\n",
+      "Epoch 1140/5000\n",
+      "919/919 - 3s - loss: 1.6866 - accuracy: 0.4584 - val_loss: 3.3951 - val_accuracy: 0.0408\n",
+      "Epoch 1141/5000\n",
+      "919/919 - 3s - loss: 1.6928 - accuracy: 0.4602 - val_loss: 3.3960 - val_accuracy: 0.0408\n",
+      "Epoch 1142/5000\n",
+      "919/919 - 3s - loss: 1.7039 - accuracy: 0.4571 - val_loss: 3.3879 - val_accuracy: 0.0407\n",
+      "Epoch 1143/5000\n",
+      "919/919 - 3s - loss: 1.7011 - accuracy: 0.4582 - val_loss: 3.3948 - val_accuracy: 0.0411\n",
+      "Epoch 1144/5000\n",
+      "919/919 - 3s - loss: 1.7088 - accuracy: 0.4578 - val_loss: 3.3892 - val_accuracy: 0.0411\n",
+      "Epoch 1145/5000\n",
+      "919/919 - 3s - loss: 1.6899 - accuracy: 0.4616 - val_loss: 3.3929 - val_accuracy: 0.0411\n",
+      "Epoch 1146/5000\n",
+      "919/919 - 3s - loss: 1.6797 - accuracy: 0.4591 - val_loss: 3.3949 - val_accuracy: 0.0411\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1147/5000\n",
+      "919/919 - 3s - loss: 1.6883 - accuracy: 0.4603 - val_loss: 3.4006 - val_accuracy: 0.0407\n",
+      "Epoch 1148/5000\n",
+      "919/919 - 3s - loss: 1.7282 - accuracy: 0.4585 - val_loss: 3.4069 - val_accuracy: 0.0409\n",
+      "Epoch 1149/5000\n",
+      "919/919 - 3s - loss: 1.6944 - accuracy: 0.4617 - val_loss: 3.3981 - val_accuracy: 0.0412\n",
+      "Epoch 1150/5000\n",
+      "919/919 - 3s - loss: 1.7885 - accuracy: 0.4580 - val_loss: 3.3976 - val_accuracy: 0.0407\n",
+      "Epoch 1151/5000\n",
+      "919/919 - 3s - loss: 1.6917 - accuracy: 0.4620 - val_loss: 3.3939 - val_accuracy: 0.0411\n",
+      "Epoch 1152/5000\n",
+      "919/919 - 3s - loss: 1.6761 - accuracy: 0.4618 - val_loss: 3.3892 - val_accuracy: 0.0417\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1153/5000\n",
+      "919/919 - 3s - loss: 1.6827 - accuracy: 0.4611 - val_loss: 3.3894 - val_accuracy: 0.0416\n",
+      "Epoch 1154/5000\n",
+      "919/919 - 3s - loss: 1.6916 - accuracy: 0.4624 - val_loss: 3.3924 - val_accuracy: 0.0417\n",
+      "Epoch 1155/5000\n",
+      "919/919 - 3s - loss: 1.6861 - accuracy: 0.4605 - val_loss: 3.3962 - val_accuracy: 0.0417\n",
+      "Epoch 1156/5000\n",
+      "919/919 - 3s - loss: 1.6746 - accuracy: 0.4599 - val_loss: 3.3933 - val_accuracy: 0.0419\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1157/5000\n",
+      "919/919 - 3s - loss: 1.7033 - accuracy: 0.4629 - val_loss: 3.3917 - val_accuracy: 0.0415\n",
+      "Epoch 1158/5000\n",
+      "919/919 - 3s - loss: 1.6746 - accuracy: 0.4670 - val_loss: 3.4035 - val_accuracy: 0.0423\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1159/5000\n",
+      "919/919 - 3s - loss: 1.6874 - accuracy: 0.4635 - val_loss: 3.4020 - val_accuracy: 0.0424\n",
+      "Epoch 1160/5000\n",
+      "919/919 - 3s - loss: 1.8228 - accuracy: 0.4602 - val_loss: 3.3927 - val_accuracy: 0.0415\n",
+      "Epoch 1161/5000\n",
+      "919/919 - 3s - loss: 1.6884 - accuracy: 0.4601 - val_loss: 3.3984 - val_accuracy: 0.0417\n",
+      "Epoch 1162/5000\n",
+      "919/919 - 3s - loss: 1.6743 - accuracy: 0.4637 - val_loss: 3.4062 - val_accuracy: 0.0417\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1163/5000\n",
+      "919/919 - 4s - loss: 1.6941 - accuracy: 0.4613 - val_loss: 3.4174 - val_accuracy: 0.0415\n",
+      "Epoch 1164/5000\n",
+      "919/919 - 3s - loss: 1.6898 - accuracy: 0.4588 - val_loss: 3.4216 - val_accuracy: 0.0414\n",
+      "Epoch 1165/5000\n",
+      "919/919 - 3s - loss: 1.7321 - accuracy: 0.4649 - val_loss: 3.4237 - val_accuracy: 0.0417\n",
+      "Epoch 1166/5000\n",
+      "919/919 - 3s - loss: 1.7080 - accuracy: 0.4609 - val_loss: 3.4141 - val_accuracy: 0.0420\n",
+      "Epoch 1167/5000\n",
+      "919/919 - 3s - loss: 1.6840 - accuracy: 0.4630 - val_loss: 3.4049 - val_accuracy: 0.0423\n",
+      "Epoch 1168/5000\n",
+      "919/919 - 3s - loss: 1.6881 - accuracy: 0.4620 - val_loss: 3.4107 - val_accuracy: 0.0425\n",
+      "Epoch 1169/5000\n",
+      "919/919 - 3s - loss: 1.6882 - accuracy: 0.4607 - val_loss: 3.4119 - val_accuracy: 0.0420\n",
+      "Epoch 1170/5000\n",
+      "919/919 - 3s - loss: 1.6842 - accuracy: 0.4615 - val_loss: 3.3975 - val_accuracy: 0.0421\n",
+      "Epoch 1171/5000\n",
+      "919/919 - 3s - loss: 1.6845 - accuracy: 0.4639 - val_loss: 3.3949 - val_accuracy: 0.0420\n",
+      "Epoch 1172/5000\n",
+      "919/919 - 3s - loss: 1.6901 - accuracy: 0.4607 - val_loss: 3.3931 - val_accuracy: 0.0418\n",
+      "Epoch 1173/5000\n",
+      "919/919 - 3s - loss: 1.6907 - accuracy: 0.4624 - val_loss: 3.3840 - val_accuracy: 0.0420\n",
+      "Epoch 1174/5000\n",
+      "919/919 - 3s - loss: 1.6813 - accuracy: 0.4635 - val_loss: 3.3839 - val_accuracy: 0.0424\n",
+      "Epoch 1175/5000\n",
+      "919/919 - 3s - loss: 1.6745 - accuracy: 0.4661 - val_loss: 3.3818 - val_accuracy: 0.0426\n",
+      "Epoch 1176/5000\n",
+      "919/919 - 3s - loss: 1.6842 - accuracy: 0.4655 - val_loss: 3.3998 - val_accuracy: 0.0426\n",
+      "Epoch 1177/5000\n",
+      "919/919 - 3s - loss: 1.6821 - accuracy: 0.4618 - val_loss: 3.3980 - val_accuracy: 0.0426\n",
+      "Epoch 1178/5000\n",
+      "919/919 - 3s - loss: 1.6766 - accuracy: 0.4640 - val_loss: 3.3972 - val_accuracy: 0.0425\n",
+      "Epoch 1179/5000\n",
+      "919/919 - 3s - loss: 1.6731 - accuracy: 0.4652 - val_loss: 3.4018 - val_accuracy: 0.0425\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1180/5000\n",
+      "919/919 - 3s - loss: 1.6755 - accuracy: 0.4637 - val_loss: 3.4092 - val_accuracy: 0.0424\n",
+      "Epoch 1181/5000\n",
+      "919/919 - 3s - loss: 1.6891 - accuracy: 0.4603 - val_loss: 3.4169 - val_accuracy: 0.0422\n",
+      "Epoch 1182/5000\n",
+      "919/919 - 3s - loss: 1.7345 - accuracy: 0.4641 - val_loss: 3.4338 - val_accuracy: 0.0421\n",
+      "Epoch 1183/5000\n",
+      "919/919 - 3s - loss: 1.6774 - accuracy: 0.4669 - val_loss: 3.4449 - val_accuracy: 0.0414\n",
+      "Epoch 1184/5000\n",
+      "919/919 - 3s - loss: 1.6940 - accuracy: 0.4608 - val_loss: 3.4378 - val_accuracy: 0.0412\n",
+      "Epoch 1185/5000\n",
+      "919/919 - 3s - loss: 1.6701 - accuracy: 0.4614 - val_loss: 3.4261 - val_accuracy: 0.0426\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1186/5000\n",
+      "919/919 - 3s - loss: 1.6724 - accuracy: 0.4641 - val_loss: 3.4174 - val_accuracy: 0.0425\n",
+      "Epoch 1187/5000\n",
+      "919/919 - 3s - loss: 1.6906 - accuracy: 0.4610 - val_loss: 3.4028 - val_accuracy: 0.0422\n",
+      "Epoch 1188/5000\n",
+      "919/919 - 3s - loss: 1.6648 - accuracy: 0.4631 - val_loss: 3.4121 - val_accuracy: 0.0419\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1189/5000\n",
+      "919/919 - 3s - loss: 1.6827 - accuracy: 0.4667 - val_loss: 3.4174 - val_accuracy: 0.0423\n",
+      "Epoch 1190/5000\n",
+      "919/919 - 3s - loss: 1.6739 - accuracy: 0.4657 - val_loss: 3.4229 - val_accuracy: 0.0416\n",
+      "Epoch 1191/5000\n",
+      "919/919 - 3s - loss: 1.6778 - accuracy: 0.4621 - val_loss: 3.4193 - val_accuracy: 0.0417\n",
+      "Epoch 1192/5000\n",
+      "919/919 - 3s - loss: 1.6707 - accuracy: 0.4669 - val_loss: 3.4107 - val_accuracy: 0.0415\n",
+      "Epoch 1193/5000\n",
+      "919/919 - 3s - loss: 1.6813 - accuracy: 0.4668 - val_loss: 3.4051 - val_accuracy: 0.0414\n",
+      "Epoch 1194/5000\n",
+      "919/919 - 3s - loss: 1.6689 - accuracy: 0.4680 - val_loss: 3.3944 - val_accuracy: 0.0415\n",
+      "Epoch 1195/5000\n",
+      "919/919 - 3s - loss: 1.6727 - accuracy: 0.4637 - val_loss: 3.3979 - val_accuracy: 0.0417\n",
+      "Epoch 1196/5000\n",
+      "919/919 - 3s - loss: 1.6726 - accuracy: 0.4648 - val_loss: 3.3992 - val_accuracy: 0.0421\n",
+      "Epoch 1197/5000\n",
+      "919/919 - 3s - loss: 1.6723 - accuracy: 0.4630 - val_loss: 3.3986 - val_accuracy: 0.0423\n",
+      "Epoch 1198/5000\n",
+      "919/919 - 3s - loss: 1.6740 - accuracy: 0.4653 - val_loss: 3.4017 - val_accuracy: 0.0424\n",
+      "Epoch 1199/5000\n",
+      "919/919 - 3s - loss: 1.7754 - accuracy: 0.4641 - val_loss: 3.4060 - val_accuracy: 0.0421\n",
+      "Epoch 1200/5000\n",
+      "919/919 - 3s - loss: 1.6707 - accuracy: 0.4690 - val_loss: 3.4129 - val_accuracy: 0.0424\n",
+      "Epoch 1201/5000\n",
+      "919/919 - 3s - loss: 1.7476 - accuracy: 0.4634 - val_loss: 3.4165 - val_accuracy: 0.0424\n",
+      "Epoch 1202/5000\n",
+      "919/919 - 3s - loss: 1.6712 - accuracy: 0.4618 - val_loss: 3.4123 - val_accuracy: 0.0426\n",
+      "Epoch 1203/5000\n",
+      "919/919 - 3s - loss: 1.6945 - accuracy: 0.4652 - val_loss: 3.4133 - val_accuracy: 0.0428\n",
+      "Epoch 1204/5000\n",
+      "919/919 - 3s - loss: 1.6613 - accuracy: 0.4637 - val_loss: 3.4259 - val_accuracy: 0.0425\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1205/5000\n",
+      "919/919 - 3s - loss: 1.7897 - accuracy: 0.4668 - val_loss: 3.4163 - val_accuracy: 0.0417\n",
+      "Epoch 1206/5000\n",
+      "919/919 - 3s - loss: 1.6717 - accuracy: 0.4654 - val_loss: 3.4199 - val_accuracy: 0.0425\n",
+      "Epoch 1207/5000\n",
+      "919/919 - 3s - loss: 1.6688 - accuracy: 0.4672 - val_loss: 3.4222 - val_accuracy: 0.0417\n",
+      "Epoch 1208/5000\n",
+      "919/919 - 3s - loss: 1.6607 - accuracy: 0.4663 - val_loss: 3.4316 - val_accuracy: 0.0419\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1209/5000\n",
+      "919/919 - 3s - loss: 1.6816 - accuracy: 0.4655 - val_loss: 3.4493 - val_accuracy: 0.0420\n",
+      "Epoch 1210/5000\n",
+      "919/919 - 3s - loss: 1.6779 - accuracy: 0.4670 - val_loss: 3.4329 - val_accuracy: 0.0417\n",
+      "Epoch 1211/5000\n",
+      "919/919 - 3s - loss: 1.6785 - accuracy: 0.4671 - val_loss: 3.4198 - val_accuracy: 0.0419\n",
+      "Epoch 1212/5000\n",
+      "919/919 - 3s - loss: 1.6608 - accuracy: 0.4684 - val_loss: 3.4275 - val_accuracy: 0.0427\n",
+      "Epoch 1213/5000\n",
+      "919/919 - 3s - loss: 1.6711 - accuracy: 0.4654 - val_loss: 3.4320 - val_accuracy: 0.0424\n",
+      "Epoch 1214/5000\n",
+      "919/919 - 3s - loss: 1.6882 - accuracy: 0.4676 - val_loss: 3.4296 - val_accuracy: 0.0422\n",
+      "Epoch 1215/5000\n",
+      "919/919 - 3s - loss: 1.6642 - accuracy: 0.4683 - val_loss: 3.4400 - val_accuracy: 0.0428\n",
+      "Epoch 1216/5000\n",
+      "919/919 - 3s - loss: 1.7020 - accuracy: 0.4664 - val_loss: 3.4192 - val_accuracy: 0.0426\n",
+      "Epoch 1217/5000\n",
+      "919/919 - 3s - loss: 1.6791 - accuracy: 0.4685 - val_loss: 3.4169 - val_accuracy: 0.0423\n",
+      "Epoch 1218/5000\n",
+      "919/919 - 3s - loss: 1.6829 - accuracy: 0.4665 - val_loss: 3.4144 - val_accuracy: 0.0426\n",
+      "Epoch 1219/5000\n",
+      "919/919 - 3s - loss: 1.6691 - accuracy: 0.4679 - val_loss: 3.4019 - val_accuracy: 0.0427\n",
+      "Epoch 1220/5000\n",
+      "919/919 - 3s - loss: 1.6628 - accuracy: 0.4667 - val_loss: 3.3983 - val_accuracy: 0.0428\n",
+      "Epoch 1221/5000\n",
+      "919/919 - 3s - loss: 1.6727 - accuracy: 0.4645 - val_loss: 3.4006 - val_accuracy: 0.0426\n",
+      "Epoch 1222/5000\n",
+      "919/919 - 3s - loss: 1.6747 - accuracy: 0.4660 - val_loss: 3.4007 - val_accuracy: 0.0424\n",
+      "Epoch 1223/5000\n",
+      "919/919 - 3s - loss: 1.6750 - accuracy: 0.4644 - val_loss: 3.4056 - val_accuracy: 0.0426\n",
+      "Epoch 1224/5000\n",
+      "919/919 - 3s - loss: 1.6731 - accuracy: 0.4667 - val_loss: 3.4033 - val_accuracy: 0.0422\n",
+      "Epoch 1225/5000\n",
+      "919/919 - 3s - loss: 1.7228 - accuracy: 0.4670 - val_loss: 3.4148 - val_accuracy: 0.0432\n",
+      "Epoch 1226/5000\n",
+      "919/919 - 3s - loss: 1.6633 - accuracy: 0.4705 - val_loss: 3.4160 - val_accuracy: 0.0424\n",
+      "Epoch 1227/5000\n",
+      "919/919 - 3s - loss: 1.6681 - accuracy: 0.4661 - val_loss: 3.4191 - val_accuracy: 0.0429\n",
+      "Epoch 1228/5000\n",
+      "919/919 - 3s - loss: 1.6561 - accuracy: 0.4667 - val_loss: 3.4212 - val_accuracy: 0.0427\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1229/5000\n",
+      "919/919 - 3s - loss: 1.6692 - accuracy: 0.4699 - val_loss: 3.4178 - val_accuracy: 0.0426\n",
+      "Epoch 1230/5000\n",
+      "919/919 - 3s - loss: 1.6817 - accuracy: 0.4646 - val_loss: 3.4116 - val_accuracy: 0.0426\n",
+      "Epoch 1231/5000\n",
+      "919/919 - 3s - loss: 1.6604 - accuracy: 0.4711 - val_loss: 3.4175 - val_accuracy: 0.0433\n",
+      "Epoch 1232/5000\n",
+      "919/919 - 3s - loss: 1.6682 - accuracy: 0.4669 - val_loss: 3.4094 - val_accuracy: 0.0433\n",
+      "Epoch 1233/5000\n",
+      "919/919 - 3s - loss: 1.6678 - accuracy: 0.4690 - val_loss: 3.4140 - val_accuracy: 0.0427\n",
+      "Epoch 1234/5000\n",
+      "919/919 - 3s - loss: 1.6758 - accuracy: 0.4658 - val_loss: 3.4135 - val_accuracy: 0.0426\n",
+      "Epoch 1235/5000\n",
+      "919/919 - 3s - loss: 1.6631 - accuracy: 0.4697 - val_loss: 3.4292 - val_accuracy: 0.0427\n",
+      "Epoch 1236/5000\n",
+      "919/919 - 3s - loss: 1.6720 - accuracy: 0.4701 - val_loss: 3.4111 - val_accuracy: 0.0427\n",
+      "Epoch 1237/5000\n",
+      "919/919 - 3s - loss: 1.7050 - accuracy: 0.4678 - val_loss: 3.4292 - val_accuracy: 0.0434\n",
+      "Epoch 1238/5000\n",
+      "919/919 - 3s - loss: 1.6677 - accuracy: 0.4638 - val_loss: 3.4296 - val_accuracy: 0.0431\n",
+      "Epoch 1239/5000\n",
+      "919/919 - 3s - loss: 1.6599 - accuracy: 0.4694 - val_loss: 3.4373 - val_accuracy: 0.0431\n",
+      "Epoch 1240/5000\n",
+      "919/919 - 3s - loss: 1.7101 - accuracy: 0.4708 - val_loss: 3.4377 - val_accuracy: 0.0427\n",
+      "Epoch 1241/5000\n",
+      "919/919 - 3s - loss: 1.6640 - accuracy: 0.4684 - val_loss: 3.4525 - val_accuracy: 0.0435\n",
+      "Epoch 1242/5000\n",
+      "919/919 - 3s - loss: 1.6742 - accuracy: 0.4710 - val_loss: 3.4569 - val_accuracy: 0.0435\n",
+      "Epoch 1243/5000\n",
+      "919/919 - 3s - loss: 1.6654 - accuracy: 0.4676 - val_loss: 3.4441 - val_accuracy: 0.0424\n",
+      "Epoch 1244/5000\n",
+      "919/919 - 3s - loss: 1.6737 - accuracy: 0.4660 - val_loss: 3.4431 - val_accuracy: 0.0428\n",
+      "Epoch 1245/5000\n",
+      "919/919 - 3s - loss: 1.6609 - accuracy: 0.4714 - val_loss: 3.4271 - val_accuracy: 0.0432\n",
+      "Epoch 1246/5000\n",
+      "919/919 - 3s - loss: 1.6674 - accuracy: 0.4670 - val_loss: 3.4205 - val_accuracy: 0.0432\n",
+      "Epoch 1247/5000\n",
+      "919/919 - 3s - loss: 1.6561 - accuracy: 0.4746 - val_loss: 3.4281 - val_accuracy: 0.0434\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1248/5000\n",
+      "919/919 - 3s - loss: 1.6614 - accuracy: 0.4692 - val_loss: 3.4334 - val_accuracy: 0.0436\n",
+      "Epoch 1249/5000\n",
+      "919/919 - 3s - loss: 1.6551 - accuracy: 0.4706 - val_loss: 3.4352 - val_accuracy: 0.0440\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1250/5000\n",
+      "919/919 - 3s - loss: 1.6621 - accuracy: 0.4695 - val_loss: 3.4363 - val_accuracy: 0.0439\n",
+      "Epoch 1251/5000\n",
+      "919/919 - 3s - loss: 1.6732 - accuracy: 0.4690 - val_loss: 3.4262 - val_accuracy: 0.0437\n",
+      "Epoch 1252/5000\n",
+      "919/919 - 3s - loss: 1.6590 - accuracy: 0.4717 - val_loss: 3.4312 - val_accuracy: 0.0438\n",
+      "Epoch 1253/5000\n",
+      "919/919 - 3s - loss: 1.6543 - accuracy: 0.4715 - val_loss: 3.4338 - val_accuracy: 0.0437\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1254/5000\n",
+      "919/919 - 3s - loss: 1.6834 - accuracy: 0.4724 - val_loss: 3.4227 - val_accuracy: 0.0438\n",
+      "Epoch 1255/5000\n",
+      "919/919 - 3s - loss: 1.6618 - accuracy: 0.4742 - val_loss: 3.4277 - val_accuracy: 0.0440\n",
+      "Epoch 1256/5000\n",
+      "919/919 - 3s - loss: 1.6627 - accuracy: 0.4714 - val_loss: 3.4379 - val_accuracy: 0.0438\n",
+      "Epoch 1257/5000\n",
+      "919/919 - 3s - loss: 1.6591 - accuracy: 0.4688 - val_loss: 3.4373 - val_accuracy: 0.0439\n",
+      "Epoch 1258/5000\n",
+      "919/919 - 3s - loss: 1.6645 - accuracy: 0.4691 - val_loss: 3.4329 - val_accuracy: 0.0438\n",
+      "Epoch 1259/5000\n",
+      "919/919 - 3s - loss: 1.6955 - accuracy: 0.4707 - val_loss: 3.4331 - val_accuracy: 0.0446\n",
+      "Epoch 1260/5000\n",
+      "919/919 - 3s - loss: 1.6660 - accuracy: 0.4690 - val_loss: 3.4231 - val_accuracy: 0.0446\n",
+      "Epoch 1261/5000\n",
+      "919/919 - 3s - loss: 1.6744 - accuracy: 0.4727 - val_loss: 3.4198 - val_accuracy: 0.0443\n",
+      "Epoch 1262/5000\n",
+      "919/919 - 3s - loss: 1.6624 - accuracy: 0.4775 - val_loss: 3.4165 - val_accuracy: 0.0444\n",
+      "Epoch 1263/5000\n",
+      "919/919 - 3s - loss: 1.6866 - accuracy: 0.4769 - val_loss: 3.4144 - val_accuracy: 0.0447\n",
+      "Epoch 1264/5000\n",
+      "919/919 - 3s - loss: 1.6463 - accuracy: 0.4741 - val_loss: 3.4254 - val_accuracy: 0.0447\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1265/5000\n",
+      "919/919 - 3s - loss: 1.6548 - accuracy: 0.4705 - val_loss: 3.4312 - val_accuracy: 0.0444\n",
+      "Epoch 1266/5000\n",
+      "919/919 - 3s - loss: 1.6444 - accuracy: 0.4759 - val_loss: 3.4459 - val_accuracy: 0.0449\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1267/5000\n",
+      "919/919 - 3s - loss: 1.6670 - accuracy: 0.4680 - val_loss: 3.4432 - val_accuracy: 0.0446\n",
+      "Epoch 1268/5000\n",
+      "919/919 - 3s - loss: 1.6680 - accuracy: 0.4716 - val_loss: 3.4226 - val_accuracy: 0.0444\n",
+      "Epoch 1269/5000\n",
+      "919/919 - 3s - loss: 1.6653 - accuracy: 0.4701 - val_loss: 3.4120 - val_accuracy: 0.0446\n",
+      "Epoch 1270/5000\n",
+      "919/919 - 3s - loss: 1.6719 - accuracy: 0.4715 - val_loss: 3.4174 - val_accuracy: 0.0444\n",
+      "Epoch 1271/5000\n",
+      "919/919 - 3s - loss: 1.6573 - accuracy: 0.4722 - val_loss: 3.4278 - val_accuracy: 0.0442\n",
+      "Epoch 1272/5000\n",
+      "919/919 - 3s - loss: 1.6464 - accuracy: 0.4720 - val_loss: 3.4342 - val_accuracy: 0.0448\n",
+      "Epoch 1273/5000\n",
+      "919/919 - 3s - loss: 1.6620 - accuracy: 0.4714 - val_loss: 3.4420 - val_accuracy: 0.0448\n",
+      "Epoch 1274/5000\n",
+      "919/919 - 3s - loss: 1.6435 - accuracy: 0.4748 - val_loss: 3.4460 - val_accuracy: 0.0444\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1275/5000\n",
+      "919/919 - 3s - loss: 1.6753 - accuracy: 0.4718 - val_loss: 3.4458 - val_accuracy: 0.0447\n",
+      "Epoch 1276/5000\n",
+      "919/919 - 3s - loss: 1.6484 - accuracy: 0.4723 - val_loss: 3.4417 - val_accuracy: 0.0450\n",
+      "Epoch 1277/5000\n",
+      "919/919 - 3s - loss: 1.7296 - accuracy: 0.4731 - val_loss: 3.4226 - val_accuracy: 0.0448\n",
+      "Epoch 1278/5000\n",
+      "919/919 - 3s - loss: 1.6593 - accuracy: 0.4699 - val_loss: 3.4271 - val_accuracy: 0.0445\n",
+      "Epoch 1279/5000\n",
+      "919/919 - 3s - loss: 1.6390 - accuracy: 0.4739 - val_loss: 3.4283 - val_accuracy: 0.0441\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1280/5000\n",
+      "919/919 - 3s - loss: 1.6536 - accuracy: 0.4729 - val_loss: 3.4352 - val_accuracy: 0.0442\n",
+      "Epoch 1281/5000\n",
+      "919/919 - 3s - loss: 1.6633 - accuracy: 0.4722 - val_loss: 3.4541 - val_accuracy: 0.0440\n",
+      "Epoch 1282/5000\n",
+      "919/919 - 3s - loss: 1.6517 - accuracy: 0.4723 - val_loss: 3.4574 - val_accuracy: 0.0436\n",
+      "Epoch 1283/5000\n",
+      "919/919 - 3s - loss: 1.6585 - accuracy: 0.4709 - val_loss: 3.4560 - val_accuracy: 0.0444\n",
+      "Epoch 1284/5000\n",
+      "919/919 - 3s - loss: 1.6441 - accuracy: 0.4754 - val_loss: 3.4539 - val_accuracy: 0.0447\n",
+      "Epoch 1285/5000\n",
+      "919/919 - 3s - loss: 1.6705 - accuracy: 0.4722 - val_loss: 3.4613 - val_accuracy: 0.0448\n",
+      "Epoch 1286/5000\n",
+      "919/919 - 3s - loss: 1.6552 - accuracy: 0.4750 - val_loss: 3.4572 - val_accuracy: 0.0445\n",
+      "Epoch 1287/5000\n",
+      "919/919 - 3s - loss: 1.6422 - accuracy: 0.4741 - val_loss: 3.4652 - val_accuracy: 0.0450\n",
+      "Epoch 1288/5000\n",
+      "919/919 - 3s - loss: 1.6583 - accuracy: 0.4750 - val_loss: 3.4703 - val_accuracy: 0.0448\n",
+      "Epoch 1289/5000\n",
+      "919/919 - 3s - loss: 1.6669 - accuracy: 0.4705 - val_loss: 3.4697 - val_accuracy: 0.0447\n",
+      "Epoch 1290/5000\n",
+      "919/919 - 3s - loss: 1.6472 - accuracy: 0.4761 - val_loss: 3.4714 - val_accuracy: 0.0447\n",
+      "Epoch 1291/5000\n",
+      "919/919 - 3s - loss: 1.6892 - accuracy: 0.4731 - val_loss: 3.4667 - val_accuracy: 0.0447\n",
+      "Epoch 1292/5000\n",
+      "919/919 - 3s - loss: 1.6517 - accuracy: 0.4727 - val_loss: 3.4625 - val_accuracy: 0.0444\n",
+      "Epoch 1293/5000\n",
+      "919/919 - 3s - loss: 1.6518 - accuracy: 0.4754 - val_loss: 3.4605 - val_accuracy: 0.0447\n",
+      "Epoch 1294/5000\n",
+      "919/919 - 3s - loss: 1.7304 - accuracy: 0.4720 - val_loss: 3.4564 - val_accuracy: 0.0450\n",
+      "Epoch 1295/5000\n",
+      "919/919 - 3s - loss: 1.6998 - accuracy: 0.4699 - val_loss: 3.4422 - val_accuracy: 0.0448\n",
+      "Epoch 1296/5000\n",
+      "919/919 - 3s - loss: 1.7370 - accuracy: 0.4749 - val_loss: 3.4460 - val_accuracy: 0.0448\n",
+      "Epoch 1297/5000\n",
+      "919/919 - 3s - loss: 1.6650 - accuracy: 0.4744 - val_loss: 3.4508 - val_accuracy: 0.0449\n",
+      "Epoch 1298/5000\n",
+      "919/919 - 3s - loss: 1.6770 - accuracy: 0.4731 - val_loss: 3.4506 - val_accuracy: 0.0453\n",
+      "Epoch 1299/5000\n",
+      "919/919 - 3s - loss: 1.6564 - accuracy: 0.4759 - val_loss: 3.4532 - val_accuracy: 0.0453\n",
+      "Epoch 1300/5000\n",
+      "919/919 - 3s - loss: 1.6530 - accuracy: 0.4755 - val_loss: 3.4493 - val_accuracy: 0.0451\n",
+      "Epoch 1301/5000\n",
+      "919/919 - 3s - loss: 1.6390 - accuracy: 0.4746 - val_loss: 3.4606 - val_accuracy: 0.0450\n",
+      "Epoch 1302/5000\n",
+      "919/919 - 3s - loss: 1.6514 - accuracy: 0.4783 - val_loss: 3.4463 - val_accuracy: 0.0448\n",
+      "Epoch 1303/5000\n",
+      "919/919 - 3s - loss: 1.6547 - accuracy: 0.4754 - val_loss: 3.4523 - val_accuracy: 0.0449\n",
+      "Epoch 1304/5000\n",
+      "919/919 - 3s - loss: 1.6462 - accuracy: 0.4741 - val_loss: 3.4434 - val_accuracy: 0.0448\n",
+      "Epoch 1305/5000\n",
+      "919/919 - 3s - loss: 1.6525 - accuracy: 0.4746 - val_loss: 3.4644 - val_accuracy: 0.0450\n",
+      "Epoch 1306/5000\n",
+      "919/919 - 3s - loss: 1.6462 - accuracy: 0.4743 - val_loss: 3.4745 - val_accuracy: 0.0450\n",
+      "Epoch 1307/5000\n",
+      "919/919 - 3s - loss: 1.6813 - accuracy: 0.4770 - val_loss: 3.4739 - val_accuracy: 0.0451\n",
+      "Epoch 1308/5000\n",
+      "919/919 - 3s - loss: 1.6615 - accuracy: 0.4755 - val_loss: 3.4692 - val_accuracy: 0.0451\n",
+      "Epoch 1309/5000\n",
+      "919/919 - 3s - loss: 1.6445 - accuracy: 0.4774 - val_loss: 3.4649 - val_accuracy: 0.0451\n",
+      "Epoch 1310/5000\n",
+      "919/919 - 3s - loss: 1.6387 - accuracy: 0.4749 - val_loss: 3.4551 - val_accuracy: 0.0448\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1311/5000\n",
+      "919/919 - 3s - loss: 1.6334 - accuracy: 0.4766 - val_loss: 3.4619 - val_accuracy: 0.0452\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1312/5000\n",
+      "919/919 - 3s - loss: 1.6495 - accuracy: 0.4724 - val_loss: 3.4608 - val_accuracy: 0.0453\n",
+      "Epoch 1313/5000\n",
+      "919/919 - 3s - loss: 1.6549 - accuracy: 0.4753 - val_loss: 3.4659 - val_accuracy: 0.0455\n",
+      "Epoch 1314/5000\n",
+      "919/919 - 3s - loss: 1.6451 - accuracy: 0.4782 - val_loss: 3.4674 - val_accuracy: 0.0457\n",
+      "Epoch 1315/5000\n",
+      "919/919 - 3s - loss: 1.6437 - accuracy: 0.4791 - val_loss: 3.4787 - val_accuracy: 0.0455\n",
+      "Epoch 1316/5000\n",
+      "919/919 - 3s - loss: 1.7356 - accuracy: 0.4714 - val_loss: 3.4704 - val_accuracy: 0.0455\n",
+      "Epoch 1317/5000\n",
+      "919/919 - 3s - loss: 1.6458 - accuracy: 0.4765 - val_loss: 3.4744 - val_accuracy: 0.0452\n",
+      "Epoch 1318/5000\n",
+      "919/919 - 3s - loss: 1.6464 - accuracy: 0.4752 - val_loss: 3.4610 - val_accuracy: 0.0452\n",
+      "Epoch 1319/5000\n",
+      "919/919 - 3s - loss: 1.6471 - accuracy: 0.4786 - val_loss: 3.4752 - val_accuracy: 0.0453\n",
+      "Epoch 1320/5000\n",
+      "919/919 - 3s - loss: 1.6312 - accuracy: 0.4780 - val_loss: 3.4678 - val_accuracy: 0.0451\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1321/5000\n",
+      "919/919 - 3s - loss: 1.6485 - accuracy: 0.4768 - val_loss: 3.4722 - val_accuracy: 0.0453\n",
+      "Epoch 1322/5000\n",
+      "919/919 - 3s - loss: 1.6415 - accuracy: 0.4737 - val_loss: 3.4848 - val_accuracy: 0.0455\n",
+      "Epoch 1323/5000\n",
+      "919/919 - 3s - loss: 1.6545 - accuracy: 0.4773 - val_loss: 3.4673 - val_accuracy: 0.0457\n",
+      "Epoch 1324/5000\n",
+      "919/919 - 3s - loss: 1.6382 - accuracy: 0.4786 - val_loss: 3.4741 - val_accuracy: 0.0455\n",
+      "Epoch 1325/5000\n",
+      "919/919 - 3s - loss: 1.6551 - accuracy: 0.4770 - val_loss: 3.4726 - val_accuracy: 0.0456\n",
+      "Epoch 1326/5000\n",
+      "919/919 - 3s - loss: 1.6483 - accuracy: 0.4724 - val_loss: 3.4862 - val_accuracy: 0.0452\n",
+      "Epoch 1327/5000\n",
+      "919/919 - 3s - loss: 1.6317 - accuracy: 0.4745 - val_loss: 3.4959 - val_accuracy: 0.0454\n",
+      "Epoch 1328/5000\n",
+      "919/919 - 3s - loss: 1.6439 - accuracy: 0.4756 - val_loss: 3.4820 - val_accuracy: 0.0454\n",
+      "Epoch 1329/5000\n",
+      "919/919 - 3s - loss: 1.6690 - accuracy: 0.4761 - val_loss: 3.4742 - val_accuracy: 0.0453\n",
+      "Epoch 1330/5000\n",
+      "919/919 - 3s - loss: 1.6354 - accuracy: 0.4751 - val_loss: 3.4886 - val_accuracy: 0.0456\n",
+      "Epoch 1331/5000\n",
+      "919/919 - 3s - loss: 1.6560 - accuracy: 0.4746 - val_loss: 3.4767 - val_accuracy: 0.0459\n",
+      "Epoch 1332/5000\n",
+      "919/919 - 3s - loss: 1.6375 - accuracy: 0.4786 - val_loss: 3.4926 - val_accuracy: 0.0459\n",
+      "Epoch 1333/5000\n",
+      "919/919 - 3s - loss: 1.6367 - accuracy: 0.4781 - val_loss: 3.5005 - val_accuracy: 0.0461\n",
+      "Epoch 1334/5000\n",
+      "919/919 - 3s - loss: 1.6417 - accuracy: 0.4786 - val_loss: 3.5010 - val_accuracy: 0.0462\n",
+      "Epoch 1335/5000\n",
+      "919/919 - 3s - loss: 1.7826 - accuracy: 0.4806 - val_loss: 3.4916 - val_accuracy: 0.0461\n",
+      "Epoch 1336/5000\n",
+      "919/919 - 3s - loss: 1.6486 - accuracy: 0.4733 - val_loss: 3.4894 - val_accuracy: 0.0456\n",
+      "Epoch 1337/5000\n",
+      "919/919 - 3s - loss: 1.6437 - accuracy: 0.4774 - val_loss: 3.4805 - val_accuracy: 0.0456\n",
+      "Epoch 1338/5000\n",
+      "919/919 - 3s - loss: 1.6325 - accuracy: 0.4756 - val_loss: 3.4852 - val_accuracy: 0.0456\n",
+      "Epoch 1339/5000\n",
+      "919/919 - 3s - loss: 1.6445 - accuracy: 0.4766 - val_loss: 3.4789 - val_accuracy: 0.0456\n",
+      "Epoch 1340/5000\n",
+      "919/919 - 3s - loss: 1.7108 - accuracy: 0.4786 - val_loss: 3.4884 - val_accuracy: 0.0453\n",
+      "Epoch 1341/5000\n",
+      "919/919 - 3s - loss: 1.6400 - accuracy: 0.4796 - val_loss: 3.4823 - val_accuracy: 0.0453\n",
+      "Epoch 1342/5000\n",
+      "919/919 - 3s - loss: 1.6366 - accuracy: 0.4765 - val_loss: 3.4866 - val_accuracy: 0.0454\n",
+      "Epoch 1343/5000\n",
+      "919/919 - 3s - loss: 1.6444 - accuracy: 0.4768 - val_loss: 3.4839 - val_accuracy: 0.0454\n",
+      "Epoch 1344/5000\n",
+      "919/919 - 3s - loss: 1.6473 - accuracy: 0.4801 - val_loss: 3.4988 - val_accuracy: 0.0456\n",
+      "Epoch 1345/5000\n",
+      "919/919 - 3s - loss: 1.6416 - accuracy: 0.4780 - val_loss: 3.4943 - val_accuracy: 0.0453\n",
+      "Epoch 1346/5000\n",
+      "919/919 - 3s - loss: 1.7197 - accuracy: 0.4756 - val_loss: 3.4910 - val_accuracy: 0.0457\n",
+      "Epoch 1347/5000\n",
+      "919/919 - 3s - loss: 1.6423 - accuracy: 0.4777 - val_loss: 3.4888 - val_accuracy: 0.0455\n",
+      "Epoch 1348/5000\n",
+      "919/919 - 3s - loss: 1.6431 - accuracy: 0.4756 - val_loss: 3.4804 - val_accuracy: 0.0462\n",
+      "Epoch 1349/5000\n",
+      "919/919 - 3s - loss: 1.6359 - accuracy: 0.4784 - val_loss: 3.4691 - val_accuracy: 0.0455\n",
+      "Epoch 1350/5000\n",
+      "919/919 - 3s - loss: 1.6205 - accuracy: 0.4782 - val_loss: 3.4681 - val_accuracy: 0.0454\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1351/5000\n",
+      "919/919 - 3s - loss: 1.6354 - accuracy: 0.4809 - val_loss: 3.4703 - val_accuracy: 0.0451\n",
+      "Epoch 1352/5000\n",
+      "919/919 - 3s - loss: 1.6351 - accuracy: 0.4813 - val_loss: 3.4824 - val_accuracy: 0.0455\n",
+      "Epoch 1353/5000\n",
+      "919/919 - 3s - loss: 1.6516 - accuracy: 0.4786 - val_loss: 3.4717 - val_accuracy: 0.0453\n",
+      "Epoch 1354/5000\n",
+      "919/919 - 3s - loss: 1.6365 - accuracy: 0.4799 - val_loss: 3.4729 - val_accuracy: 0.0457\n",
+      "Epoch 1355/5000\n",
+      "919/919 - 3s - loss: 1.6657 - accuracy: 0.4803 - val_loss: 3.4645 - val_accuracy: 0.0451\n",
+      "Epoch 1356/5000\n",
+      "919/919 - 3s - loss: 1.6505 - accuracy: 0.4751 - val_loss: 3.4663 - val_accuracy: 0.0451\n",
+      "Epoch 1357/5000\n",
+      "919/919 - 3s - loss: 1.6321 - accuracy: 0.4798 - val_loss: 3.4709 - val_accuracy: 0.0455\n",
+      "Epoch 1358/5000\n",
+      "919/919 - 3s - loss: 1.6274 - accuracy: 0.4799 - val_loss: 3.4671 - val_accuracy: 0.0450\n",
+      "Epoch 1359/5000\n",
+      "919/919 - 3s - loss: 1.6412 - accuracy: 0.4758 - val_loss: 3.4451 - val_accuracy: 0.0453\n",
+      "Epoch 1360/5000\n",
+      "919/919 - 3s - loss: 1.6481 - accuracy: 0.4782 - val_loss: 3.4503 - val_accuracy: 0.0453\n",
+      "Epoch 1361/5000\n",
+      "919/919 - 3s - loss: 1.6224 - accuracy: 0.4788 - val_loss: 3.4630 - val_accuracy: 0.0457\n",
+      "Epoch 1362/5000\n",
+      "919/919 - 3s - loss: 1.6343 - accuracy: 0.4793 - val_loss: 3.4670 - val_accuracy: 0.0456\n",
+      "Epoch 1363/5000\n",
+      "919/919 - 3s - loss: 1.6230 - accuracy: 0.4807 - val_loss: 3.4678 - val_accuracy: 0.0458\n",
+      "Epoch 1364/5000\n",
+      "919/919 - 3s - loss: 1.6122 - accuracy: 0.4813 - val_loss: 3.4790 - val_accuracy: 0.0458\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1365/5000\n",
+      "919/919 - 3s - loss: 1.6307 - accuracy: 0.4806 - val_loss: 3.4664 - val_accuracy: 0.0454\n",
+      "Epoch 1366/5000\n",
+      "919/919 - 3s - loss: 1.6481 - accuracy: 0.4768 - val_loss: 3.4619 - val_accuracy: 0.0454\n",
+      "Epoch 1367/5000\n",
+      "919/919 - 3s - loss: 1.6212 - accuracy: 0.4796 - val_loss: 3.4736 - val_accuracy: 0.0455\n",
+      "Epoch 1368/5000\n",
+      "919/919 - 3s - loss: 1.6361 - accuracy: 0.4746 - val_loss: 3.4809 - val_accuracy: 0.0456\n",
+      "Epoch 1369/5000\n",
+      "919/919 - 3s - loss: 1.6373 - accuracy: 0.4799 - val_loss: 3.4818 - val_accuracy: 0.0457\n",
+      "Epoch 1370/5000\n",
+      "919/919 - 3s - loss: 1.6280 - accuracy: 0.4818 - val_loss: 3.4853 - val_accuracy: 0.0458\n",
+      "Epoch 1371/5000\n",
+      "919/919 - 3s - loss: 1.6363 - accuracy: 0.4767 - val_loss: 3.4812 - val_accuracy: 0.0456\n",
+      "Epoch 1372/5000\n",
+      "919/919 - 3s - loss: 1.6499 - accuracy: 0.4814 - val_loss: 3.4737 - val_accuracy: 0.0456\n",
+      "Epoch 1373/5000\n",
+      "919/919 - 3s - loss: 1.6248 - accuracy: 0.4812 - val_loss: 3.4677 - val_accuracy: 0.0455\n",
+      "Epoch 1374/5000\n",
+      "919/919 - 3s - loss: 1.6170 - accuracy: 0.4833 - val_loss: 3.4780 - val_accuracy: 0.0454\n",
+      "Epoch 1375/5000\n",
+      "919/919 - 3s - loss: 1.6328 - accuracy: 0.4812 - val_loss: 3.4746 - val_accuracy: 0.0456\n",
+      "Epoch 1376/5000\n",
+      "919/919 - 3s - loss: 1.6166 - accuracy: 0.4820 - val_loss: 3.4769 - val_accuracy: 0.0462\n",
+      "Epoch 1377/5000\n",
+      "919/919 - 3s - loss: 1.6341 - accuracy: 0.4790 - val_loss: 3.4780 - val_accuracy: 0.0456\n",
+      "Epoch 1378/5000\n",
+      "919/919 - 3s - loss: 1.6264 - accuracy: 0.4792 - val_loss: 3.4819 - val_accuracy: 0.0458\n",
+      "Epoch 1379/5000\n",
+      "919/919 - 3s - loss: 1.6251 - accuracy: 0.4796 - val_loss: 3.4822 - val_accuracy: 0.0460\n",
+      "Epoch 1380/5000\n",
+      "919/919 - 3s - loss: 1.6308 - accuracy: 0.4825 - val_loss: 3.4976 - val_accuracy: 0.0461\n",
+      "Epoch 1381/5000\n",
+      "919/919 - 3s - loss: 1.6244 - accuracy: 0.4819 - val_loss: 3.5160 - val_accuracy: 0.0455\n",
+      "Epoch 1382/5000\n",
+      "919/919 - 3s - loss: 1.6456 - accuracy: 0.4782 - val_loss: 3.4937 - val_accuracy: 0.0453\n",
+      "Epoch 1383/5000\n",
+      "919/919 - 3s - loss: 1.6205 - accuracy: 0.4840 - val_loss: 3.5024 - val_accuracy: 0.0456\n",
+      "Epoch 1384/5000\n",
+      "919/919 - 3s - loss: 1.6283 - accuracy: 0.4816 - val_loss: 3.4990 - val_accuracy: 0.0458\n",
+      "Epoch 1385/5000\n",
+      "919/919 - 3s - loss: 1.6254 - accuracy: 0.4799 - val_loss: 3.4974 - val_accuracy: 0.0462\n",
+      "Epoch 1386/5000\n",
+      "919/919 - 3s - loss: 1.6293 - accuracy: 0.4788 - val_loss: 3.5025 - val_accuracy: 0.0461\n",
+      "Epoch 1387/5000\n",
+      "919/919 - 3s - loss: 1.6242 - accuracy: 0.4820 - val_loss: 3.4952 - val_accuracy: 0.0463\n",
+      "Epoch 1388/5000\n",
+      "919/919 - 3s - loss: 1.6298 - accuracy: 0.4766 - val_loss: 3.4976 - val_accuracy: 0.0462\n",
+      "Epoch 1389/5000\n",
+      "919/919 - 3s - loss: 1.6219 - accuracy: 0.4814 - val_loss: 3.4833 - val_accuracy: 0.0463\n",
+      "Epoch 1390/5000\n",
+      "919/919 - 3s - loss: 1.6367 - accuracy: 0.4790 - val_loss: 3.4906 - val_accuracy: 0.0462\n",
+      "Epoch 1391/5000\n",
+      "919/919 - 3s - loss: 1.6391 - accuracy: 0.4795 - val_loss: 3.4871 - val_accuracy: 0.0460\n",
+      "Epoch 1392/5000\n",
+      "919/919 - 3s - loss: 1.6246 - accuracy: 0.4817 - val_loss: 3.4902 - val_accuracy: 0.0463\n",
+      "Epoch 1393/5000\n",
+      "919/919 - 3s - loss: 1.6290 - accuracy: 0.4812 - val_loss: 3.4909 - val_accuracy: 0.0462\n",
+      "Epoch 1394/5000\n",
+      "919/919 - 3s - loss: 1.6536 - accuracy: 0.4795 - val_loss: 3.4913 - val_accuracy: 0.0463\n",
+      "Epoch 1395/5000\n",
+      "919/919 - 3s - loss: 1.6266 - accuracy: 0.4834 - val_loss: 3.4910 - val_accuracy: 0.0461\n",
+      "Epoch 1396/5000\n",
+      "919/919 - 3s - loss: 1.6325 - accuracy: 0.4800 - val_loss: 3.5042 - val_accuracy: 0.0459\n",
+      "Epoch 1397/5000\n",
+      "919/919 - 3s - loss: 1.6223 - accuracy: 0.4825 - val_loss: 3.5058 - val_accuracy: 0.0465\n",
+      "Epoch 1398/5000\n",
+      "919/919 - 3s - loss: 1.6295 - accuracy: 0.4757 - val_loss: 3.5094 - val_accuracy: 0.0459\n",
+      "Epoch 1399/5000\n",
+      "919/919 - 3s - loss: 1.6304 - accuracy: 0.4821 - val_loss: 3.5011 - val_accuracy: 0.0463\n",
+      "Epoch 1400/5000\n",
+      "919/919 - 3s - loss: 1.6265 - accuracy: 0.4808 - val_loss: 3.5126 - val_accuracy: 0.0465\n",
+      "Epoch 1401/5000\n",
+      "919/919 - 3s - loss: 1.6211 - accuracy: 0.4810 - val_loss: 3.5130 - val_accuracy: 0.0465\n",
+      "Epoch 1402/5000\n",
+      "919/919 - 3s - loss: 1.6219 - accuracy: 0.4848 - val_loss: 3.5066 - val_accuracy: 0.0467\n",
+      "Epoch 1403/5000\n",
+      "919/919 - 3s - loss: 1.6268 - accuracy: 0.4801 - val_loss: 3.5011 - val_accuracy: 0.0466\n",
+      "Epoch 1404/5000\n",
+      "919/919 - 3s - loss: 1.6298 - accuracy: 0.4795 - val_loss: 3.5022 - val_accuracy: 0.0461\n",
+      "Epoch 1405/5000\n",
+      "919/919 - 3s - loss: 1.7149 - accuracy: 0.4847 - val_loss: 3.4981 - val_accuracy: 0.0463\n",
+      "Epoch 1406/5000\n",
+      "919/919 - 3s - loss: 1.6283 - accuracy: 0.4838 - val_loss: 3.4987 - val_accuracy: 0.0465\n",
+      "Epoch 1407/5000\n",
+      "919/919 - 3s - loss: 1.6263 - accuracy: 0.4823 - val_loss: 3.5202 - val_accuracy: 0.0463\n",
+      "Epoch 1408/5000\n",
+      "919/919 - 3s - loss: 1.6911 - accuracy: 0.4809 - val_loss: 3.5244 - val_accuracy: 0.0463\n",
+      "Epoch 1409/5000\n",
+      "919/919 - 3s - loss: 1.6184 - accuracy: 0.4817 - val_loss: 3.5279 - val_accuracy: 0.0462\n",
+      "Epoch 1410/5000\n",
+      "919/919 - 3s - loss: 1.6344 - accuracy: 0.4827 - val_loss: 3.5301 - val_accuracy: 0.0455\n",
+      "Epoch 1411/5000\n",
+      "919/919 - 3s - loss: 1.6253 - accuracy: 0.4822 - val_loss: 3.5369 - val_accuracy: 0.0459\n",
+      "Epoch 1412/5000\n",
+      "919/919 - 3s - loss: 1.6205 - accuracy: 0.4812 - val_loss: 3.5501 - val_accuracy: 0.0457\n",
+      "Epoch 1413/5000\n",
+      "919/919 - 3s - loss: 1.6240 - accuracy: 0.4810 - val_loss: 3.5430 - val_accuracy: 0.0458\n",
+      "Epoch 1414/5000\n",
+      "919/919 - 3s - loss: 1.6656 - accuracy: 0.4824 - val_loss: 3.5490 - val_accuracy: 0.0458\n",
+      "Epoch 1415/5000\n",
+      "919/919 - 3s - loss: 1.6269 - accuracy: 0.4813 - val_loss: 3.5502 - val_accuracy: 0.0460\n",
+      "Epoch 1416/5000\n",
+      "919/919 - 3s - loss: 1.6616 - accuracy: 0.4835 - val_loss: 3.5434 - val_accuracy: 0.0461\n",
+      "Epoch 1417/5000\n",
+      "919/919 - 3s - loss: 1.6181 - accuracy: 0.4827 - val_loss: 3.5339 - val_accuracy: 0.0463\n",
+      "Epoch 1418/5000\n",
+      "919/919 - 3s - loss: 1.6144 - accuracy: 0.4859 - val_loss: 3.5322 - val_accuracy: 0.0466\n",
+      "Epoch 1419/5000\n",
+      "919/919 - 3s - loss: 1.6672 - accuracy: 0.4880 - val_loss: 3.5336 - val_accuracy: 0.0460\n",
+      "Epoch 1420/5000\n",
+      "919/919 - 3s - loss: 1.6586 - accuracy: 0.4841 - val_loss: 3.5376 - val_accuracy: 0.0459\n",
+      "Epoch 1421/5000\n",
+      "919/919 - 3s - loss: 1.6287 - accuracy: 0.4782 - val_loss: 3.5274 - val_accuracy: 0.0466\n",
+      "Epoch 1422/5000\n",
+      "919/919 - 3s - loss: 1.6195 - accuracy: 0.4818 - val_loss: 3.5350 - val_accuracy: 0.0470\n",
+      "Epoch 1423/5000\n",
+      "919/919 - 3s - loss: 1.6472 - accuracy: 0.4807 - val_loss: 3.5499 - val_accuracy: 0.0464\n",
+      "Epoch 1424/5000\n",
+      "919/919 - 3s - loss: 1.6438 - accuracy: 0.4797 - val_loss: 3.5530 - val_accuracy: 0.0461\n",
+      "Epoch 1425/5000\n",
+      "919/919 - 3s - loss: 1.6178 - accuracy: 0.4814 - val_loss: 3.5487 - val_accuracy: 0.0459\n",
+      "Epoch 1426/5000\n",
+      "919/919 - 3s - loss: 1.6102 - accuracy: 0.4818 - val_loss: 3.5545 - val_accuracy: 0.0462\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1427/5000\n",
+      "919/919 - 3s - loss: 1.6257 - accuracy: 0.4846 - val_loss: 3.5377 - val_accuracy: 0.0466\n",
+      "Epoch 1428/5000\n",
+      "919/919 - 3s - loss: 1.6390 - accuracy: 0.4830 - val_loss: 3.5202 - val_accuracy: 0.0464\n",
+      "Epoch 1429/5000\n",
+      "919/919 - 3s - loss: 1.6262 - accuracy: 0.4801 - val_loss: 3.5214 - val_accuracy: 0.0471\n",
+      "Epoch 1430/5000\n",
+      "919/919 - 3s - loss: 1.6353 - accuracy: 0.4796 - val_loss: 3.5122 - val_accuracy: 0.0469\n",
+      "Epoch 1431/5000\n",
+      "919/919 - 3s - loss: 1.6174 - accuracy: 0.4822 - val_loss: 3.5231 - val_accuracy: 0.0470\n",
+      "Epoch 1432/5000\n",
+      "919/919 - 3s - loss: 1.6160 - accuracy: 0.4826 - val_loss: 3.5320 - val_accuracy: 0.0468\n",
+      "Epoch 1433/5000\n",
+      "919/919 - 3s - loss: 1.6279 - accuracy: 0.4846 - val_loss: 3.5204 - val_accuracy: 0.0470\n",
+      "Epoch 1434/5000\n",
+      "919/919 - 3s - loss: 1.6212 - accuracy: 0.4822 - val_loss: 3.5095 - val_accuracy: 0.0471\n",
+      "Epoch 1435/5000\n",
+      "919/919 - 3s - loss: 1.6416 - accuracy: 0.4844 - val_loss: 3.5148 - val_accuracy: 0.0467\n",
+      "Epoch 1436/5000\n",
+      "919/919 - 3s - loss: 1.6292 - accuracy: 0.4825 - val_loss: 3.5129 - val_accuracy: 0.0468\n",
+      "Epoch 1437/5000\n",
+      "919/919 - 3s - loss: 1.6510 - accuracy: 0.4818 - val_loss: 3.5217 - val_accuracy: 0.0469\n",
+      "Epoch 1438/5000\n",
+      "919/919 - 3s - loss: 1.6112 - accuracy: 0.4839 - val_loss: 3.5068 - val_accuracy: 0.0466\n",
+      "Epoch 1439/5000\n",
+      "919/919 - 3s - loss: 1.6199 - accuracy: 0.4841 - val_loss: 3.4918 - val_accuracy: 0.0465\n",
+      "Epoch 1440/5000\n",
+      "919/919 - 3s - loss: 1.6152 - accuracy: 0.4827 - val_loss: 3.4952 - val_accuracy: 0.0471\n",
+      "Epoch 1441/5000\n",
+      "919/919 - 3s - loss: 1.6236 - accuracy: 0.4844 - val_loss: 3.4831 - val_accuracy: 0.0468\n",
+      "Epoch 1442/5000\n",
+      "919/919 - 3s - loss: 1.6133 - accuracy: 0.4854 - val_loss: 3.5058 - val_accuracy: 0.0468\n",
+      "Epoch 1443/5000\n",
+      "919/919 - 3s - loss: 1.6372 - accuracy: 0.4820 - val_loss: 3.5074 - val_accuracy: 0.0466\n",
+      "Epoch 1444/5000\n",
+      "919/919 - 3s - loss: 1.6305 - accuracy: 0.4840 - val_loss: 3.5164 - val_accuracy: 0.0466\n",
+      "Epoch 1445/5000\n",
+      "919/919 - 3s - loss: 1.6260 - accuracy: 0.4831 - val_loss: 3.5213 - val_accuracy: 0.0471\n",
+      "Epoch 1446/5000\n",
+      "919/919 - 3s - loss: 1.6280 - accuracy: 0.4841 - val_loss: 3.5331 - val_accuracy: 0.0469\n",
+      "Epoch 1447/5000\n",
+      "919/919 - 3s - loss: 1.6251 - accuracy: 0.4831 - val_loss: 3.5311 - val_accuracy: 0.0471\n",
+      "Epoch 1448/5000\n",
+      "919/919 - 3s - loss: 1.6148 - accuracy: 0.4863 - val_loss: 3.5346 - val_accuracy: 0.0468\n",
+      "Epoch 1449/5000\n",
+      "919/919 - 3s - loss: 1.6192 - accuracy: 0.4844 - val_loss: 3.5384 - val_accuracy: 0.0469\n",
+      "Epoch 1450/5000\n",
+      "919/919 - 3s - loss: 1.6230 - accuracy: 0.4854 - val_loss: 3.5284 - val_accuracy: 0.0468\n",
+      "Epoch 1451/5000\n",
+      "919/919 - 3s - loss: 1.6174 - accuracy: 0.4845 - val_loss: 3.5300 - val_accuracy: 0.0467\n",
+      "Epoch 1452/5000\n",
+      "919/919 - 3s - loss: 1.6243 - accuracy: 0.4819 - val_loss: 3.5182 - val_accuracy: 0.0466\n",
+      "Epoch 1453/5000\n",
+      "919/919 - 3s - loss: 1.6240 - accuracy: 0.4868 - val_loss: 3.5235 - val_accuracy: 0.0469\n",
+      "Epoch 1454/5000\n",
+      "919/919 - 3s - loss: 1.6199 - accuracy: 0.4844 - val_loss: 3.5264 - val_accuracy: 0.0471\n",
+      "Epoch 1455/5000\n",
+      "919/919 - 3s - loss: 1.6712 - accuracy: 0.4840 - val_loss: 3.5224 - val_accuracy: 0.0471\n",
+      "Epoch 1456/5000\n",
+      "919/919 - 3s - loss: 1.6154 - accuracy: 0.4839 - val_loss: 3.5075 - val_accuracy: 0.0471\n",
+      "Epoch 1457/5000\n",
+      "919/919 - 3s - loss: 1.6311 - accuracy: 0.4843 - val_loss: 3.5150 - val_accuracy: 0.0471\n",
+      "Epoch 1458/5000\n",
+      "919/919 - 3s - loss: 1.6492 - accuracy: 0.4856 - val_loss: 3.5129 - val_accuracy: 0.0471\n",
+      "Epoch 1459/5000\n",
+      "919/919 - 3s - loss: 1.6144 - accuracy: 0.4804 - val_loss: 3.4911 - val_accuracy: 0.0468\n",
+      "Epoch 1460/5000\n",
+      "919/919 - 3s - loss: 1.6150 - accuracy: 0.4819 - val_loss: 3.4952 - val_accuracy: 0.0468\n",
+      "Epoch 1461/5000\n",
+      "919/919 - 3s - loss: 1.6195 - accuracy: 0.4855 - val_loss: 3.5085 - val_accuracy: 0.0471\n",
+      "Epoch 1462/5000\n",
+      "919/919 - 3s - loss: 1.6102 - accuracy: 0.4836 - val_loss: 3.5054 - val_accuracy: 0.0471\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1463/5000\n",
+      "919/919 - 3s - loss: 1.6454 - accuracy: 0.4852 - val_loss: 3.5188 - val_accuracy: 0.0470\n",
+      "Epoch 1464/5000\n",
+      "919/919 - 3s - loss: 1.6067 - accuracy: 0.4859 - val_loss: 3.5056 - val_accuracy: 0.0469\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1465/5000\n",
+      "919/919 - 3s - loss: 1.6269 - accuracy: 0.4834 - val_loss: 3.5021 - val_accuracy: 0.0471\n",
+      "Epoch 1466/5000\n",
+      "919/919 - 3s - loss: 1.6294 - accuracy: 0.4867 - val_loss: 3.5151 - val_accuracy: 0.0471\n",
+      "Epoch 1467/5000\n",
+      "919/919 - 3s - loss: 1.6142 - accuracy: 0.4822 - val_loss: 3.5064 - val_accuracy: 0.0473\n",
+      "Epoch 1468/5000\n",
+      "919/919 - 3s - loss: 1.6098 - accuracy: 0.4875 - val_loss: 3.5151 - val_accuracy: 0.0473\n",
+      "Epoch 1469/5000\n",
+      "919/919 - 3s - loss: 1.6153 - accuracy: 0.4846 - val_loss: 3.5214 - val_accuracy: 0.0471\n",
+      "Epoch 1470/5000\n",
+      "919/919 - 3s - loss: 1.6201 - accuracy: 0.4805 - val_loss: 3.5190 - val_accuracy: 0.0471\n",
+      "Epoch 1471/5000\n",
+      "919/919 - 3s - loss: 1.6127 - accuracy: 0.4833 - val_loss: 3.5167 - val_accuracy: 0.0471\n",
+      "Epoch 1472/5000\n",
+      "919/919 - 3s - loss: 1.6459 - accuracy: 0.4869 - val_loss: 3.5239 - val_accuracy: 0.0470\n",
+      "Epoch 1473/5000\n",
+      "919/919 - 3s - loss: 1.5999 - accuracy: 0.4849 - val_loss: 3.5384 - val_accuracy: 0.0471\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1474/5000\n",
+      "919/919 - 3s - loss: 1.6090 - accuracy: 0.4835 - val_loss: 3.5331 - val_accuracy: 0.0470\n",
+      "Epoch 1475/5000\n",
+      "919/919 - 3s - loss: 1.6146 - accuracy: 0.4846 - val_loss: 3.5325 - val_accuracy: 0.0468\n",
+      "Epoch 1476/5000\n",
+      "919/919 - 3s - loss: 1.6110 - accuracy: 0.4864 - val_loss: 3.5248 - val_accuracy: 0.0465\n",
+      "Epoch 1477/5000\n",
+      "919/919 - 3s - loss: 1.7405 - accuracy: 0.4878 - val_loss: 3.5202 - val_accuracy: 0.0468\n",
+      "Epoch 1478/5000\n",
+      "919/919 - 3s - loss: 1.6029 - accuracy: 0.4882 - val_loss: 3.5474 - val_accuracy: 0.0476\n",
+      "Epoch 1479/5000\n",
+      "919/919 - 3s - loss: 1.6455 - accuracy: 0.4859 - val_loss: 3.5402 - val_accuracy: 0.0471\n",
+      "Epoch 1480/5000\n",
+      "919/919 - 3s - loss: 1.6022 - accuracy: 0.4888 - val_loss: 3.5377 - val_accuracy: 0.0469\n",
+      "Epoch 1481/5000\n",
+      "919/919 - 3s - loss: 1.6117 - accuracy: 0.4865 - val_loss: 3.5300 - val_accuracy: 0.0468\n",
+      "Epoch 1482/5000\n",
+      "919/919 - 3s - loss: 1.6486 - accuracy: 0.4875 - val_loss: 3.5316 - val_accuracy: 0.0471\n",
+      "Epoch 1483/5000\n",
+      "919/919 - 3s - loss: 1.6218 - accuracy: 0.4824 - val_loss: 3.5273 - val_accuracy: 0.0473\n",
+      "Epoch 1484/5000\n",
+      "919/919 - 3s - loss: 1.6046 - accuracy: 0.4886 - val_loss: 3.5277 - val_accuracy: 0.0473\n",
+      "Epoch 1485/5000\n",
+      "919/919 - 3s - loss: 1.5961 - accuracy: 0.4898 - val_loss: 3.5063 - val_accuracy: 0.0475\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1486/5000\n",
+      "919/919 - 3s - loss: 1.6117 - accuracy: 0.4876 - val_loss: 3.5160 - val_accuracy: 0.0475\n",
+      "Epoch 1487/5000\n",
+      "919/919 - 3s - loss: 1.6088 - accuracy: 0.4861 - val_loss: 3.5084 - val_accuracy: 0.0472\n",
+      "Epoch 1488/5000\n",
+      "919/919 - 3s - loss: 1.6365 - accuracy: 0.4863 - val_loss: 3.5171 - val_accuracy: 0.0468\n",
+      "Epoch 1489/5000\n",
+      "919/919 - 3s - loss: 1.6083 - accuracy: 0.4859 - val_loss: 3.5243 - val_accuracy: 0.0469\n",
+      "Epoch 1490/5000\n",
+      "919/919 - 3s - loss: 1.6069 - accuracy: 0.4886 - val_loss: 3.5227 - val_accuracy: 0.0469\n",
+      "Epoch 1491/5000\n",
+      "919/919 - 3s - loss: 1.6008 - accuracy: 0.4859 - val_loss: 3.5332 - val_accuracy: 0.0468\n",
+      "Epoch 1492/5000\n",
+      "919/919 - 3s - loss: 1.5939 - accuracy: 0.4871 - val_loss: 3.5348 - val_accuracy: 0.0470\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1493/5000\n",
+      "919/919 - 3s - loss: 1.6194 - accuracy: 0.4875 - val_loss: 3.5333 - val_accuracy: 0.0468\n",
+      "Epoch 1494/5000\n",
+      "919/919 - 3s - loss: 1.6721 - accuracy: 0.4848 - val_loss: 3.5445 - val_accuracy: 0.0469\n",
+      "Epoch 1495/5000\n",
+      "919/919 - 3s - loss: 1.6073 - accuracy: 0.4851 - val_loss: 3.5470 - val_accuracy: 0.0471\n",
+      "Epoch 1496/5000\n",
+      "919/919 - 3s - loss: 1.6039 - accuracy: 0.4882 - val_loss: 3.5660 - val_accuracy: 0.0471\n",
+      "Epoch 1497/5000\n",
+      "919/919 - 3s - loss: 1.5974 - accuracy: 0.4878 - val_loss: 3.5486 - val_accuracy: 0.0472\n",
+      "Epoch 1498/5000\n",
+      "919/919 - 3s - loss: 1.6076 - accuracy: 0.4871 - val_loss: 3.5503 - val_accuracy: 0.0473\n",
+      "Epoch 1499/5000\n",
+      "919/919 - 3s - loss: 1.5949 - accuracy: 0.4852 - val_loss: 3.5515 - val_accuracy: 0.0474\n",
+      "Epoch 1500/5000\n",
+      "919/919 - 3s - loss: 1.5975 - accuracy: 0.4877 - val_loss: 3.5442 - val_accuracy: 0.0471\n",
+      "Epoch 1501/5000\n",
+      "919/919 - 3s - loss: 1.6036 - accuracy: 0.4861 - val_loss: 3.5334 - val_accuracy: 0.0477\n",
+      "Epoch 1502/5000\n",
+      "919/919 - 3s - loss: 1.6058 - accuracy: 0.4893 - val_loss: 3.5276 - val_accuracy: 0.0480\n",
+      "Epoch 1503/5000\n",
+      "919/919 - 3s - loss: 1.6110 - accuracy: 0.4879 - val_loss: 3.5340 - val_accuracy: 0.0472\n",
+      "Epoch 1504/5000\n",
+      "919/919 - 3s - loss: 1.5934 - accuracy: 0.4885 - val_loss: 3.5526 - val_accuracy: 0.0474\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1505/5000\n",
+      "919/919 - 3s - loss: 1.6151 - accuracy: 0.4900 - val_loss: 3.5633 - val_accuracy: 0.0479\n",
+      "Epoch 1506/5000\n",
+      "919/919 - 3s - loss: 1.6068 - accuracy: 0.4882 - val_loss: 3.5621 - val_accuracy: 0.0476\n",
+      "Epoch 1507/5000\n",
+      "919/919 - 3s - loss: 1.6285 - accuracy: 0.4834 - val_loss: 3.5530 - val_accuracy: 0.0478\n",
+      "Epoch 1508/5000\n",
+      "919/919 - 3s - loss: 1.6382 - accuracy: 0.4881 - val_loss: 3.5471 - val_accuracy: 0.0482\n",
+      "Epoch 1509/5000\n",
+      "919/919 - 3s - loss: 1.5926 - accuracy: 0.4918 - val_loss: 3.5372 - val_accuracy: 0.0480\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1510/5000\n",
+      "919/919 - 3s - loss: 1.5953 - accuracy: 0.4907 - val_loss: 3.5407 - val_accuracy: 0.0478\n",
+      "Epoch 1511/5000\n",
+      "919/919 - 3s - loss: 1.5954 - accuracy: 0.4876 - val_loss: 3.5458 - val_accuracy: 0.0477\n",
+      "Epoch 1512/5000\n",
+      "919/919 - 3s - loss: 1.6031 - accuracy: 0.4884 - val_loss: 3.5306 - val_accuracy: 0.0477\n",
+      "Epoch 1513/5000\n",
+      "919/919 - 3s - loss: 1.6026 - accuracy: 0.4895 - val_loss: 3.5439 - val_accuracy: 0.0474\n",
+      "Epoch 1514/5000\n",
+      "919/919 - 3s - loss: 1.6051 - accuracy: 0.4886 - val_loss: 3.5376 - val_accuracy: 0.0474\n",
+      "Epoch 1515/5000\n",
+      "919/919 - 3s - loss: 1.8145 - accuracy: 0.4901 - val_loss: 3.5407 - val_accuracy: 0.0471\n",
+      "Epoch 1516/5000\n",
+      "919/919 - 3s - loss: 1.5917 - accuracy: 0.4907 - val_loss: 3.5485 - val_accuracy: 0.0471\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1517/5000\n",
+      "919/919 - 3s - loss: 1.6044 - accuracy: 0.4847 - val_loss: 3.5440 - val_accuracy: 0.0472\n",
+      "Epoch 1518/5000\n",
+      "919/919 - 3s - loss: 1.5871 - accuracy: 0.4937 - val_loss: 3.5452 - val_accuracy: 0.0472\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1519/5000\n",
+      "919/919 - 3s - loss: 1.6101 - accuracy: 0.4882 - val_loss: 3.5337 - val_accuracy: 0.0475\n",
+      "Epoch 1520/5000\n",
+      "919/919 - 3s - loss: 1.6037 - accuracy: 0.4852 - val_loss: 3.5341 - val_accuracy: 0.0477\n",
+      "Epoch 1521/5000\n",
+      "919/919 - 3s - loss: 1.6234 - accuracy: 0.4840 - val_loss: 3.5356 - val_accuracy: 0.0473\n",
+      "Epoch 1522/5000\n",
+      "919/919 - 3s - loss: 1.6333 - accuracy: 0.4869 - val_loss: 3.5407 - val_accuracy: 0.0473\n",
+      "Epoch 1523/5000\n",
+      "919/919 - 3s - loss: 1.6542 - accuracy: 0.4887 - val_loss: 3.5418 - val_accuracy: 0.0475\n",
+      "Epoch 1524/5000\n",
+      "919/919 - 3s - loss: 1.6004 - accuracy: 0.4878 - val_loss: 3.5668 - val_accuracy: 0.0478\n",
+      "Epoch 1525/5000\n",
+      "919/919 - 3s - loss: 1.6026 - accuracy: 0.4850 - val_loss: 3.5590 - val_accuracy: 0.0479\n",
+      "Epoch 1526/5000\n",
+      "919/919 - 3s - loss: 1.6040 - accuracy: 0.4888 - val_loss: 3.5427 - val_accuracy: 0.0477\n",
+      "Epoch 1527/5000\n",
+      "919/919 - 3s - loss: 1.5989 - accuracy: 0.4867 - val_loss: 3.5426 - val_accuracy: 0.0477\n",
+      "Epoch 1528/5000\n",
+      "919/919 - 3s - loss: 1.5941 - accuracy: 0.4913 - val_loss: 3.5413 - val_accuracy: 0.0480\n",
+      "Epoch 1529/5000\n",
+      "919/919 - 3s - loss: 1.6176 - accuracy: 0.4895 - val_loss: 3.5310 - val_accuracy: 0.0481\n",
+      "Epoch 1530/5000\n",
+      "919/919 - 3s - loss: 1.5929 - accuracy: 0.4866 - val_loss: 3.5512 - val_accuracy: 0.0478\n",
+      "Epoch 1531/5000\n",
+      "919/919 - 3s - loss: 1.5981 - accuracy: 0.4891 - val_loss: 3.5698 - val_accuracy: 0.0484\n",
+      "Epoch 1532/5000\n",
+      "919/919 - 3s - loss: 1.6074 - accuracy: 0.4880 - val_loss: 3.5670 - val_accuracy: 0.0479\n",
+      "Epoch 1533/5000\n",
+      "919/919 - 3s - loss: 1.6337 - accuracy: 0.4903 - val_loss: 3.5718 - val_accuracy: 0.0479\n",
+      "Epoch 1534/5000\n",
+      "919/919 - 3s - loss: 1.6339 - accuracy: 0.4897 - val_loss: 3.5749 - val_accuracy: 0.0482\n",
+      "Epoch 1535/5000\n",
+      "919/919 - 3s - loss: 1.5944 - accuracy: 0.4905 - val_loss: 3.5686 - val_accuracy: 0.0489\n",
+      "Epoch 1536/5000\n",
+      "919/919 - 3s - loss: 1.6247 - accuracy: 0.4909 - val_loss: 3.5770 - val_accuracy: 0.0485\n",
+      "Epoch 1537/5000\n",
+      "919/919 - 3s - loss: 1.6014 - accuracy: 0.4902 - val_loss: 3.5706 - val_accuracy: 0.0482\n",
+      "Epoch 1538/5000\n",
+      "919/919 - 3s - loss: 1.5944 - accuracy: 0.4916 - val_loss: 3.5755 - val_accuracy: 0.0486\n",
+      "Epoch 1539/5000\n",
+      "919/919 - 3s - loss: 1.6073 - accuracy: 0.4890 - val_loss: 3.5782 - val_accuracy: 0.0482\n",
+      "Epoch 1540/5000\n",
+      "919/919 - 3s - loss: 1.6393 - accuracy: 0.4878 - val_loss: 3.5727 - val_accuracy: 0.0480\n",
+      "Epoch 1541/5000\n",
+      "919/919 - 3s - loss: 1.5998 - accuracy: 0.4907 - val_loss: 3.5690 - val_accuracy: 0.0484\n",
+      "Epoch 1542/5000\n",
+      "919/919 - 3s - loss: 1.5986 - accuracy: 0.4871 - val_loss: 3.5696 - val_accuracy: 0.0484\n",
+      "Epoch 1543/5000\n",
+      "919/919 - 3s - loss: 1.5890 - accuracy: 0.4907 - val_loss: 3.5583 - val_accuracy: 0.0482\n",
+      "Epoch 1544/5000\n",
+      "919/919 - 3s - loss: 1.6063 - accuracy: 0.4890 - val_loss: 3.5591 - val_accuracy: 0.0484\n",
+      "Epoch 1545/5000\n",
+      "919/919 - 3s - loss: 1.5881 - accuracy: 0.4926 - val_loss: 3.5557 - val_accuracy: 0.0481\n",
+      "Epoch 1546/5000\n",
+      "919/919 - 3s - loss: 1.5910 - accuracy: 0.4895 - val_loss: 3.5519 - val_accuracy: 0.0481\n",
+      "Epoch 1547/5000\n",
+      "919/919 - 3s - loss: 1.5949 - accuracy: 0.4888 - val_loss: 3.5536 - val_accuracy: 0.0484\n",
+      "Epoch 1548/5000\n",
+      "919/919 - 3s - loss: 1.6171 - accuracy: 0.4877 - val_loss: 3.5620 - val_accuracy: 0.0486\n",
+      "Epoch 1549/5000\n",
+      "919/919 - 3s - loss: 1.6327 - accuracy: 0.4905 - val_loss: 3.5511 - val_accuracy: 0.0480\n",
+      "Epoch 1550/5000\n",
+      "919/919 - 3s - loss: 1.6049 - accuracy: 0.4933 - val_loss: 3.5658 - val_accuracy: 0.0482\n",
+      "Epoch 1551/5000\n",
+      "919/919 - 3s - loss: 1.5867 - accuracy: 0.4912 - val_loss: 3.5642 - val_accuracy: 0.0481\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1552/5000\n",
+      "919/919 - 3s - loss: 1.6126 - accuracy: 0.4904 - val_loss: 3.5524 - val_accuracy: 0.0480\n",
+      "Epoch 1553/5000\n",
+      "919/919 - 3s - loss: 1.5876 - accuracy: 0.4903 - val_loss: 3.5424 - val_accuracy: 0.0482\n",
+      "Epoch 1554/5000\n",
+      "919/919 - 3s - loss: 1.5982 - accuracy: 0.4915 - val_loss: 3.5526 - val_accuracy: 0.0483\n",
+      "Epoch 1555/5000\n",
+      "919/919 - 3s - loss: 1.5811 - accuracy: 0.4897 - val_loss: 3.5461 - val_accuracy: 0.0481\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1556/5000\n",
+      "919/919 - 3s - loss: 1.5922 - accuracy: 0.4884 - val_loss: 3.5549 - val_accuracy: 0.0481\n",
+      "Epoch 1557/5000\n",
+      "919/919 - 3s - loss: 1.5906 - accuracy: 0.4918 - val_loss: 3.5598 - val_accuracy: 0.0483\n",
+      "Epoch 1558/5000\n",
+      "919/919 - 3s - loss: 1.5818 - accuracy: 0.4892 - val_loss: 3.5561 - val_accuracy: 0.0482\n",
+      "Epoch 1559/5000\n",
+      "919/919 - 3s - loss: 1.6035 - accuracy: 0.4869 - val_loss: 3.5715 - val_accuracy: 0.0483\n",
+      "Epoch 1560/5000\n",
+      "919/919 - 3s - loss: 1.6885 - accuracy: 0.4914 - val_loss: 3.5581 - val_accuracy: 0.0484\n",
+      "Epoch 1561/5000\n",
+      "919/919 - 3s - loss: 1.5815 - accuracy: 0.4947 - val_loss: 3.5517 - val_accuracy: 0.0487\n",
+      "Epoch 1562/5000\n",
+      "919/919 - 3s - loss: 1.6012 - accuracy: 0.4913 - val_loss: 3.5425 - val_accuracy: 0.0485\n",
+      "Epoch 1563/5000\n",
+      "919/919 - 3s - loss: 1.5950 - accuracy: 0.4909 - val_loss: 3.5444 - val_accuracy: 0.0486\n",
+      "Epoch 1564/5000\n",
+      "919/919 - 3s - loss: 1.5871 - accuracy: 0.4908 - val_loss: 3.5572 - val_accuracy: 0.0487\n",
+      "Epoch 1565/5000\n",
+      "919/919 - 3s - loss: 1.5991 - accuracy: 0.4870 - val_loss: 3.5713 - val_accuracy: 0.0484\n",
+      "Epoch 1566/5000\n",
+      "919/919 - 3s - loss: 1.5896 - accuracy: 0.4874 - val_loss: 3.5661 - val_accuracy: 0.0481\n",
+      "Epoch 1567/5000\n",
+      "919/919 - 3s - loss: 1.6166 - accuracy: 0.4898 - val_loss: 3.5548 - val_accuracy: 0.0481\n",
+      "Epoch 1568/5000\n",
+      "919/919 - 3s - loss: 1.5863 - accuracy: 0.4890 - val_loss: 3.5696 - val_accuracy: 0.0478\n",
+      "Epoch 1569/5000\n",
+      "919/919 - 3s - loss: 1.5901 - accuracy: 0.4910 - val_loss: 3.5539 - val_accuracy: 0.0483\n",
+      "Epoch 1570/5000\n",
+      "919/919 - 3s - loss: 1.6518 - accuracy: 0.4913 - val_loss: 3.5622 - val_accuracy: 0.0482\n",
+      "Epoch 1571/5000\n",
+      "919/919 - 3s - loss: 1.5803 - accuracy: 0.4937 - val_loss: 3.5746 - val_accuracy: 0.0487\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1572/5000\n",
+      "919/919 - 3s - loss: 1.5953 - accuracy: 0.4890 - val_loss: 3.5662 - val_accuracy: 0.0486\n",
+      "Epoch 1573/5000\n",
+      "919/919 - 3s - loss: 1.5986 - accuracy: 0.4904 - val_loss: 3.5502 - val_accuracy: 0.0484\n",
+      "Epoch 1574/5000\n",
+      "919/919 - 3s - loss: 1.5896 - accuracy: 0.4927 - val_loss: 3.5599 - val_accuracy: 0.0480\n",
+      "Epoch 1575/5000\n",
+      "919/919 - 3s - loss: 1.5900 - accuracy: 0.4953 - val_loss: 3.5586 - val_accuracy: 0.0485\n",
+      "Epoch 1576/5000\n",
+      "919/919 - 3s - loss: 1.6061 - accuracy: 0.4859 - val_loss: 3.5603 - val_accuracy: 0.0487\n",
+      "Epoch 1577/5000\n",
+      "919/919 - 3s - loss: 1.5808 - accuracy: 0.4899 - val_loss: 3.5638 - val_accuracy: 0.0485\n",
+      "Epoch 1578/5000\n",
+      "919/919 - 3s - loss: 1.5838 - accuracy: 0.4916 - val_loss: 3.5592 - val_accuracy: 0.0486\n",
+      "Epoch 1579/5000\n",
+      "919/919 - 3s - loss: 1.6376 - accuracy: 0.4903 - val_loss: 3.5621 - val_accuracy: 0.0490\n",
+      "Epoch 1580/5000\n",
+      "919/919 - 3s - loss: 1.5970 - accuracy: 0.4908 - val_loss: 3.5598 - val_accuracy: 0.0486\n",
+      "Epoch 1581/5000\n",
+      "919/919 - 3s - loss: 1.5761 - accuracy: 0.4931 - val_loss: 3.5649 - val_accuracy: 0.0486\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1582/5000\n",
+      "919/919 - 3s - loss: 1.5760 - accuracy: 0.4881 - val_loss: 3.5594 - val_accuracy: 0.0486\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1583/5000\n",
+      "919/919 - 3s - loss: 1.5903 - accuracy: 0.4914 - val_loss: 3.5589 - val_accuracy: 0.0486\n",
+      "Epoch 1584/5000\n",
+      "919/919 - 3s - loss: 1.6184 - accuracy: 0.4910 - val_loss: 3.5501 - val_accuracy: 0.0481\n",
+      "Epoch 1585/5000\n",
+      "919/919 - 3s - loss: 1.5813 - accuracy: 0.4945 - val_loss: 3.5558 - val_accuracy: 0.0483\n",
+      "Epoch 1586/5000\n",
+      "919/919 - 3s - loss: 1.5891 - accuracy: 0.4897 - val_loss: 3.5635 - val_accuracy: 0.0487\n",
+      "Epoch 1587/5000\n",
+      "919/919 - 3s - loss: 1.5864 - accuracy: 0.4918 - val_loss: 3.5652 - val_accuracy: 0.0483\n",
+      "Epoch 1588/5000\n",
+      "919/919 - 3s - loss: 1.6111 - accuracy: 0.4915 - val_loss: 3.5655 - val_accuracy: 0.0483\n",
+      "Epoch 1589/5000\n",
+      "919/919 - 3s - loss: 1.7808 - accuracy: 0.4935 - val_loss: 3.5656 - val_accuracy: 0.0484\n",
+      "Epoch 1590/5000\n",
+      "919/919 - 3s - loss: 1.5826 - accuracy: 0.4940 - val_loss: 3.5760 - val_accuracy: 0.0484\n",
+      "Epoch 1591/5000\n",
+      "919/919 - 3s - loss: 1.7355 - accuracy: 0.4907 - val_loss: 3.5793 - val_accuracy: 0.0484\n",
+      "Epoch 1592/5000\n",
+      "919/919 - 3s - loss: 1.5898 - accuracy: 0.4891 - val_loss: 3.5705 - val_accuracy: 0.0484\n",
+      "Epoch 1593/5000\n",
+      "919/919 - 3s - loss: 1.6048 - accuracy: 0.4874 - val_loss: 3.5515 - val_accuracy: 0.0482\n",
+      "Epoch 1594/5000\n",
+      "919/919 - 3s - loss: 1.6020 - accuracy: 0.4869 - val_loss: 3.5436 - val_accuracy: 0.0481\n",
+      "Epoch 1595/5000\n",
+      "919/919 - 3s - loss: 1.5844 - accuracy: 0.4914 - val_loss: 3.5522 - val_accuracy: 0.0482\n",
+      "Epoch 1596/5000\n",
+      "919/919 - 3s - loss: 1.5793 - accuracy: 0.4933 - val_loss: 3.5335 - val_accuracy: 0.0485\n",
+      "Epoch 1597/5000\n",
+      "919/919 - 3s - loss: 1.5911 - accuracy: 0.4929 - val_loss: 3.5381 - val_accuracy: 0.0489\n",
+      "Epoch 1598/5000\n",
+      "919/919 - 3s - loss: 1.5869 - accuracy: 0.4908 - val_loss: 3.5395 - val_accuracy: 0.0485\n",
+      "Epoch 1599/5000\n",
+      "919/919 - 3s - loss: 1.5808 - accuracy: 0.4910 - val_loss: 3.5462 - val_accuracy: 0.0483\n",
+      "Epoch 1600/5000\n",
+      "919/919 - 3s - loss: 1.5943 - accuracy: 0.4918 - val_loss: 3.5509 - val_accuracy: 0.0486\n",
+      "Epoch 1601/5000\n",
+      "919/919 - 3s - loss: 1.5791 - accuracy: 0.4943 - val_loss: 3.5524 - val_accuracy: 0.0487\n",
+      "Epoch 1602/5000\n",
+      "919/919 - 3s - loss: 1.5903 - accuracy: 0.4948 - val_loss: 3.5670 - val_accuracy: 0.0493\n",
+      "Epoch 1603/5000\n",
+      "919/919 - 3s - loss: 1.5871 - accuracy: 0.4954 - val_loss: 3.5778 - val_accuracy: 0.0492\n",
+      "Epoch 1604/5000\n",
+      "919/919 - 3s - loss: 1.5764 - accuracy: 0.4956 - val_loss: 3.5584 - val_accuracy: 0.0492\n",
+      "Epoch 1605/5000\n",
+      "919/919 - 3s - loss: 1.5768 - accuracy: 0.4933 - val_loss: 3.5797 - val_accuracy: 0.0493\n",
+      "Epoch 1606/5000\n",
+      "919/919 - 3s - loss: 1.6239 - accuracy: 0.4963 - val_loss: 3.5873 - val_accuracy: 0.0493\n",
+      "Epoch 1607/5000\n",
+      "919/919 - 3s - loss: 1.5935 - accuracy: 0.4918 - val_loss: 3.5889 - val_accuracy: 0.0492\n",
+      "Epoch 1608/5000\n",
+      "919/919 - 3s - loss: 1.5823 - accuracy: 0.4912 - val_loss: 3.5753 - val_accuracy: 0.0490\n",
+      "Epoch 1609/5000\n",
+      "919/919 - 3s - loss: 1.5813 - accuracy: 0.4938 - val_loss: 3.5687 - val_accuracy: 0.0488\n",
+      "Epoch 1610/5000\n",
+      "919/919 - 3s - loss: 1.6035 - accuracy: 0.4882 - val_loss: 3.5551 - val_accuracy: 0.0491\n",
+      "Epoch 1611/5000\n",
+      "919/919 - 3s - loss: 1.5766 - accuracy: 0.4911 - val_loss: 3.5701 - val_accuracy: 0.0492\n",
+      "Epoch 1612/5000\n",
+      "919/919 - 3s - loss: 1.5899 - accuracy: 0.4937 - val_loss: 3.5891 - val_accuracy: 0.0496\n",
+      "Epoch 1613/5000\n",
+      "919/919 - 3s - loss: 1.5830 - accuracy: 0.4941 - val_loss: 3.5965 - val_accuracy: 0.0495\n",
+      "Epoch 1614/5000\n",
+      "919/919 - 3s - loss: 1.5752 - accuracy: 0.4932 - val_loss: 3.5887 - val_accuracy: 0.0498\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1615/5000\n",
+      "919/919 - 3s - loss: 1.5768 - accuracy: 0.4937 - val_loss: 3.5852 - val_accuracy: 0.0496\n",
+      "Epoch 1616/5000\n",
+      "919/919 - 3s - loss: 1.6120 - accuracy: 0.4915 - val_loss: 3.5685 - val_accuracy: 0.0492\n",
+      "Epoch 1617/5000\n",
+      "919/919 - 3s - loss: 1.5856 - accuracy: 0.4986 - val_loss: 3.5623 - val_accuracy: 0.0497\n",
+      "Epoch 1618/5000\n",
+      "919/919 - 3s - loss: 1.5867 - accuracy: 0.4888 - val_loss: 3.5586 - val_accuracy: 0.0493\n",
+      "Epoch 1619/5000\n",
+      "919/919 - 3s - loss: 1.5748 - accuracy: 0.4924 - val_loss: 3.5614 - val_accuracy: 0.0490\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1620/5000\n",
+      "919/919 - 3s - loss: 1.5821 - accuracy: 0.4907 - val_loss: 3.5748 - val_accuracy: 0.0491\n",
+      "Epoch 1621/5000\n",
+      "919/919 - 3s - loss: 1.5705 - accuracy: 0.4928 - val_loss: 3.5736 - val_accuracy: 0.0495\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1622/5000\n",
+      "919/919 - 3s - loss: 1.5844 - accuracy: 0.4918 - val_loss: 3.5711 - val_accuracy: 0.0494\n",
+      "Epoch 1623/5000\n",
+      "919/919 - 3s - loss: 1.5812 - accuracy: 0.4962 - val_loss: 3.5666 - val_accuracy: 0.0491\n",
+      "Epoch 1624/5000\n",
+      "919/919 - 3s - loss: 1.5754 - accuracy: 0.4929 - val_loss: 3.5669 - val_accuracy: 0.0493\n",
+      "Epoch 1625/5000\n",
+      "919/919 - 3s - loss: 1.6085 - accuracy: 0.4934 - val_loss: 3.5516 - val_accuracy: 0.0489\n",
+      "Epoch 1626/5000\n",
+      "919/919 - 3s - loss: 1.5854 - accuracy: 0.4909 - val_loss: 3.5595 - val_accuracy: 0.0493\n",
+      "Epoch 1627/5000\n",
+      "919/919 - 3s - loss: 1.6197 - accuracy: 0.4931 - val_loss: 3.5574 - val_accuracy: 0.0489\n",
+      "Epoch 1628/5000\n",
+      "919/919 - 3s - loss: 1.5734 - accuracy: 0.4933 - val_loss: 3.5710 - val_accuracy: 0.0493\n",
+      "Epoch 1629/5000\n",
+      "919/919 - 3s - loss: 1.5780 - accuracy: 0.4933 - val_loss: 3.5707 - val_accuracy: 0.0494\n",
+      "Epoch 1630/5000\n",
+      "919/919 - 3s - loss: 1.5821 - accuracy: 0.4914 - val_loss: 3.5752 - val_accuracy: 0.0494\n",
+      "Epoch 1631/5000\n",
+      "919/919 - 3s - loss: 1.5862 - accuracy: 0.4909 - val_loss: 3.5823 - val_accuracy: 0.0492\n",
+      "Epoch 1632/5000\n",
+      "919/919 - 3s - loss: 1.5747 - accuracy: 0.4903 - val_loss: 3.5814 - val_accuracy: 0.0496\n",
+      "Epoch 1633/5000\n",
+      "919/919 - 3s - loss: 1.6131 - accuracy: 0.4929 - val_loss: 3.5646 - val_accuracy: 0.0495\n",
+      "Epoch 1634/5000\n",
+      "919/919 - 3s - loss: 1.5730 - accuracy: 0.4934 - val_loss: 3.5645 - val_accuracy: 0.0493\n",
+      "Epoch 1635/5000\n",
+      "919/919 - 3s - loss: 1.5693 - accuracy: 0.4939 - val_loss: 3.5796 - val_accuracy: 0.0489\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1636/5000\n",
+      "919/919 - 3s - loss: 1.5896 - accuracy: 0.4929 - val_loss: 3.5889 - val_accuracy: 0.0491\n",
+      "Epoch 1637/5000\n",
+      "919/919 - 3s - loss: 1.5779 - accuracy: 0.4914 - val_loss: 3.5757 - val_accuracy: 0.0489\n",
+      "Epoch 1638/5000\n",
+      "919/919 - 3s - loss: 1.5659 - accuracy: 0.4933 - val_loss: 3.5757 - val_accuracy: 0.0499\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1639/5000\n",
+      "919/919 - 3s - loss: 1.5657 - accuracy: 0.4978 - val_loss: 3.5861 - val_accuracy: 0.0499\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1640/5000\n",
+      "919/919 - 3s - loss: 1.5804 - accuracy: 0.4904 - val_loss: 3.5601 - val_accuracy: 0.0492\n",
+      "Epoch 1641/5000\n",
+      "919/919 - 3s - loss: 1.5760 - accuracy: 0.4927 - val_loss: 3.5654 - val_accuracy: 0.0489\n",
+      "Epoch 1642/5000\n",
+      "919/919 - 3s - loss: 1.5778 - accuracy: 0.4925 - val_loss: 3.5777 - val_accuracy: 0.0489\n",
+      "Epoch 1643/5000\n",
+      "919/919 - 3s - loss: 1.5737 - accuracy: 0.4959 - val_loss: 3.5805 - val_accuracy: 0.0489\n",
+      "Epoch 1644/5000\n",
+      "919/919 - 3s - loss: 1.5704 - accuracy: 0.4945 - val_loss: 3.5903 - val_accuracy: 0.0495\n",
+      "Epoch 1645/5000\n",
+      "919/919 - 3s - loss: 1.5824 - accuracy: 0.4907 - val_loss: 3.5847 - val_accuracy: 0.0500\n",
+      "Epoch 1646/5000\n",
+      "919/919 - 3s - loss: 1.5717 - accuracy: 0.4967 - val_loss: 3.6105 - val_accuracy: 0.0497\n",
+      "Epoch 1647/5000\n",
+      "919/919 - 3s - loss: 1.5654 - accuracy: 0.4980 - val_loss: 3.6049 - val_accuracy: 0.0496\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1648/5000\n",
+      "919/919 - 3s - loss: 1.5738 - accuracy: 0.4913 - val_loss: 3.6061 - val_accuracy: 0.0495\n",
+      "Epoch 1649/5000\n",
+      "919/919 - 3s - loss: 1.5840 - accuracy: 0.4932 - val_loss: 3.5838 - val_accuracy: 0.0499\n",
+      "Epoch 1650/5000\n",
+      "919/919 - 3s - loss: 1.5864 - accuracy: 0.4948 - val_loss: 3.5901 - val_accuracy: 0.0501\n",
+      "Epoch 1651/5000\n",
+      "919/919 - 3s - loss: 1.5601 - accuracy: 0.4973 - val_loss: 3.5826 - val_accuracy: 0.0502\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1652/5000\n",
+      "919/919 - 3s - loss: 1.5693 - accuracy: 0.4929 - val_loss: 3.5721 - val_accuracy: 0.0501\n",
+      "Epoch 1653/5000\n",
+      "919/919 - 3s - loss: 1.5851 - accuracy: 0.4948 - val_loss: 3.5559 - val_accuracy: 0.0498\n",
+      "Epoch 1654/5000\n",
+      "919/919 - 3s - loss: 1.5754 - accuracy: 0.4917 - val_loss: 3.5488 - val_accuracy: 0.0498\n",
+      "Epoch 1655/5000\n",
+      "919/919 - 3s - loss: 1.6208 - accuracy: 0.4961 - val_loss: 3.5397 - val_accuracy: 0.0494\n",
+      "Epoch 1656/5000\n",
+      "919/919 - 3s - loss: 1.5751 - accuracy: 0.4916 - val_loss: 3.5618 - val_accuracy: 0.0495\n",
+      "Epoch 1657/5000\n",
+      "919/919 - 3s - loss: 1.5684 - accuracy: 0.4932 - val_loss: 3.5677 - val_accuracy: 0.0498\n",
+      "Epoch 1658/5000\n",
+      "919/919 - 3s - loss: 1.5735 - accuracy: 0.4963 - val_loss: 3.5671 - val_accuracy: 0.0494\n",
+      "Epoch 1659/5000\n",
+      "919/919 - 3s - loss: 1.5853 - accuracy: 0.4924 - val_loss: 3.5846 - val_accuracy: 0.0495\n",
+      "Epoch 1660/5000\n",
+      "919/919 - 3s - loss: 1.6906 - accuracy: 0.4961 - val_loss: 3.5816 - val_accuracy: 0.0494\n",
+      "Epoch 1661/5000\n",
+      "919/919 - 3s - loss: 1.6048 - accuracy: 0.4937 - val_loss: 3.5850 - val_accuracy: 0.0496\n",
+      "Epoch 1662/5000\n",
+      "919/919 - 3s - loss: 1.5931 - accuracy: 0.4941 - val_loss: 3.5746 - val_accuracy: 0.0496\n",
+      "Epoch 1663/5000\n",
+      "919/919 - 3s - loss: 1.5675 - accuracy: 0.4959 - val_loss: 3.5909 - val_accuracy: 0.0498\n",
+      "Epoch 1664/5000\n",
+      "919/919 - 3s - loss: 1.5692 - accuracy: 0.4924 - val_loss: 3.6165 - val_accuracy: 0.0498\n",
+      "Epoch 1665/5000\n",
+      "919/919 - 3s - loss: 1.5912 - accuracy: 0.4913 - val_loss: 3.6087 - val_accuracy: 0.0494\n",
+      "Epoch 1666/5000\n",
+      "919/919 - 3s - loss: 1.5646 - accuracy: 0.4944 - val_loss: 3.6192 - val_accuracy: 0.0498\n",
+      "Epoch 1667/5000\n",
+      "919/919 - 3s - loss: 1.6109 - accuracy: 0.4927 - val_loss: 3.6069 - val_accuracy: 0.0496\n",
+      "Epoch 1668/5000\n",
+      "919/919 - 3s - loss: 1.5913 - accuracy: 0.4929 - val_loss: 3.6016 - val_accuracy: 0.0491\n",
+      "Epoch 1669/5000\n",
+      "919/919 - 3s - loss: 1.5665 - accuracy: 0.4963 - val_loss: 3.5990 - val_accuracy: 0.0497\n",
+      "Epoch 1670/5000\n",
+      "919/919 - 3s - loss: 1.5865 - accuracy: 0.4924 - val_loss: 3.5747 - val_accuracy: 0.0493\n",
+      "Epoch 1671/5000\n",
+      "919/919 - 3s - loss: 1.5564 - accuracy: 0.4965 - val_loss: 3.5847 - val_accuracy: 0.0495\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1672/5000\n",
+      "919/919 - 3s - loss: 1.5912 - accuracy: 0.4903 - val_loss: 3.5715 - val_accuracy: 0.0496\n",
+      "Epoch 1673/5000\n",
+      "919/919 - 3s - loss: 1.5688 - accuracy: 0.4931 - val_loss: 3.5687 - val_accuracy: 0.0496\n",
+      "Epoch 1674/5000\n",
+      "919/919 - 3s - loss: 1.5654 - accuracy: 0.4952 - val_loss: 3.5697 - val_accuracy: 0.0498\n",
+      "Epoch 1675/5000\n",
+      "919/919 - 3s - loss: 1.5606 - accuracy: 0.4939 - val_loss: 3.5825 - val_accuracy: 0.0498\n",
+      "Epoch 1676/5000\n",
+      "919/919 - 3s - loss: 1.5768 - accuracy: 0.4898 - val_loss: 3.5723 - val_accuracy: 0.0494\n",
+      "Epoch 1677/5000\n",
+      "919/919 - 3s - loss: 1.5770 - accuracy: 0.4920 - val_loss: 3.5831 - val_accuracy: 0.0496\n",
+      "Epoch 1678/5000\n",
+      "919/919 - 3s - loss: 1.5612 - accuracy: 0.4922 - val_loss: 3.5795 - val_accuracy: 0.0496\n",
+      "Epoch 1679/5000\n",
+      "919/919 - 3s - loss: 1.5865 - accuracy: 0.4967 - val_loss: 3.5759 - val_accuracy: 0.0495\n",
+      "Epoch 1680/5000\n",
+      "919/919 - 3s - loss: 1.5727 - accuracy: 0.4939 - val_loss: 3.5817 - val_accuracy: 0.0492\n",
+      "Epoch 1681/5000\n",
+      "919/919 - 3s - loss: 1.6053 - accuracy: 0.4945 - val_loss: 3.5830 - val_accuracy: 0.0498\n",
+      "Epoch 1682/5000\n",
+      "919/919 - 3s - loss: 1.5655 - accuracy: 0.4948 - val_loss: 3.5915 - val_accuracy: 0.0499\n",
+      "Epoch 1683/5000\n",
+      "919/919 - 3s - loss: 1.5644 - accuracy: 0.4952 - val_loss: 3.5959 - val_accuracy: 0.0499\n",
+      "Epoch 1684/5000\n",
+      "919/919 - 3s - loss: 1.5774 - accuracy: 0.4962 - val_loss: 3.5917 - val_accuracy: 0.0498\n",
+      "Epoch 1685/5000\n",
+      "919/919 - 3s - loss: 1.5629 - accuracy: 0.4964 - val_loss: 3.5711 - val_accuracy: 0.0502\n",
+      "Epoch 1686/5000\n",
+      "919/919 - 3s - loss: 1.5697 - accuracy: 0.4946 - val_loss: 3.5960 - val_accuracy: 0.0500\n",
+      "Epoch 1687/5000\n",
+      "919/919 - 3s - loss: 1.5689 - accuracy: 0.4953 - val_loss: 3.5901 - val_accuracy: 0.0499\n",
+      "Epoch 1688/5000\n",
+      "919/919 - 3s - loss: 1.5627 - accuracy: 0.4951 - val_loss: 3.6027 - val_accuracy: 0.0500\n",
+      "Epoch 1689/5000\n",
+      "919/919 - 3s - loss: 1.5832 - accuracy: 0.4934 - val_loss: 3.5886 - val_accuracy: 0.0496\n",
+      "Epoch 1690/5000\n",
+      "919/919 - 3s - loss: 1.5758 - accuracy: 0.4918 - val_loss: 3.5789 - val_accuracy: 0.0493\n",
+      "Epoch 1691/5000\n",
+      "919/919 - 3s - loss: 1.5762 - accuracy: 0.4944 - val_loss: 3.5905 - val_accuracy: 0.0493\n",
+      "Epoch 1692/5000\n",
+      "919/919 - 3s - loss: 1.5650 - accuracy: 0.4941 - val_loss: 3.5951 - val_accuracy: 0.0490\n",
+      "Epoch 1693/5000\n",
+      "919/919 - 3s - loss: 1.5526 - accuracy: 0.4980 - val_loss: 3.6003 - val_accuracy: 0.0495\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1694/5000\n",
+      "919/919 - 3s - loss: 1.5965 - accuracy: 0.4967 - val_loss: 3.6013 - val_accuracy: 0.0497\n",
+      "Epoch 1695/5000\n",
+      "919/919 - 3s - loss: 1.5833 - accuracy: 0.4951 - val_loss: 3.6101 - val_accuracy: 0.0501\n",
+      "Epoch 1696/5000\n",
+      "919/919 - 3s - loss: 1.5695 - accuracy: 0.4973 - val_loss: 3.6028 - val_accuracy: 0.0498\n",
+      "Epoch 1697/5000\n",
+      "919/919 - 3s - loss: 1.5643 - accuracy: 0.4974 - val_loss: 3.6172 - val_accuracy: 0.0500\n",
+      "Epoch 1698/5000\n",
+      "919/919 - 3s - loss: 1.5941 - accuracy: 0.4931 - val_loss: 3.6130 - val_accuracy: 0.0500\n",
+      "Epoch 1699/5000\n",
+      "919/919 - 3s - loss: 1.5748 - accuracy: 0.4968 - val_loss: 3.6032 - val_accuracy: 0.0498\n",
+      "Epoch 1700/5000\n",
+      "919/919 - 3s - loss: 1.5721 - accuracy: 0.4914 - val_loss: 3.6109 - val_accuracy: 0.0499\n",
+      "Epoch 1701/5000\n",
+      "919/919 - 3s - loss: 1.5581 - accuracy: 0.4949 - val_loss: 3.6040 - val_accuracy: 0.0499\n",
+      "Epoch 1702/5000\n",
+      "919/919 - 3s - loss: 1.5915 - accuracy: 0.4980 - val_loss: 3.6007 - val_accuracy: 0.0500\n",
+      "Epoch 1703/5000\n",
+      "919/919 - 3s - loss: 1.5708 - accuracy: 0.4948 - val_loss: 3.6089 - val_accuracy: 0.0500\n",
+      "Epoch 1704/5000\n",
+      "919/919 - 3s - loss: 1.5604 - accuracy: 0.4997 - val_loss: 3.5995 - val_accuracy: 0.0498\n",
+      "Epoch 1705/5000\n",
+      "919/919 - 3s - loss: 1.5704 - accuracy: 0.4933 - val_loss: 3.5817 - val_accuracy: 0.0498\n",
+      "Epoch 1706/5000\n",
+      "919/919 - 3s - loss: 1.5574 - accuracy: 0.4980 - val_loss: 3.5914 - val_accuracy: 0.0497\n",
+      "Epoch 1707/5000\n",
+      "919/919 - 3s - loss: 1.5562 - accuracy: 0.4973 - val_loss: 3.5757 - val_accuracy: 0.0498\n",
+      "Epoch 1708/5000\n",
+      "919/919 - 3s - loss: 1.5675 - accuracy: 0.4918 - val_loss: 3.5559 - val_accuracy: 0.0498\n",
+      "Epoch 1709/5000\n",
+      "919/919 - 3s - loss: 1.5602 - accuracy: 0.4948 - val_loss: 3.5783 - val_accuracy: 0.0500\n",
+      "Epoch 1710/5000\n",
+      "919/919 - 3s - loss: 1.5746 - accuracy: 0.4959 - val_loss: 3.5784 - val_accuracy: 0.0494\n",
+      "Epoch 1711/5000\n",
+      "919/919 - 3s - loss: 1.5629 - accuracy: 0.4969 - val_loss: 3.5761 - val_accuracy: 0.0496\n",
+      "Epoch 1712/5000\n",
+      "919/919 - 3s - loss: 1.5528 - accuracy: 0.4972 - val_loss: 3.5769 - val_accuracy: 0.0496\n",
+      "Epoch 1713/5000\n",
+      "919/919 - 3s - loss: 1.5716 - accuracy: 0.4959 - val_loss: 3.5877 - val_accuracy: 0.0497\n",
+      "Epoch 1714/5000\n",
+      "919/919 - 3s - loss: 1.5601 - accuracy: 0.4950 - val_loss: 3.5943 - val_accuracy: 0.0499\n",
+      "Epoch 1715/5000\n",
+      "919/919 - 3s - loss: 1.5523 - accuracy: 0.4940 - val_loss: 3.5881 - val_accuracy: 0.0499\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1716/5000\n",
+      "919/919 - 3s - loss: 1.5649 - accuracy: 0.4967 - val_loss: 3.5953 - val_accuracy: 0.0499\n",
+      "Epoch 1717/5000\n",
+      "919/919 - 3s - loss: 1.5695 - accuracy: 0.4940 - val_loss: 3.5913 - val_accuracy: 0.0502\n",
+      "Epoch 1718/5000\n",
+      "919/919 - 3s - loss: 1.5670 - accuracy: 0.4967 - val_loss: 3.5848 - val_accuracy: 0.0498\n",
+      "Epoch 1719/5000\n",
+      "919/919 - 3s - loss: 1.5504 - accuracy: 0.4981 - val_loss: 3.5950 - val_accuracy: 0.0501\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1720/5000\n",
+      "919/919 - 3s - loss: 1.5715 - accuracy: 0.4948 - val_loss: 3.5877 - val_accuracy: 0.0509\n",
+      "Epoch 1721/5000\n",
+      "919/919 - 3s - loss: 1.5516 - accuracy: 0.4937 - val_loss: 3.5924 - val_accuracy: 0.0504\n",
+      "Epoch 1722/5000\n",
+      "919/919 - 3s - loss: 1.5502 - accuracy: 0.4990 - val_loss: 3.6001 - val_accuracy: 0.0507\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1723/5000\n",
+      "919/919 - 3s - loss: 1.5668 - accuracy: 0.4982 - val_loss: 3.6057 - val_accuracy: 0.0503\n",
+      "Epoch 1724/5000\n",
+      "919/919 - 3s - loss: 1.5778 - accuracy: 0.4916 - val_loss: 3.5909 - val_accuracy: 0.0504\n",
+      "Epoch 1725/5000\n",
+      "919/919 - 3s - loss: 1.5654 - accuracy: 0.4982 - val_loss: 3.5895 - val_accuracy: 0.0507\n",
+      "Epoch 1726/5000\n",
+      "919/919 - 3s - loss: 1.5556 - accuracy: 0.4962 - val_loss: 3.5919 - val_accuracy: 0.0506\n",
+      "Epoch 1727/5000\n",
+      "919/919 - 3s - loss: 1.5594 - accuracy: 0.4953 - val_loss: 3.5967 - val_accuracy: 0.0505\n",
+      "Epoch 1728/5000\n",
+      "919/919 - 3s - loss: 1.5543 - accuracy: 0.4956 - val_loss: 3.5848 - val_accuracy: 0.0504\n",
+      "Epoch 1729/5000\n",
+      "919/919 - 3s - loss: 1.5649 - accuracy: 0.4910 - val_loss: 3.5872 - val_accuracy: 0.0498\n",
+      "Epoch 1730/5000\n",
+      "919/919 - 3s - loss: 1.5540 - accuracy: 0.4938 - val_loss: 3.5859 - val_accuracy: 0.0499\n",
+      "Epoch 1731/5000\n",
+      "919/919 - 3s - loss: 1.5670 - accuracy: 0.4931 - val_loss: 3.5912 - val_accuracy: 0.0498\n",
+      "Epoch 1732/5000\n",
+      "919/919 - 3s - loss: 1.5490 - accuracy: 0.4964 - val_loss: 3.5860 - val_accuracy: 0.0501\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1733/5000\n",
+      "919/919 - 3s - loss: 1.5591 - accuracy: 0.4939 - val_loss: 3.5900 - val_accuracy: 0.0501\n",
+      "Epoch 1734/5000\n",
+      "919/919 - 3s - loss: 1.5989 - accuracy: 0.4954 - val_loss: 3.5903 - val_accuracy: 0.0504\n",
+      "Epoch 1735/5000\n",
+      "919/919 - 3s - loss: 1.5598 - accuracy: 0.4962 - val_loss: 3.5869 - val_accuracy: 0.0507\n",
+      "Epoch 1736/5000\n",
+      "919/919 - 3s - loss: 1.5585 - accuracy: 0.4967 - val_loss: 3.5825 - val_accuracy: 0.0507\n",
+      "Epoch 1737/5000\n",
+      "919/919 - 3s - loss: 1.5587 - accuracy: 0.4980 - val_loss: 3.5662 - val_accuracy: 0.0504\n",
+      "Epoch 1738/5000\n",
+      "919/919 - 3s - loss: 1.5590 - accuracy: 0.4998 - val_loss: 3.5573 - val_accuracy: 0.0503\n",
+      "Epoch 1739/5000\n",
+      "919/919 - 3s - loss: 1.5580 - accuracy: 0.4981 - val_loss: 3.5567 - val_accuracy: 0.0503\n",
+      "Epoch 1740/5000\n",
+      "919/919 - 3s - loss: 1.5870 - accuracy: 0.4938 - val_loss: 3.5680 - val_accuracy: 0.0500\n",
+      "Epoch 1741/5000\n",
+      "919/919 - 3s - loss: 1.5644 - accuracy: 0.4968 - val_loss: 3.5816 - val_accuracy: 0.0506\n",
+      "Epoch 1742/5000\n",
+      "919/919 - 3s - loss: 1.5688 - accuracy: 0.4944 - val_loss: 3.5933 - val_accuracy: 0.0507\n",
+      "Epoch 1743/5000\n",
+      "919/919 - 3s - loss: 1.5457 - accuracy: 0.4939 - val_loss: 3.6046 - val_accuracy: 0.0507\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1744/5000\n",
+      "919/919 - 3s - loss: 1.5725 - accuracy: 0.4965 - val_loss: 3.6105 - val_accuracy: 0.0509\n",
+      "Epoch 1745/5000\n",
+      "919/919 - 3s - loss: 1.5709 - accuracy: 0.4945 - val_loss: 3.6072 - val_accuracy: 0.0507\n",
+      "Epoch 1746/5000\n",
+      "919/919 - 3s - loss: 1.5642 - accuracy: 0.4959 - val_loss: 3.5911 - val_accuracy: 0.0508\n",
+      "Epoch 1747/5000\n",
+      "919/919 - 3s - loss: 1.5372 - accuracy: 0.5000 - val_loss: 3.5860 - val_accuracy: 0.0507\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1748/5000\n",
+      "919/919 - 3s - loss: 1.6188 - accuracy: 0.4990 - val_loss: 3.5812 - val_accuracy: 0.0507\n",
+      "Epoch 1749/5000\n",
+      "919/919 - 3s - loss: 1.5518 - accuracy: 0.4979 - val_loss: 3.5897 - val_accuracy: 0.0507\n",
+      "Epoch 1750/5000\n",
+      "919/919 - 3s - loss: 1.5412 - accuracy: 0.5014 - val_loss: 3.6146 - val_accuracy: 0.0508\n",
+      "Epoch 1751/5000\n",
+      "919/919 - 3s - loss: 1.5570 - accuracy: 0.4942 - val_loss: 3.6005 - val_accuracy: 0.0507\n",
+      "Epoch 1752/5000\n",
+      "919/919 - 3s - loss: 1.5624 - accuracy: 0.5002 - val_loss: 3.5828 - val_accuracy: 0.0507\n",
+      "Epoch 1753/5000\n",
+      "919/919 - 3s - loss: 1.5432 - accuracy: 0.4991 - val_loss: 3.5828 - val_accuracy: 0.0504\n",
+      "Epoch 1754/5000\n",
+      "919/919 - 3s - loss: 1.5720 - accuracy: 0.4971 - val_loss: 3.5823 - val_accuracy: 0.0507\n",
+      "Epoch 1755/5000\n",
+      "919/919 - 3s - loss: 1.5929 - accuracy: 0.4997 - val_loss: 3.5936 - val_accuracy: 0.0512\n",
+      "Epoch 1756/5000\n",
+      "919/919 - 3s - loss: 1.5610 - accuracy: 0.4969 - val_loss: 3.5833 - val_accuracy: 0.0509\n",
+      "Epoch 1757/5000\n",
+      "919/919 - 3s - loss: 1.5812 - accuracy: 0.4966 - val_loss: 3.5971 - val_accuracy: 0.0507\n",
+      "Epoch 1758/5000\n",
+      "919/919 - 3s - loss: 1.5517 - accuracy: 0.4971 - val_loss: 3.6072 - val_accuracy: 0.0506\n",
+      "Epoch 1759/5000\n",
+      "919/919 - 3s - loss: 1.5455 - accuracy: 0.4967 - val_loss: 3.6101 - val_accuracy: 0.0511\n",
+      "Epoch 1760/5000\n",
+      "919/919 - 3s - loss: 1.5657 - accuracy: 0.5012 - val_loss: 3.6080 - val_accuracy: 0.0510\n",
+      "Epoch 1761/5000\n",
+      "919/919 - 3s - loss: 1.5752 - accuracy: 0.4946 - val_loss: 3.6156 - val_accuracy: 0.0503\n",
+      "Epoch 1762/5000\n",
+      "919/919 - 3s - loss: 1.5525 - accuracy: 0.4993 - val_loss: 3.6098 - val_accuracy: 0.0507\n",
+      "Epoch 1763/5000\n",
+      "919/919 - 3s - loss: 1.5819 - accuracy: 0.4958 - val_loss: 3.6183 - val_accuracy: 0.0510\n",
+      "Epoch 1764/5000\n",
+      "919/919 - 3s - loss: 1.5676 - accuracy: 0.4960 - val_loss: 3.6218 - val_accuracy: 0.0515\n",
+      "Epoch 1765/5000\n",
+      "919/919 - 3s - loss: 1.5489 - accuracy: 0.5013 - val_loss: 3.6198 - val_accuracy: 0.0512\n",
+      "Epoch 1766/5000\n",
+      "919/919 - 3s - loss: 1.5527 - accuracy: 0.4939 - val_loss: 3.6167 - val_accuracy: 0.0516\n",
+      "Epoch 1767/5000\n",
+      "919/919 - 3s - loss: 1.5600 - accuracy: 0.4975 - val_loss: 3.6068 - val_accuracy: 0.0511\n",
+      "Epoch 1768/5000\n",
+      "919/919 - 3s - loss: 1.5439 - accuracy: 0.5001 - val_loss: 3.6019 - val_accuracy: 0.0513\n",
+      "Epoch 1769/5000\n",
+      "919/919 - 3s - loss: 1.6879 - accuracy: 0.4986 - val_loss: 3.6065 - val_accuracy: 0.0515\n",
+      "Epoch 1770/5000\n",
+      "919/919 - 3s - loss: 1.5474 - accuracy: 0.5031 - val_loss: 3.6034 - val_accuracy: 0.0520\n",
+      "Epoch 1771/5000\n",
+      "919/919 - 3s - loss: 1.6870 - accuracy: 0.4987 - val_loss: 3.6155 - val_accuracy: 0.0521\n",
+      "Epoch 1772/5000\n",
+      "919/919 - 3s - loss: 1.6105 - accuracy: 0.4972 - val_loss: 3.6178 - val_accuracy: 0.0514\n",
+      "Epoch 1773/5000\n",
+      "919/919 - 3s - loss: 1.5690 - accuracy: 0.4995 - val_loss: 3.6167 - val_accuracy: 0.0515\n",
+      "Epoch 1774/5000\n",
+      "919/919 - 3s - loss: 1.5574 - accuracy: 0.4967 - val_loss: 3.6126 - val_accuracy: 0.0513\n",
+      "Epoch 1775/5000\n",
+      "919/919 - 3s - loss: 1.5634 - accuracy: 0.4995 - val_loss: 3.6178 - val_accuracy: 0.0513\n",
+      "Epoch 1776/5000\n",
+      "919/919 - 3s - loss: 1.5401 - accuracy: 0.4993 - val_loss: 3.6083 - val_accuracy: 0.0517\n",
+      "Epoch 1777/5000\n",
+      "919/919 - 3s - loss: 1.5448 - accuracy: 0.4989 - val_loss: 3.6037 - val_accuracy: 0.0519\n",
+      "Epoch 1778/5000\n",
+      "919/919 - 3s - loss: 1.5451 - accuracy: 0.5006 - val_loss: 3.6097 - val_accuracy: 0.0519\n",
+      "Epoch 1779/5000\n",
+      "919/919 - 3s - loss: 1.5427 - accuracy: 0.5006 - val_loss: 3.6092 - val_accuracy: 0.0521\n",
+      "Epoch 1780/5000\n",
+      "919/919 - 3s - loss: 1.5391 - accuracy: 0.5032 - val_loss: 3.6059 - val_accuracy: 0.0521\n",
+      "Epoch 1781/5000\n",
+      "919/919 - 3s - loss: 1.5506 - accuracy: 0.4937 - val_loss: 3.6001 - val_accuracy: 0.0519\n",
+      "Epoch 1782/5000\n",
+      "919/919 - 3s - loss: 1.5460 - accuracy: 0.4997 - val_loss: 3.6069 - val_accuracy: 0.0516\n",
+      "Epoch 1783/5000\n",
+      "919/919 - 3s - loss: 1.5495 - accuracy: 0.4990 - val_loss: 3.6064 - val_accuracy: 0.0517\n",
+      "Epoch 1784/5000\n",
+      "919/919 - 3s - loss: 1.5769 - accuracy: 0.4953 - val_loss: 3.6023 - val_accuracy: 0.0523\n",
+      "Epoch 1785/5000\n",
+      "919/919 - 3s - loss: 1.5429 - accuracy: 0.4984 - val_loss: 3.6033 - val_accuracy: 0.0518\n",
+      "Epoch 1786/5000\n",
+      "919/919 - 3s - loss: 1.5623 - accuracy: 0.4986 - val_loss: 3.5881 - val_accuracy: 0.0517\n",
+      "Epoch 1787/5000\n",
+      "919/919 - 3s - loss: 1.7321 - accuracy: 0.4982 - val_loss: 3.5982 - val_accuracy: 0.0521\n",
+      "Epoch 1788/5000\n",
+      "919/919 - 3s - loss: 1.5479 - accuracy: 0.5003 - val_loss: 3.5890 - val_accuracy: 0.0523\n",
+      "Epoch 1789/5000\n",
+      "919/919 - 3s - loss: 1.5807 - accuracy: 0.4963 - val_loss: 3.5800 - val_accuracy: 0.0521\n",
+      "Epoch 1790/5000\n",
+      "919/919 - 3s - loss: 1.5447 - accuracy: 0.5008 - val_loss: 3.5788 - val_accuracy: 0.0517\n",
+      "Epoch 1791/5000\n",
+      "919/919 - 3s - loss: 1.5450 - accuracy: 0.4981 - val_loss: 3.5824 - val_accuracy: 0.0522\n",
+      "Epoch 1792/5000\n",
+      "919/919 - 3s - loss: 1.5611 - accuracy: 0.4946 - val_loss: 3.5866 - val_accuracy: 0.0518\n",
+      "Epoch 1793/5000\n",
+      "919/919 - 3s - loss: 1.5415 - accuracy: 0.5005 - val_loss: 3.5921 - val_accuracy: 0.0520\n",
+      "Epoch 1794/5000\n",
+      "919/919 - 3s - loss: 1.5403 - accuracy: 0.4993 - val_loss: 3.5981 - val_accuracy: 0.0523\n",
+      "Epoch 1795/5000\n",
+      "919/919 - 3s - loss: 1.5746 - accuracy: 0.4982 - val_loss: 3.6065 - val_accuracy: 0.0520\n",
+      "Epoch 1796/5000\n",
+      "919/919 - 3s - loss: 1.5697 - accuracy: 0.4986 - val_loss: 3.6046 - val_accuracy: 0.0524\n",
+      "Epoch 1797/5000\n",
+      "919/919 - 3s - loss: 1.5421 - accuracy: 0.4965 - val_loss: 3.6244 - val_accuracy: 0.0531\n",
+      "Epoch 1798/5000\n",
+      "919/919 - 3s - loss: 1.5626 - accuracy: 0.4967 - val_loss: 3.6152 - val_accuracy: 0.0529\n",
+      "Epoch 1799/5000\n",
+      "919/919 - 3s - loss: 1.5628 - accuracy: 0.5010 - val_loss: 3.6158 - val_accuracy: 0.0526\n",
+      "Epoch 1800/5000\n",
+      "919/919 - 3s - loss: 1.6177 - accuracy: 0.4982 - val_loss: 3.6288 - val_accuracy: 0.0526\n",
+      "Epoch 1801/5000\n",
+      "919/919 - 3s - loss: 1.5477 - accuracy: 0.5014 - val_loss: 3.6175 - val_accuracy: 0.0527\n",
+      "Epoch 1802/5000\n",
+      "919/919 - 3s - loss: 1.5435 - accuracy: 0.5041 - val_loss: 3.6227 - val_accuracy: 0.0529\n",
+      "Epoch 1803/5000\n",
+      "919/919 - 3s - loss: 1.5301 - accuracy: 0.5046 - val_loss: 3.6275 - val_accuracy: 0.0528\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1804/5000\n",
+      "919/919 - 3s - loss: 1.5481 - accuracy: 0.5013 - val_loss: 3.6083 - val_accuracy: 0.0530\n",
+      "Epoch 1805/5000\n",
+      "919/919 - 3s - loss: 1.5795 - accuracy: 0.5020 - val_loss: 3.6251 - val_accuracy: 0.0529\n",
+      "Epoch 1806/5000\n",
+      "919/919 - 3s - loss: 1.5369 - accuracy: 0.5017 - val_loss: 3.6057 - val_accuracy: 0.0529\n",
+      "Epoch 1807/5000\n",
+      "919/919 - 3s - loss: 1.5357 - accuracy: 0.5028 - val_loss: 3.5948 - val_accuracy: 0.0530\n",
+      "Epoch 1808/5000\n",
+      "919/919 - 3s - loss: 1.5365 - accuracy: 0.4991 - val_loss: 3.6048 - val_accuracy: 0.0532\n",
+      "Epoch 1809/5000\n",
+      "919/919 - 3s - loss: 1.5431 - accuracy: 0.4967 - val_loss: 3.6030 - val_accuracy: 0.0527\n",
+      "Epoch 1810/5000\n",
+      "919/919 - 3s - loss: 1.5358 - accuracy: 0.5034 - val_loss: 3.6015 - val_accuracy: 0.0533\n",
+      "Epoch 1811/5000\n",
+      "919/919 - 3s - loss: 1.5420 - accuracy: 0.5009 - val_loss: 3.5873 - val_accuracy: 0.0531\n",
+      "Epoch 1812/5000\n",
+      "919/919 - 3s - loss: 1.5355 - accuracy: 0.5008 - val_loss: 3.5915 - val_accuracy: 0.0532\n",
+      "Epoch 1813/5000\n",
+      "919/919 - 3s - loss: 1.5402 - accuracy: 0.5046 - val_loss: 3.5974 - val_accuracy: 0.0526\n",
+      "Epoch 1814/5000\n",
+      "919/919 - 3s - loss: 1.5472 - accuracy: 0.5008 - val_loss: 3.5975 - val_accuracy: 0.0532\n",
+      "Epoch 1815/5000\n",
+      "919/919 - 3s - loss: 1.5444 - accuracy: 0.5015 - val_loss: 3.5823 - val_accuracy: 0.0534\n",
+      "Epoch 1816/5000\n",
+      "919/919 - 3s - loss: 1.5367 - accuracy: 0.5015 - val_loss: 3.5819 - val_accuracy: 0.0534\n",
+      "Epoch 1817/5000\n",
+      "919/919 - 4s - loss: 1.5782 - accuracy: 0.4974 - val_loss: 3.5832 - val_accuracy: 0.0535\n",
+      "Epoch 1818/5000\n",
+      "919/919 - 4s - loss: 1.5405 - accuracy: 0.5006 - val_loss: 3.5945 - val_accuracy: 0.0534\n",
+      "Epoch 1819/5000\n",
+      "919/919 - 3s - loss: 1.5308 - accuracy: 0.5001 - val_loss: 3.6116 - val_accuracy: 0.0532\n",
+      "Epoch 1820/5000\n",
+      "919/919 - 3s - loss: 1.5396 - accuracy: 0.5007 - val_loss: 3.6163 - val_accuracy: 0.0534\n",
+      "Epoch 1821/5000\n",
+      "919/919 - 4s - loss: 1.5445 - accuracy: 0.5022 - val_loss: 3.6039 - val_accuracy: 0.0533\n",
+      "Epoch 1822/5000\n",
+      "919/919 - 4s - loss: 1.5368 - accuracy: 0.5016 - val_loss: 3.5986 - val_accuracy: 0.0540\n",
+      "Epoch 1823/5000\n",
+      "919/919 - 3s - loss: 1.5227 - accuracy: 0.5008 - val_loss: 3.6002 - val_accuracy: 0.0541\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1824/5000\n",
+      "919/919 - 4s - loss: 1.5484 - accuracy: 0.4980 - val_loss: 3.5991 - val_accuracy: 0.0539\n",
+      "Epoch 1825/5000\n",
+      "919/919 - 4s - loss: 1.5407 - accuracy: 0.5056 - val_loss: 3.5998 - val_accuracy: 0.0543\n",
+      "Epoch 1826/5000\n",
+      "919/919 - 3s - loss: 1.5298 - accuracy: 0.5031 - val_loss: 3.6086 - val_accuracy: 0.0543\n",
+      "Epoch 1827/5000\n",
+      "919/919 - 3s - loss: 1.6193 - accuracy: 0.5016 - val_loss: 3.5953 - val_accuracy: 0.0541\n",
+      "Epoch 1828/5000\n",
+      "919/919 - 4s - loss: 1.5129 - accuracy: 0.5052 - val_loss: 3.6005 - val_accuracy: 0.0546\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1829/5000\n",
+      "919/919 - 4s - loss: 1.5192 - accuracy: 0.5056 - val_loss: 3.5900 - val_accuracy: 0.0550\n",
+      "Epoch 1830/5000\n",
+      "919/919 - 3s - loss: 1.5387 - accuracy: 0.4965 - val_loss: 3.5900 - val_accuracy: 0.0549\n",
+      "Epoch 1831/5000\n",
+      "919/919 - 3s - loss: 1.5335 - accuracy: 0.5058 - val_loss: 3.6158 - val_accuracy: 0.0552\n",
+      "Epoch 1832/5000\n",
+      "919/919 - 4s - loss: 1.5643 - accuracy: 0.5009 - val_loss: 3.6035 - val_accuracy: 0.0552\n",
+      "Epoch 1833/5000\n",
+      "919/919 - 4s - loss: 1.5348 - accuracy: 0.5052 - val_loss: 3.6073 - val_accuracy: 0.0552\n",
+      "Epoch 1834/5000\n",
+      "919/919 - 3s - loss: 1.5244 - accuracy: 0.5036 - val_loss: 3.6126 - val_accuracy: 0.0556\n",
+      "Epoch 1835/5000\n",
+      "919/919 - 3s - loss: 1.5329 - accuracy: 0.5039 - val_loss: 3.6071 - val_accuracy: 0.0552\n",
+      "Epoch 1836/5000\n",
+      "919/919 - 4s - loss: 1.5339 - accuracy: 0.5044 - val_loss: 3.6008 - val_accuracy: 0.0551\n",
+      "Epoch 1837/5000\n",
+      "919/919 - 4s - loss: 1.5339 - accuracy: 0.5067 - val_loss: 3.6117 - val_accuracy: 0.0546\n",
+      "Epoch 1838/5000\n",
+      "919/919 - 3s - loss: 1.5257 - accuracy: 0.5046 - val_loss: 3.6266 - val_accuracy: 0.0550\n",
+      "Epoch 1839/5000\n",
+      "919/919 - 3s - loss: 1.5753 - accuracy: 0.5048 - val_loss: 3.6352 - val_accuracy: 0.0548\n",
+      "Epoch 1840/5000\n",
+      "919/919 - 4s - loss: 1.5328 - accuracy: 0.5037 - val_loss: 3.6241 - val_accuracy: 0.0545\n",
+      "Epoch 1841/5000\n",
+      "919/919 - 4s - loss: 1.5332 - accuracy: 0.5042 - val_loss: 3.6314 - val_accuracy: 0.0546\n",
+      "Epoch 1842/5000\n",
+      "919/919 - 3s - loss: 1.5403 - accuracy: 0.5035 - val_loss: 3.6191 - val_accuracy: 0.0549\n",
+      "Epoch 1843/5000\n",
+      "919/919 - 3s - loss: 1.5482 - accuracy: 0.5029 - val_loss: 3.6149 - val_accuracy: 0.0548\n",
+      "Epoch 1844/5000\n",
+      "919/919 - 4s - loss: 1.5269 - accuracy: 0.5057 - val_loss: 3.5917 - val_accuracy: 0.0549\n",
+      "Epoch 1845/5000\n",
+      "919/919 - 4s - loss: 1.5554 - accuracy: 0.5037 - val_loss: 3.5955 - val_accuracy: 0.0553\n",
+      "Epoch 1846/5000\n",
+      "919/919 - 3s - loss: 1.5677 - accuracy: 0.5037 - val_loss: 3.6101 - val_accuracy: 0.0554\n",
+      "Epoch 1847/5000\n",
+      "919/919 - 3s - loss: 1.5349 - accuracy: 0.5052 - val_loss: 3.5999 - val_accuracy: 0.0556\n",
+      "Epoch 1848/5000\n",
+      "919/919 - 4s - loss: 1.5395 - accuracy: 0.5093 - val_loss: 3.5991 - val_accuracy: 0.0558\n",
+      "Epoch 1849/5000\n",
+      "919/919 - 4s - loss: 1.5375 - accuracy: 0.5042 - val_loss: 3.6010 - val_accuracy: 0.0559\n",
+      "Epoch 1850/5000\n",
+      "919/919 - 3s - loss: 1.5373 - accuracy: 0.5053 - val_loss: 3.5940 - val_accuracy: 0.0558\n",
+      "Epoch 1851/5000\n",
+      "919/919 - 3s - loss: 1.5875 - accuracy: 0.5056 - val_loss: 3.5887 - val_accuracy: 0.0561\n",
+      "Epoch 1852/5000\n",
+      "919/919 - 4s - loss: 1.5545 - accuracy: 0.5028 - val_loss: 3.5948 - val_accuracy: 0.0555\n",
+      "Epoch 1853/5000\n",
+      "919/919 - 3s - loss: 1.5329 - accuracy: 0.5015 - val_loss: 3.5979 - val_accuracy: 0.0554\n",
+      "Epoch 1854/5000\n",
+      "919/919 - 3s - loss: 1.5158 - accuracy: 0.5024 - val_loss: 3.6065 - val_accuracy: 0.0554\n",
+      "Epoch 1855/5000\n",
+      "919/919 - 3s - loss: 1.5310 - accuracy: 0.5089 - val_loss: 3.5998 - val_accuracy: 0.0553\n",
+      "Epoch 1856/5000\n",
+      "919/919 - 4s - loss: 1.5281 - accuracy: 0.5067 - val_loss: 3.6232 - val_accuracy: 0.0553\n",
+      "Epoch 1857/5000\n",
+      "919/919 - 3s - loss: 1.5288 - accuracy: 0.5078 - val_loss: 3.6197 - val_accuracy: 0.0552\n",
+      "Epoch 1858/5000\n",
+      "919/919 - 3s - loss: 1.5628 - accuracy: 0.5069 - val_loss: 3.6267 - val_accuracy: 0.0558\n",
+      "Epoch 1859/5000\n",
+      "919/919 - 4s - loss: 1.5337 - accuracy: 0.5039 - val_loss: 3.6019 - val_accuracy: 0.0555\n",
+      "Epoch 1860/5000\n",
+      "919/919 - 4s - loss: 1.5230 - accuracy: 0.5097 - val_loss: 3.6077 - val_accuracy: 0.0558\n",
+      "Epoch 1861/5000\n",
+      "919/919 - 3s - loss: 1.6026 - accuracy: 0.5047 - val_loss: 3.6052 - val_accuracy: 0.0553\n",
+      "Epoch 1862/5000\n",
+      "919/919 - 3s - loss: 1.5323 - accuracy: 0.5040 - val_loss: 3.5944 - val_accuracy: 0.0555\n",
+      "Epoch 1863/5000\n",
+      "919/919 - 4s - loss: 1.5274 - accuracy: 0.5037 - val_loss: 3.5997 - val_accuracy: 0.0555\n",
+      "Epoch 1864/5000\n",
+      "919/919 - 4s - loss: 1.5297 - accuracy: 0.5061 - val_loss: 3.6031 - val_accuracy: 0.0557\n",
+      "Epoch 1865/5000\n",
+      "919/919 - 3s - loss: 1.5374 - accuracy: 0.5074 - val_loss: 3.6123 - val_accuracy: 0.0553\n",
+      "Epoch 1866/5000\n",
+      "919/919 - 3s - loss: 1.5307 - accuracy: 0.5029 - val_loss: 3.6257 - val_accuracy: 0.0552\n",
+      "Epoch 1867/5000\n",
+      "919/919 - 4s - loss: 1.5209 - accuracy: 0.5084 - val_loss: 3.6118 - val_accuracy: 0.0556\n",
+      "Epoch 1868/5000\n",
+      "919/919 - 4s - loss: 1.5075 - accuracy: 0.5090 - val_loss: 3.6228 - val_accuracy: 0.0559\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1869/5000\n",
+      "919/919 - 3s - loss: 1.5592 - accuracy: 0.5057 - val_loss: 3.6052 - val_accuracy: 0.0560\n",
+      "Epoch 1870/5000\n",
+      "919/919 - 3s - loss: 1.5284 - accuracy: 0.5031 - val_loss: 3.6098 - val_accuracy: 0.0558\n",
+      "Epoch 1871/5000\n",
+      "919/919 - 4s - loss: 1.5139 - accuracy: 0.5054 - val_loss: 3.6115 - val_accuracy: 0.0559\n",
+      "Epoch 1872/5000\n",
+      "919/919 - 4s - loss: 1.5332 - accuracy: 0.5037 - val_loss: 3.6122 - val_accuracy: 0.0558\n",
+      "Epoch 1873/5000\n",
+      "919/919 - 3s - loss: 1.5523 - accuracy: 0.5072 - val_loss: 3.6123 - val_accuracy: 0.0560\n",
+      "Epoch 1874/5000\n",
+      "919/919 - 3s - loss: 1.5481 - accuracy: 0.5134 - val_loss: 3.6043 - val_accuracy: 0.0562\n",
+      "Epoch 1875/5000\n",
+      "919/919 - 4s - loss: 1.5176 - accuracy: 0.5074 - val_loss: 3.5914 - val_accuracy: 0.0552\n",
+      "Epoch 1876/5000\n",
+      "919/919 - 4s - loss: 1.5291 - accuracy: 0.5119 - val_loss: 3.5922 - val_accuracy: 0.0555\n",
+      "Epoch 1877/5000\n",
+      "919/919 - 3s - loss: 1.5310 - accuracy: 0.5037 - val_loss: 3.5900 - val_accuracy: 0.0556\n",
+      "Epoch 1878/5000\n",
+      "919/919 - 3s - loss: 1.5259 - accuracy: 0.5054 - val_loss: 3.5997 - val_accuracy: 0.0560\n",
+      "Epoch 1879/5000\n",
+      "919/919 - 4s - loss: 1.5494 - accuracy: 0.5054 - val_loss: 3.5839 - val_accuracy: 0.0560\n",
+      "Epoch 1880/5000\n",
+      "919/919 - 4s - loss: 1.5620 - accuracy: 0.5105 - val_loss: 3.5755 - val_accuracy: 0.0560\n",
+      "Epoch 1881/5000\n",
+      "919/919 - 3s - loss: 1.5269 - accuracy: 0.5086 - val_loss: 3.5849 - val_accuracy: 0.0561\n",
+      "Epoch 1882/5000\n",
+      "919/919 - 3s - loss: 1.5139 - accuracy: 0.5103 - val_loss: 3.5883 - val_accuracy: 0.0565\n",
+      "Epoch 1883/5000\n",
+      "919/919 - 4s - loss: 1.5265 - accuracy: 0.5088 - val_loss: 3.6019 - val_accuracy: 0.0564\n",
+      "Epoch 1884/5000\n",
+      "919/919 - 4s - loss: 1.5323 - accuracy: 0.5064 - val_loss: 3.5976 - val_accuracy: 0.0560\n",
+      "Epoch 1885/5000\n",
+      "919/919 - 3s - loss: 1.5217 - accuracy: 0.5061 - val_loss: 3.6037 - val_accuracy: 0.0559\n",
+      "Epoch 1886/5000\n",
+      "919/919 - 3s - loss: 1.5313 - accuracy: 0.5073 - val_loss: 3.5895 - val_accuracy: 0.0561\n",
+      "Epoch 1887/5000\n",
+      "919/919 - 4s - loss: 1.5259 - accuracy: 0.5062 - val_loss: 3.6003 - val_accuracy: 0.0557\n",
+      "Epoch 1888/5000\n",
+      "919/919 - 4s - loss: 1.5274 - accuracy: 0.5106 - val_loss: 3.5955 - val_accuracy: 0.0560\n",
+      "Epoch 1889/5000\n",
+      "919/919 - 3s - loss: 1.5168 - accuracy: 0.5088 - val_loss: 3.6053 - val_accuracy: 0.0563\n",
+      "Epoch 1890/5000\n",
+      "919/919 - 3s - loss: 1.5198 - accuracy: 0.5131 - val_loss: 3.6029 - val_accuracy: 0.0569\n",
+      "Epoch 1891/5000\n",
+      "919/919 - 4s - loss: 1.5218 - accuracy: 0.5091 - val_loss: 3.6044 - val_accuracy: 0.0566\n",
+      "Epoch 1892/5000\n",
+      "919/919 - 4s - loss: 1.5192 - accuracy: 0.5103 - val_loss: 3.6206 - val_accuracy: 0.0567\n",
+      "Epoch 1893/5000\n",
+      "919/919 - 3s - loss: 1.6097 - accuracy: 0.5109 - val_loss: 3.6242 - val_accuracy: 0.0563\n",
+      "Epoch 1894/5000\n",
+      "919/919 - 3s - loss: 1.5439 - accuracy: 0.5094 - val_loss: 3.6238 - val_accuracy: 0.0563\n",
+      "Epoch 1895/5000\n",
+      "919/919 - 4s - loss: 1.5291 - accuracy: 0.5118 - val_loss: 3.6241 - val_accuracy: 0.0561\n",
+      "Epoch 1896/5000\n",
+      "919/919 - 4s - loss: 1.5379 - accuracy: 0.5050 - val_loss: 3.6268 - val_accuracy: 0.0557\n",
+      "Epoch 1897/5000\n",
+      "919/919 - 3s - loss: 1.5222 - accuracy: 0.5089 - val_loss: 3.6309 - val_accuracy: 0.0561\n",
+      "Epoch 1898/5000\n",
+      "919/919 - 3s - loss: 1.5136 - accuracy: 0.5105 - val_loss: 3.6194 - val_accuracy: 0.0561\n",
+      "Epoch 1899/5000\n",
+      "919/919 - 4s - loss: 1.5016 - accuracy: 0.5141 - val_loss: 3.6213 - val_accuracy: 0.0562\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1900/5000\n",
+      "919/919 - 3s - loss: 1.5353 - accuracy: 0.5090 - val_loss: 3.6192 - val_accuracy: 0.0557\n",
+      "Epoch 1901/5000\n",
+      "919/919 - 3s - loss: 1.6352 - accuracy: 0.5110 - val_loss: 3.6281 - val_accuracy: 0.0604\n",
+      "Epoch 1902/5000\n",
+      "919/919 - 4s - loss: 1.5624 - accuracy: 0.5069 - val_loss: 3.6327 - val_accuracy: 0.0598\n",
+      "Epoch 1903/5000\n",
+      "919/919 - 4s - loss: 1.5097 - accuracy: 0.5141 - val_loss: 3.6278 - val_accuracy: 0.0602\n",
+      "Epoch 1904/5000\n",
+      "919/919 - 3s - loss: 1.5203 - accuracy: 0.5090 - val_loss: 3.6199 - val_accuracy: 0.0604\n",
+      "Epoch 1905/5000\n",
+      "919/919 - 3s - loss: 1.5064 - accuracy: 0.5108 - val_loss: 3.6093 - val_accuracy: 0.0603\n",
+      "Epoch 1906/5000\n",
+      "919/919 - 4s - loss: 1.5233 - accuracy: 0.5088 - val_loss: 3.6125 - val_accuracy: 0.0609\n",
+      "Epoch 1907/5000\n",
+      "919/919 - 4s - loss: 1.5123 - accuracy: 0.5120 - val_loss: 3.5885 - val_accuracy: 0.0601\n",
+      "Epoch 1908/5000\n",
+      "919/919 - 3s - loss: 1.5199 - accuracy: 0.5077 - val_loss: 3.5928 - val_accuracy: 0.0606\n",
+      "Epoch 1909/5000\n",
+      "919/919 - 3s - loss: 1.5159 - accuracy: 0.5069 - val_loss: 3.5909 - val_accuracy: 0.0605\n",
+      "Epoch 1910/5000\n",
+      "919/919 - 4s - loss: 1.5703 - accuracy: 0.5100 - val_loss: 3.5982 - val_accuracy: 0.0609\n",
+      "Epoch 1911/5000\n",
+      "919/919 - 4s - loss: 1.5054 - accuracy: 0.5107 - val_loss: 3.6058 - val_accuracy: 0.0600\n",
+      "Epoch 1912/5000\n",
+      "919/919 - 3s - loss: 1.5423 - accuracy: 0.5140 - val_loss: 3.6025 - val_accuracy: 0.0601\n",
+      "Epoch 1913/5000\n",
+      "919/919 - 3s - loss: 1.5661 - accuracy: 0.5107 - val_loss: 3.6051 - val_accuracy: 0.0605\n",
+      "Epoch 1914/5000\n",
+      "919/919 - 4s - loss: 1.5498 - accuracy: 0.5073 - val_loss: 3.6051 - val_accuracy: 0.0606\n",
+      "Epoch 1915/5000\n",
+      "919/919 - 4s - loss: 1.5534 - accuracy: 0.5124 - val_loss: 3.6112 - val_accuracy: 0.0608\n",
+      "Epoch 1916/5000\n",
+      "919/919 - 3s - loss: 1.5391 - accuracy: 0.5093 - val_loss: 3.6040 - val_accuracy: 0.0605\n",
+      "Epoch 1917/5000\n",
+      "919/919 - 3s - loss: 1.5180 - accuracy: 0.5128 - val_loss: 3.6037 - val_accuracy: 0.0600\n",
+      "Epoch 1918/5000\n",
+      "919/919 - 4s - loss: 1.5018 - accuracy: 0.5142 - val_loss: 3.6171 - val_accuracy: 0.0608\n",
+      "Epoch 1919/5000\n",
+      "919/919 - 4s - loss: 1.5253 - accuracy: 0.5094 - val_loss: 3.6204 - val_accuracy: 0.0611\n",
+      "Epoch 1920/5000\n",
+      "919/919 - 3s - loss: 1.5114 - accuracy: 0.5114 - val_loss: 3.6161 - val_accuracy: 0.0619\n",
+      "Epoch 1921/5000\n",
+      "919/919 - 3s - loss: 1.5080 - accuracy: 0.5129 - val_loss: 3.6090 - val_accuracy: 0.0611\n",
+      "Epoch 1922/5000\n",
+      "919/919 - 4s - loss: 1.5044 - accuracy: 0.5103 - val_loss: 3.6181 - val_accuracy: 0.0615\n",
+      "Epoch 1923/5000\n",
+      "919/919 - 4s - loss: 1.5761 - accuracy: 0.5109 - val_loss: 3.6375 - val_accuracy: 0.0614\n",
+      "Epoch 1924/5000\n",
+      "919/919 - 3s - loss: 1.5820 - accuracy: 0.5085 - val_loss: 3.6208 - val_accuracy: 0.0612\n",
+      "Epoch 1925/5000\n",
+      "919/919 - 3s - loss: 1.5431 - accuracy: 0.5135 - val_loss: 3.6340 - val_accuracy: 0.0614\n",
+      "Epoch 1926/5000\n",
+      "919/919 - 4s - loss: 1.5048 - accuracy: 0.5097 - val_loss: 3.6339 - val_accuracy: 0.0611\n",
+      "Epoch 1927/5000\n",
+      "919/919 - 4s - loss: 1.5006 - accuracy: 0.5136 - val_loss: 3.6291 - val_accuracy: 0.0606\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1928/5000\n",
+      "919/919 - 3s - loss: 1.5380 - accuracy: 0.5119 - val_loss: 3.6401 - val_accuracy: 0.0605\n",
+      "Epoch 1929/5000\n",
+      "919/919 - 3s - loss: 1.5173 - accuracy: 0.5141 - val_loss: 3.6338 - val_accuracy: 0.0612\n",
+      "Epoch 1930/5000\n",
+      "919/919 - 4s - loss: 1.5414 - accuracy: 0.5095 - val_loss: 3.6105 - val_accuracy: 0.0613\n",
+      "Epoch 1931/5000\n",
+      "919/919 - 4s - loss: 1.5133 - accuracy: 0.5096 - val_loss: 3.6198 - val_accuracy: 0.0611\n",
+      "Epoch 1932/5000\n",
+      "919/919 - 3s - loss: 1.5207 - accuracy: 0.5083 - val_loss: 3.6278 - val_accuracy: 0.0609\n",
+      "Epoch 1933/5000\n",
+      "919/919 - 3s - loss: 1.5004 - accuracy: 0.5147 - val_loss: 3.6280 - val_accuracy: 0.0607\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1934/5000\n",
+      "919/919 - 4s - loss: 1.5233 - accuracy: 0.5136 - val_loss: 3.6179 - val_accuracy: 0.0614\n",
+      "Epoch 1935/5000\n",
+      "919/919 - 3s - loss: 1.5103 - accuracy: 0.5124 - val_loss: 3.6178 - val_accuracy: 0.0612\n",
+      "Epoch 1936/5000\n",
+      "919/919 - 3s - loss: 1.5100 - accuracy: 0.5116 - val_loss: 3.6164 - val_accuracy: 0.0613\n",
+      "Epoch 1937/5000\n",
+      "919/919 - 4s - loss: 1.5369 - accuracy: 0.5134 - val_loss: 3.6217 - val_accuracy: 0.0608\n",
+      "Epoch 1938/5000\n",
+      "919/919 - 4s - loss: 1.5528 - accuracy: 0.5128 - val_loss: 3.6210 - val_accuracy: 0.0608\n",
+      "Epoch 1939/5000\n",
+      "919/919 - 3s - loss: 1.5133 - accuracy: 0.5127 - val_loss: 3.6200 - val_accuracy: 0.0616\n",
+      "Epoch 1940/5000\n",
+      "919/919 - 3s - loss: 1.5026 - accuracy: 0.5122 - val_loss: 3.6204 - val_accuracy: 0.0615\n",
+      "Epoch 1941/5000\n",
+      "919/919 - 4s - loss: 1.5203 - accuracy: 0.5141 - val_loss: 3.6109 - val_accuracy: 0.0617\n",
+      "Epoch 1942/5000\n",
+      "919/919 - 4s - loss: 1.5074 - accuracy: 0.5137 - val_loss: 3.6136 - val_accuracy: 0.0614\n",
+      "Epoch 1943/5000\n",
+      "919/919 - 3s - loss: 1.5099 - accuracy: 0.5129 - val_loss: 3.5962 - val_accuracy: 0.0614\n",
+      "Epoch 1944/5000\n",
+      "919/919 - 3s - loss: 1.4977 - accuracy: 0.5127 - val_loss: 3.6186 - val_accuracy: 0.0610\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1945/5000\n",
+      "919/919 - 4s - loss: 1.5120 - accuracy: 0.5135 - val_loss: 3.6318 - val_accuracy: 0.0609\n",
+      "Epoch 1946/5000\n",
+      "919/919 - 4s - loss: 1.5078 - accuracy: 0.5142 - val_loss: 3.6500 - val_accuracy: 0.0607\n",
+      "Epoch 1947/5000\n",
+      "919/919 - 3s - loss: 1.5064 - accuracy: 0.5161 - val_loss: 3.6457 - val_accuracy: 0.0614\n",
+      "Epoch 1948/5000\n",
+      "919/919 - 3s - loss: 1.5286 - accuracy: 0.5156 - val_loss: 3.6282 - val_accuracy: 0.0611\n",
+      "Epoch 1949/5000\n",
+      "919/919 - 4s - loss: 1.5343 - accuracy: 0.5130 - val_loss: 3.6064 - val_accuracy: 0.0612\n",
+      "Epoch 1950/5000\n",
+      "919/919 - 4s - loss: 1.5066 - accuracy: 0.5117 - val_loss: 3.6266 - val_accuracy: 0.0617\n",
+      "Epoch 1951/5000\n",
+      "919/919 - 3s - loss: 1.5108 - accuracy: 0.5101 - val_loss: 3.6139 - val_accuracy: 0.0623\n",
+      "Epoch 1952/5000\n",
+      "919/919 - 3s - loss: 1.5062 - accuracy: 0.5145 - val_loss: 3.6110 - val_accuracy: 0.0621\n",
+      "Epoch 1953/5000\n",
+      "919/919 - 4s - loss: 1.5199 - accuracy: 0.5111 - val_loss: 3.6165 - val_accuracy: 0.0610\n",
+      "Epoch 1954/5000\n",
+      "919/919 - 4s - loss: 1.5478 - accuracy: 0.5074 - val_loss: 3.6098 - val_accuracy: 0.0614\n",
+      "Epoch 1955/5000\n",
+      "919/919 - 3s - loss: 1.4973 - accuracy: 0.5120 - val_loss: 3.6145 - val_accuracy: 0.0626\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1956/5000\n",
+      "919/919 - 3s - loss: 1.4866 - accuracy: 0.5167 - val_loss: 3.6182 - val_accuracy: 0.0620\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1957/5000\n",
+      "919/919 - 4s - loss: 1.5035 - accuracy: 0.5139 - val_loss: 3.6070 - val_accuracy: 0.0622\n",
+      "Epoch 1958/5000\n",
+      "919/919 - 4s - loss: 1.5071 - accuracy: 0.5120 - val_loss: 3.6144 - val_accuracy: 0.0612\n",
+      "Epoch 1959/5000\n",
+      "919/919 - 3s - loss: 1.5337 - accuracy: 0.5135 - val_loss: 3.6188 - val_accuracy: 0.0622\n",
+      "Epoch 1960/5000\n",
+      "919/919 - 3s - loss: 1.5184 - accuracy: 0.5133 - val_loss: 3.6209 - val_accuracy: 0.0618\n",
+      "Epoch 1961/5000\n",
+      "919/919 - 4s - loss: 1.5049 - accuracy: 0.5154 - val_loss: 3.6344 - val_accuracy: 0.0618\n",
+      "Epoch 1962/5000\n",
+      "919/919 - 4s - loss: 1.5377 - accuracy: 0.5157 - val_loss: 3.6341 - val_accuracy: 0.0619\n",
+      "Epoch 1963/5000\n",
+      "919/919 - 3s - loss: 1.5047 - accuracy: 0.5137 - val_loss: 3.6459 - val_accuracy: 0.0610\n",
+      "Epoch 1964/5000\n",
+      "919/919 - 3s - loss: 1.4914 - accuracy: 0.5159 - val_loss: 3.6496 - val_accuracy: 0.0611\n",
+      "Epoch 1965/5000\n",
+      "919/919 - 4s - loss: 1.5044 - accuracy: 0.5143 - val_loss: 3.6461 - val_accuracy: 0.0608\n",
+      "Epoch 1966/5000\n",
+      "919/919 - 3s - loss: 1.4986 - accuracy: 0.5159 - val_loss: 3.6338 - val_accuracy: 0.0621\n",
+      "Epoch 1967/5000\n",
+      "919/919 - 3s - loss: 1.5096 - accuracy: 0.5116 - val_loss: 3.6372 - val_accuracy: 0.0618\n",
+      "Epoch 1968/5000\n",
+      "919/919 - 3s - loss: 1.5120 - accuracy: 0.5165 - val_loss: 3.6344 - val_accuracy: 0.0610\n",
+      "Epoch 1969/5000\n",
+      "919/919 - 4s - loss: 1.4989 - accuracy: 0.5180 - val_loss: 3.6272 - val_accuracy: 0.0614\n",
+      "Epoch 1970/5000\n",
+      "919/919 - 3s - loss: 1.5083 - accuracy: 0.5122 - val_loss: 3.6309 - val_accuracy: 0.0621\n",
+      "Epoch 1971/5000\n",
+      "919/919 - 3s - loss: 1.4968 - accuracy: 0.5144 - val_loss: 3.6220 - val_accuracy: 0.0618\n",
+      "Epoch 1972/5000\n",
+      "919/919 - 3s - loss: 1.5279 - accuracy: 0.5174 - val_loss: 3.6424 - val_accuracy: 0.0617\n",
+      "Epoch 1973/5000\n",
+      "919/919 - 4s - loss: 1.5026 - accuracy: 0.5158 - val_loss: 3.6396 - val_accuracy: 0.0613\n",
+      "Epoch 1974/5000\n",
+      "919/919 - 4s - loss: 1.4938 - accuracy: 0.5165 - val_loss: 3.6382 - val_accuracy: 0.0616\n",
+      "Epoch 1975/5000\n",
+      "919/919 - 3s - loss: 1.4999 - accuracy: 0.5120 - val_loss: 3.6330 - val_accuracy: 0.0613\n",
+      "Epoch 1976/5000\n",
+      "919/919 - 3s - loss: 1.4903 - accuracy: 0.5162 - val_loss: 3.6418 - val_accuracy: 0.0614\n",
+      "Epoch 1977/5000\n",
+      "919/919 - 4s - loss: 1.5423 - accuracy: 0.5173 - val_loss: 3.6323 - val_accuracy: 0.0619\n",
+      "Epoch 1978/5000\n",
+      "919/919 - 3s - loss: 1.5005 - accuracy: 0.5131 - val_loss: 3.6150 - val_accuracy: 0.0617\n",
+      "Epoch 1979/5000\n",
+      "919/919 - 3s - loss: 1.5095 - accuracy: 0.5118 - val_loss: 3.6117 - val_accuracy: 0.0619\n",
+      "Epoch 1980/5000\n",
+      "919/919 - 3s - loss: 1.5076 - accuracy: 0.5090 - val_loss: 3.5888 - val_accuracy: 0.0614\n",
+      "Epoch 1981/5000\n",
+      "919/919 - 4s - loss: 1.5293 - accuracy: 0.5131 - val_loss: 3.6047 - val_accuracy: 0.0619\n",
+      "Epoch 1982/5000\n",
+      "919/919 - 3s - loss: 1.5045 - accuracy: 0.5118 - val_loss: 3.6037 - val_accuracy: 0.0614\n",
+      "Epoch 1983/5000\n",
+      "919/919 - 3s - loss: 1.5010 - accuracy: 0.5143 - val_loss: 3.6079 - val_accuracy: 0.0622\n",
+      "Epoch 1984/5000\n",
+      "919/919 - 4s - loss: 1.5000 - accuracy: 0.5133 - val_loss: 3.6204 - val_accuracy: 0.0623\n",
+      "Epoch 1985/5000\n",
+      "919/919 - 4s - loss: 1.4896 - accuracy: 0.5179 - val_loss: 3.6284 - val_accuracy: 0.0620\n",
+      "Epoch 1986/5000\n",
+      "919/919 - 3s - loss: 1.5054 - accuracy: 0.5141 - val_loss: 3.6264 - val_accuracy: 0.0618\n",
+      "Epoch 1987/5000\n",
+      "919/919 - 3s - loss: 1.5455 - accuracy: 0.5176 - val_loss: 3.6249 - val_accuracy: 0.0623\n",
+      "Epoch 1988/5000\n",
+      "919/919 - 4s - loss: 1.5013 - accuracy: 0.5138 - val_loss: 3.6243 - val_accuracy: 0.0619\n",
+      "Epoch 1989/5000\n",
+      "919/919 - 4s - loss: 1.4854 - accuracy: 0.5199 - val_loss: 3.6229 - val_accuracy: 0.0627\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 1990/5000\n",
+      "919/919 - 3s - loss: 1.5164 - accuracy: 0.5137 - val_loss: 3.6134 - val_accuracy: 0.0624\n",
+      "Epoch 1991/5000\n",
+      "919/919 - 3s - loss: 1.4981 - accuracy: 0.5159 - val_loss: 3.6242 - val_accuracy: 0.0624\n",
+      "Epoch 1992/5000\n",
+      "919/919 - 4s - loss: 1.5293 - accuracy: 0.5164 - val_loss: 3.6214 - val_accuracy: 0.0623\n",
+      "Epoch 1993/5000\n",
+      "919/919 - 4s - loss: 1.5550 - accuracy: 0.5204 - val_loss: 3.6107 - val_accuracy: 0.0621\n",
+      "Epoch 1994/5000\n",
+      "919/919 - 3s - loss: 1.5001 - accuracy: 0.5171 - val_loss: 3.6172 - val_accuracy: 0.0622\n",
+      "Epoch 1995/5000\n",
+      "919/919 - 3s - loss: 1.4943 - accuracy: 0.5184 - val_loss: 3.6165 - val_accuracy: 0.0623\n",
+      "Epoch 1996/5000\n",
+      "919/919 - 4s - loss: 1.5029 - accuracy: 0.5165 - val_loss: 3.6266 - val_accuracy: 0.0623\n",
+      "Epoch 1997/5000\n",
+      "919/919 - 4s - loss: 1.4981 - accuracy: 0.5162 - val_loss: 3.6133 - val_accuracy: 0.0625\n",
+      "Epoch 1998/5000\n",
+      "919/919 - 3s - loss: 1.4860 - accuracy: 0.5178 - val_loss: 3.6155 - val_accuracy: 0.0622\n",
+      "Epoch 1999/5000\n",
+      "919/919 - 3s - loss: 1.4983 - accuracy: 0.5201 - val_loss: 3.6327 - val_accuracy: 0.0623\n",
+      "Epoch 2000/5000\n",
+      "919/919 - 4s - loss: 1.5011 - accuracy: 0.5157 - val_loss: 3.6378 - val_accuracy: 0.0619\n",
+      "Epoch 2001/5000\n",
+      "919/919 - 4s - loss: 1.5056 - accuracy: 0.5137 - val_loss: 3.6211 - val_accuracy: 0.0619\n",
+      "Epoch 2002/5000\n",
+      "919/919 - 3s - loss: 1.5044 - accuracy: 0.5154 - val_loss: 3.6146 - val_accuracy: 0.0622\n",
+      "Epoch 2003/5000\n",
+      "919/919 - 3s - loss: 1.4919 - accuracy: 0.5178 - val_loss: 3.6210 - val_accuracy: 0.0620\n",
+      "Epoch 2004/5000\n",
+      "919/919 - 4s - loss: 1.4962 - accuracy: 0.5157 - val_loss: 3.6153 - val_accuracy: 0.0615\n",
+      "Epoch 2005/5000\n",
+      "919/919 - 4s - loss: 1.5036 - accuracy: 0.5141 - val_loss: 3.6150 - val_accuracy: 0.0620\n",
+      "Epoch 2006/5000\n",
+      "919/919 - 3s - loss: 1.4983 - accuracy: 0.5176 - val_loss: 3.6199 - val_accuracy: 0.0618\n",
+      "Epoch 2007/5000\n",
+      "919/919 - 3s - loss: 1.5080 - accuracy: 0.5164 - val_loss: 3.6247 - val_accuracy: 0.0619\n",
+      "Epoch 2008/5000\n",
+      "919/919 - 4s - loss: 1.4913 - accuracy: 0.5165 - val_loss: 3.6258 - val_accuracy: 0.0618\n",
+      "Epoch 2009/5000\n",
+      "919/919 - 4s - loss: 1.4958 - accuracy: 0.5168 - val_loss: 3.6307 - val_accuracy: 0.0623\n",
+      "Epoch 2010/5000\n",
+      "919/919 - 3s - loss: 1.5271 - accuracy: 0.5122 - val_loss: 3.6309 - val_accuracy: 0.0621\n",
+      "Epoch 2011/5000\n",
+      "919/919 - 3s - loss: 1.4867 - accuracy: 0.5195 - val_loss: 3.6216 - val_accuracy: 0.0629\n",
+      "Epoch 2012/5000\n",
+      "919/919 - 4s - loss: 1.4877 - accuracy: 0.5167 - val_loss: 3.6208 - val_accuracy: 0.0625\n",
+      "Epoch 2013/5000\n",
+      "919/919 - 4s - loss: 1.4881 - accuracy: 0.5187 - val_loss: 3.6307 - val_accuracy: 0.0623\n",
+      "Epoch 2014/5000\n",
+      "919/919 - 3s - loss: 1.4989 - accuracy: 0.5178 - val_loss: 3.6383 - val_accuracy: 0.0630\n",
+      "Epoch 2015/5000\n",
+      "919/919 - 3s - loss: 1.5289 - accuracy: 0.5194 - val_loss: 3.6228 - val_accuracy: 0.0627\n",
+      "Epoch 2016/5000\n",
+      "919/919 - 4s - loss: 1.4810 - accuracy: 0.5222 - val_loss: 3.6231 - val_accuracy: 0.0632\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2017/5000\n",
+      "919/919 - 4s - loss: 1.4883 - accuracy: 0.5199 - val_loss: 3.6360 - val_accuracy: 0.0623\n",
+      "Epoch 2018/5000\n",
+      "919/919 - 3s - loss: 1.5366 - accuracy: 0.5168 - val_loss: 3.6479 - val_accuracy: 0.0632\n",
+      "Epoch 2019/5000\n",
+      "919/919 - 3s - loss: 1.4969 - accuracy: 0.5199 - val_loss: 3.6521 - val_accuracy: 0.0634\n",
+      "Epoch 2020/5000\n",
+      "919/919 - 4s - loss: 1.5024 - accuracy: 0.5180 - val_loss: 3.6437 - val_accuracy: 0.0631\n",
+      "Epoch 2021/5000\n",
+      "919/919 - 4s - loss: 1.4857 - accuracy: 0.5195 - val_loss: 3.6420 - val_accuracy: 0.0632\n",
+      "Epoch 2022/5000\n",
+      "919/919 - 3s - loss: 1.4950 - accuracy: 0.5192 - val_loss: 3.6379 - val_accuracy: 0.0633\n",
+      "Epoch 2023/5000\n",
+      "919/919 - 3s - loss: 1.4987 - accuracy: 0.5184 - val_loss: 3.6495 - val_accuracy: 0.0627\n",
+      "Epoch 2024/5000\n",
+      "919/919 - 4s - loss: 1.4899 - accuracy: 0.5186 - val_loss: 3.6432 - val_accuracy: 0.0627\n",
+      "Epoch 2025/5000\n",
+      "919/919 - 4s - loss: 1.5225 - accuracy: 0.5153 - val_loss: 3.6494 - val_accuracy: 0.0623\n",
+      "Epoch 2026/5000\n",
+      "919/919 - 3s - loss: 1.5012 - accuracy: 0.5224 - val_loss: 3.6327 - val_accuracy: 0.0626\n",
+      "Epoch 2027/5000\n",
+      "919/919 - 3s - loss: 1.4824 - accuracy: 0.5205 - val_loss: 3.6183 - val_accuracy: 0.0627\n",
+      "Epoch 2028/5000\n",
+      "919/919 - 4s - loss: 1.5123 - accuracy: 0.5163 - val_loss: 3.6234 - val_accuracy: 0.0620\n",
+      "Epoch 2029/5000\n",
+      "919/919 - 3s - loss: 1.4756 - accuracy: 0.5218 - val_loss: 3.6218 - val_accuracy: 0.0623\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2030/5000\n",
+      "919/919 - 3s - loss: 1.5037 - accuracy: 0.5163 - val_loss: 3.6242 - val_accuracy: 0.0623\n",
+      "Epoch 2031/5000\n",
+      "919/919 - 4s - loss: 1.4769 - accuracy: 0.5236 - val_loss: 3.6182 - val_accuracy: 0.0623\n",
+      "Epoch 2032/5000\n",
+      "919/919 - 4s - loss: 1.5213 - accuracy: 0.5224 - val_loss: 3.6229 - val_accuracy: 0.0620\n",
+      "Epoch 2033/5000\n",
+      "919/919 - 3s - loss: 1.4656 - accuracy: 0.5207 - val_loss: 3.6347 - val_accuracy: 0.0619\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2034/5000\n",
+      "919/919 - 3s - loss: 1.4952 - accuracy: 0.5195 - val_loss: 3.6160 - val_accuracy: 0.0623\n",
+      "Epoch 2035/5000\n",
+      "919/919 - 4s - loss: 1.4860 - accuracy: 0.5229 - val_loss: 3.6322 - val_accuracy: 0.0632\n",
+      "Epoch 2036/5000\n",
+      "919/919 - 4s - loss: 1.4880 - accuracy: 0.5182 - val_loss: 3.6209 - val_accuracy: 0.0631\n",
+      "Epoch 2037/5000\n",
+      "919/919 - 3s - loss: 1.4879 - accuracy: 0.5242 - val_loss: 3.6313 - val_accuracy: 0.0631\n",
+      "Epoch 2038/5000\n",
+      "919/919 - 3s - loss: 1.4810 - accuracy: 0.5203 - val_loss: 3.6279 - val_accuracy: 0.0627\n",
+      "Epoch 2039/5000\n",
+      "919/919 - 4s - loss: 1.4986 - accuracy: 0.5190 - val_loss: 3.6221 - val_accuracy: 0.0629\n",
+      "Epoch 2040/5000\n",
+      "919/919 - 4s - loss: 1.4918 - accuracy: 0.5195 - val_loss: 3.6365 - val_accuracy: 0.0628\n",
+      "Epoch 2041/5000\n",
+      "919/919 - 3s - loss: 1.4856 - accuracy: 0.5227 - val_loss: 3.6351 - val_accuracy: 0.0624\n",
+      "Epoch 2042/5000\n",
+      "919/919 - 3s - loss: 1.5295 - accuracy: 0.5192 - val_loss: 3.6341 - val_accuracy: 0.0628\n",
+      "Epoch 2043/5000\n",
+      "919/919 - 4s - loss: 1.4886 - accuracy: 0.5203 - val_loss: 3.6560 - val_accuracy: 0.0623\n",
+      "Epoch 2044/5000\n",
+      "919/919 - 4s - loss: 1.4943 - accuracy: 0.5220 - val_loss: 3.6431 - val_accuracy: 0.0623\n",
+      "Epoch 2045/5000\n",
+      "919/919 - 3s - loss: 1.5750 - accuracy: 0.5209 - val_loss: 3.6313 - val_accuracy: 0.0623\n",
+      "Epoch 2046/5000\n",
+      "919/919 - 3s - loss: 1.4917 - accuracy: 0.5214 - val_loss: 3.6310 - val_accuracy: 0.0624\n",
+      "Epoch 2047/5000\n",
+      "919/919 - 4s - loss: 1.4832 - accuracy: 0.5231 - val_loss: 3.6325 - val_accuracy: 0.0628\n",
+      "Epoch 2048/5000\n",
+      "919/919 - 4s - loss: 1.5162 - accuracy: 0.5210 - val_loss: 3.6229 - val_accuracy: 0.0632\n",
+      "Epoch 2049/5000\n",
+      "919/919 - 3s - loss: 1.4690 - accuracy: 0.5247 - val_loss: 3.6284 - val_accuracy: 0.0633\n",
+      "Epoch 2050/5000\n",
+      "919/919 - 3s - loss: 1.4713 - accuracy: 0.5259 - val_loss: 3.6480 - val_accuracy: 0.0631\n",
+      "Epoch 2051/5000\n",
+      "919/919 - 4s - loss: 1.4906 - accuracy: 0.5205 - val_loss: 3.6328 - val_accuracy: 0.0636\n",
+      "Epoch 2052/5000\n",
+      "919/919 - 4s - loss: 1.4903 - accuracy: 0.5183 - val_loss: 3.6271 - val_accuracy: 0.0634\n",
+      "Epoch 2053/5000\n",
+      "919/919 - 3s - loss: 1.4751 - accuracy: 0.5223 - val_loss: 3.6446 - val_accuracy: 0.0641\n",
+      "Epoch 2054/5000\n",
+      "919/919 - 5s - loss: 1.4797 - accuracy: 0.5203 - val_loss: 3.6295 - val_accuracy: 0.0641\n",
+      "Epoch 2055/5000\n",
+      "919/919 - 3s - loss: 1.4890 - accuracy: 0.5206 - val_loss: 3.6312 - val_accuracy: 0.0640\n",
+      "Epoch 2056/5000\n",
+      "919/919 - 3s - loss: 1.4933 - accuracy: 0.5194 - val_loss: 3.6346 - val_accuracy: 0.0640\n",
+      "Epoch 2057/5000\n",
+      "919/919 - 3s - loss: 1.4982 - accuracy: 0.5187 - val_loss: 3.6253 - val_accuracy: 0.0633\n",
+      "Epoch 2058/5000\n",
+      "919/919 - 3s - loss: 1.4733 - accuracy: 0.5235 - val_loss: 3.6358 - val_accuracy: 0.0636\n",
+      "Epoch 2059/5000\n",
+      "919/919 - 3s - loss: 1.4932 - accuracy: 0.5237 - val_loss: 3.6333 - val_accuracy: 0.0637\n",
+      "Epoch 2060/5000\n",
+      "919/919 - 3s - loss: 1.4957 - accuracy: 0.5235 - val_loss: 3.6238 - val_accuracy: 0.0630\n",
+      "Epoch 2061/5000\n",
+      "919/919 - 3s - loss: 1.4946 - accuracy: 0.5210 - val_loss: 3.6339 - val_accuracy: 0.0637\n",
+      "Epoch 2062/5000\n",
+      "919/919 - 3s - loss: 1.4970 - accuracy: 0.5184 - val_loss: 3.6268 - val_accuracy: 0.0631\n",
+      "Epoch 2063/5000\n",
+      "919/919 - 3s - loss: 1.4808 - accuracy: 0.5251 - val_loss: 3.6336 - val_accuracy: 0.0629\n",
+      "Epoch 2064/5000\n",
+      "919/919 - 3s - loss: 1.4750 - accuracy: 0.5246 - val_loss: 3.6290 - val_accuracy: 0.0635\n",
+      "Epoch 2065/5000\n",
+      "919/919 - 3s - loss: 1.4863 - accuracy: 0.5216 - val_loss: 3.6226 - val_accuracy: 0.0640\n",
+      "Epoch 2066/5000\n",
+      "919/919 - 3s - loss: 1.4893 - accuracy: 0.5185 - val_loss: 3.6152 - val_accuracy: 0.0637\n",
+      "Epoch 2067/5000\n",
+      "919/919 - 3s - loss: 1.4737 - accuracy: 0.5252 - val_loss: 3.6216 - val_accuracy: 0.0633\n",
+      "Epoch 2068/5000\n",
+      "919/919 - 3s - loss: 1.5211 - accuracy: 0.5240 - val_loss: 3.6188 - val_accuracy: 0.0626\n",
+      "Epoch 2069/5000\n",
+      "919/919 - 3s - loss: 1.5052 - accuracy: 0.5216 - val_loss: 3.6325 - val_accuracy: 0.0625\n",
+      "Epoch 2070/5000\n",
+      "919/919 - 3s - loss: 1.4883 - accuracy: 0.5226 - val_loss: 3.6281 - val_accuracy: 0.0631\n",
+      "Epoch 2071/5000\n",
+      "919/919 - 3s - loss: 1.4840 - accuracy: 0.5246 - val_loss: 3.6316 - val_accuracy: 0.0632\n",
+      "Epoch 2072/5000\n",
+      "919/919 - 3s - loss: 1.4913 - accuracy: 0.5221 - val_loss: 3.6202 - val_accuracy: 0.0630\n",
+      "Epoch 2073/5000\n",
+      "919/919 - 3s - loss: 1.5393 - accuracy: 0.5213 - val_loss: 3.6288 - val_accuracy: 0.0639\n",
+      "Epoch 2074/5000\n",
+      "919/919 - 3s - loss: 1.4910 - accuracy: 0.5191 - val_loss: 3.6257 - val_accuracy: 0.0635\n",
+      "Epoch 2075/5000\n",
+      "919/919 - 3s - loss: 1.5789 - accuracy: 0.5265 - val_loss: 3.6255 - val_accuracy: 0.0632\n",
+      "Epoch 2076/5000\n",
+      "919/919 - 3s - loss: 1.5577 - accuracy: 0.5254 - val_loss: 3.6327 - val_accuracy: 0.0629\n",
+      "Epoch 2077/5000\n",
+      "919/919 - 3s - loss: 1.5252 - accuracy: 0.5257 - val_loss: 3.6314 - val_accuracy: 0.0629\n",
+      "Epoch 2078/5000\n",
+      "919/919 - 3s - loss: 1.5364 - accuracy: 0.5230 - val_loss: 3.6322 - val_accuracy: 0.0636\n",
+      "Epoch 2079/5000\n",
+      "919/919 - 3s - loss: 1.4925 - accuracy: 0.5199 - val_loss: 3.6434 - val_accuracy: 0.0640\n",
+      "Epoch 2080/5000\n",
+      "919/919 - 3s - loss: 1.4946 - accuracy: 0.5270 - val_loss: 3.6656 - val_accuracy: 0.0635\n",
+      "Epoch 2081/5000\n",
+      "919/919 - 3s - loss: 1.4915 - accuracy: 0.5217 - val_loss: 3.6510 - val_accuracy: 0.0632\n",
+      "Epoch 2082/5000\n",
+      "919/919 - 3s - loss: 1.4764 - accuracy: 0.5250 - val_loss: 3.6585 - val_accuracy: 0.0628\n",
+      "Epoch 2083/5000\n",
+      "919/919 - 3s - loss: 1.4645 - accuracy: 0.5274 - val_loss: 3.6493 - val_accuracy: 0.0631\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2084/5000\n",
+      "919/919 - 3s - loss: 1.5047 - accuracy: 0.5220 - val_loss: 3.6382 - val_accuracy: 0.0635\n",
+      "Epoch 2085/5000\n",
+      "919/919 - 3s - loss: 1.4700 - accuracy: 0.5267 - val_loss: 3.6512 - val_accuracy: 0.0632\n",
+      "Epoch 2086/5000\n",
+      "919/919 - 3s - loss: 1.4863 - accuracy: 0.5239 - val_loss: 3.6549 - val_accuracy: 0.0637\n",
+      "Epoch 2087/5000\n",
+      "919/919 - 3s - loss: 1.4830 - accuracy: 0.5263 - val_loss: 3.6580 - val_accuracy: 0.0635\n",
+      "Epoch 2088/5000\n",
+      "919/919 - 3s - loss: 1.4844 - accuracy: 0.5252 - val_loss: 3.6509 - val_accuracy: 0.0632\n",
+      "Epoch 2089/5000\n",
+      "919/919 - 3s - loss: 1.4722 - accuracy: 0.5222 - val_loss: 3.6564 - val_accuracy: 0.0634\n",
+      "Epoch 2090/5000\n",
+      "919/919 - 3s - loss: 1.5292 - accuracy: 0.5267 - val_loss: 3.6389 - val_accuracy: 0.0636\n",
+      "Epoch 2091/5000\n",
+      "919/919 - 3s - loss: 1.5226 - accuracy: 0.5197 - val_loss: 3.6242 - val_accuracy: 0.0635\n",
+      "Epoch 2092/5000\n",
+      "919/919 - 3s - loss: 1.4813 - accuracy: 0.5261 - val_loss: 3.6173 - val_accuracy: 0.0634\n",
+      "Epoch 2093/5000\n",
+      "919/919 - 3s - loss: 1.4777 - accuracy: 0.5253 - val_loss: 3.6091 - val_accuracy: 0.0632\n",
+      "Epoch 2094/5000\n",
+      "919/919 - 3s - loss: 1.4876 - accuracy: 0.5222 - val_loss: 3.6126 - val_accuracy: 0.0634\n",
+      "Epoch 2095/5000\n",
+      "919/919 - 3s - loss: 1.4691 - accuracy: 0.5283 - val_loss: 3.6213 - val_accuracy: 0.0638\n",
+      "Epoch 2096/5000\n",
+      "919/919 - 3s - loss: 1.4793 - accuracy: 0.5275 - val_loss: 3.6279 - val_accuracy: 0.0636\n",
+      "Epoch 2097/5000\n",
+      "919/919 - 3s - loss: 1.4772 - accuracy: 0.5259 - val_loss: 3.6218 - val_accuracy: 0.0639\n",
+      "Epoch 2098/5000\n",
+      "919/919 - 3s - loss: 1.4851 - accuracy: 0.5282 - val_loss: 3.6295 - val_accuracy: 0.0644\n",
+      "Epoch 2099/5000\n",
+      "919/919 - 3s - loss: 1.4665 - accuracy: 0.5284 - val_loss: 3.6395 - val_accuracy: 0.0643\n",
+      "Epoch 2100/5000\n",
+      "919/919 - 3s - loss: 1.4795 - accuracy: 0.5256 - val_loss: 3.6409 - val_accuracy: 0.0646\n",
+      "Epoch 2101/5000\n",
+      "919/919 - 3s - loss: 1.5789 - accuracy: 0.5246 - val_loss: 3.6290 - val_accuracy: 0.0645\n",
+      "Epoch 2102/5000\n",
+      "919/919 - 3s - loss: 1.4483 - accuracy: 0.5297 - val_loss: 3.6324 - val_accuracy: 0.0644\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2103/5000\n",
+      "919/919 - 3s - loss: 1.4849 - accuracy: 0.5263 - val_loss: 3.6302 - val_accuracy: 0.0642\n",
+      "Epoch 2104/5000\n",
+      "919/919 - 3s - loss: 1.5287 - accuracy: 0.5246 - val_loss: 3.6269 - val_accuracy: 0.0638\n",
+      "Epoch 2105/5000\n",
+      "919/919 - 3s - loss: 1.4703 - accuracy: 0.5261 - val_loss: 3.6280 - val_accuracy: 0.0640\n",
+      "Epoch 2106/5000\n",
+      "919/919 - 3s - loss: 1.4961 - accuracy: 0.5276 - val_loss: 3.6350 - val_accuracy: 0.0639\n",
+      "Epoch 2107/5000\n",
+      "919/919 - 3s - loss: 1.4547 - accuracy: 0.5267 - val_loss: 3.6453 - val_accuracy: 0.0648\n",
+      "Epoch 2108/5000\n",
+      "919/919 - 3s - loss: 1.4755 - accuracy: 0.5235 - val_loss: 3.6284 - val_accuracy: 0.0643\n",
+      "Epoch 2109/5000\n",
+      "919/919 - 3s - loss: 1.4857 - accuracy: 0.5270 - val_loss: 3.6198 - val_accuracy: 0.0648\n",
+      "Epoch 2110/5000\n",
+      "919/919 - 3s - loss: 1.4707 - accuracy: 0.5287 - val_loss: 3.6200 - val_accuracy: 0.0641\n",
+      "Epoch 2111/5000\n",
+      "919/919 - 3s - loss: 1.4821 - accuracy: 0.5270 - val_loss: 3.6211 - val_accuracy: 0.0641\n",
+      "Epoch 2112/5000\n",
+      "919/919 - 3s - loss: 1.4530 - accuracy: 0.5301 - val_loss: 3.6224 - val_accuracy: 0.0637\n",
+      "Epoch 2113/5000\n",
+      "919/919 - 3s - loss: 1.4685 - accuracy: 0.5272 - val_loss: 3.6308 - val_accuracy: 0.0636\n",
+      "Epoch 2114/5000\n",
+      "919/919 - 3s - loss: 1.4986 - accuracy: 0.5246 - val_loss: 3.6301 - val_accuracy: 0.0637\n",
+      "Epoch 2115/5000\n",
+      "919/919 - 3s - loss: 1.4653 - accuracy: 0.5287 - val_loss: 3.6360 - val_accuracy: 0.0634\n",
+      "Epoch 2116/5000\n",
+      "919/919 - 3s - loss: 1.4726 - accuracy: 0.5284 - val_loss: 3.6379 - val_accuracy: 0.0641\n",
+      "Epoch 2117/5000\n",
+      "919/919 - 3s - loss: 1.4867 - accuracy: 0.5286 - val_loss: 3.6356 - val_accuracy: 0.0639\n",
+      "Epoch 2118/5000\n",
+      "919/919 - 3s - loss: 1.4660 - accuracy: 0.5288 - val_loss: 3.6391 - val_accuracy: 0.0641\n",
+      "Epoch 2119/5000\n",
+      "919/919 - 3s - loss: 1.4725 - accuracy: 0.5323 - val_loss: 3.6393 - val_accuracy: 0.0648\n",
+      "Epoch 2120/5000\n",
+      "919/919 - 3s - loss: 1.4729 - accuracy: 0.5269 - val_loss: 3.6508 - val_accuracy: 0.0650\n",
+      "Epoch 2121/5000\n",
+      "919/919 - 3s - loss: 1.4791 - accuracy: 0.5295 - val_loss: 3.6543 - val_accuracy: 0.0648\n",
+      "Epoch 2122/5000\n",
+      "919/919 - 3s - loss: 1.4559 - accuracy: 0.5294 - val_loss: 3.6517 - val_accuracy: 0.0646\n",
+      "Epoch 2123/5000\n",
+      "919/919 - 3s - loss: 1.4721 - accuracy: 0.5286 - val_loss: 3.6445 - val_accuracy: 0.0650\n",
+      "Epoch 2124/5000\n",
+      "919/919 - 3s - loss: 1.4710 - accuracy: 0.5294 - val_loss: 3.6317 - val_accuracy: 0.0648\n",
+      "Epoch 2125/5000\n",
+      "919/919 - 3s - loss: 1.5524 - accuracy: 0.5263 - val_loss: 3.6317 - val_accuracy: 0.0650\n",
+      "Epoch 2126/5000\n",
+      "919/919 - 3s - loss: 1.4757 - accuracy: 0.5267 - val_loss: 3.6411 - val_accuracy: 0.0649\n",
+      "Epoch 2127/5000\n",
+      "919/919 - 3s - loss: 1.4839 - accuracy: 0.5267 - val_loss: 3.6508 - val_accuracy: 0.0647\n",
+      "Epoch 2128/5000\n",
+      "919/919 - 3s - loss: 1.4615 - accuracy: 0.5280 - val_loss: 3.6584 - val_accuracy: 0.0647\n",
+      "Epoch 2129/5000\n",
+      "919/919 - 3s - loss: 1.7335 - accuracy: 0.5252 - val_loss: 3.6625 - val_accuracy: 0.0650\n",
+      "Epoch 2130/5000\n",
+      "919/919 - 3s - loss: 1.4620 - accuracy: 0.5302 - val_loss: 3.6657 - val_accuracy: 0.0644\n",
+      "Epoch 2131/5000\n",
+      "919/919 - 3s - loss: 1.4748 - accuracy: 0.5274 - val_loss: 3.6563 - val_accuracy: 0.0650\n",
+      "Epoch 2132/5000\n",
+      "919/919 - 3s - loss: 1.5115 - accuracy: 0.5284 - val_loss: 3.6539 - val_accuracy: 0.0643\n",
+      "Epoch 2133/5000\n",
+      "919/919 - 3s - loss: 1.4804 - accuracy: 0.5276 - val_loss: 3.6453 - val_accuracy: 0.0650\n",
+      "Epoch 2134/5000\n",
+      "919/919 - 3s - loss: 1.4588 - accuracy: 0.5273 - val_loss: 3.6484 - val_accuracy: 0.0641\n",
+      "Epoch 2135/5000\n",
+      "919/919 - 3s - loss: 1.4877 - accuracy: 0.5252 - val_loss: 3.6440 - val_accuracy: 0.0642\n",
+      "Epoch 2136/5000\n",
+      "919/919 - 3s - loss: 1.4619 - accuracy: 0.5295 - val_loss: 3.6489 - val_accuracy: 0.0648\n",
+      "Epoch 2137/5000\n",
+      "919/919 - 3s - loss: 1.4658 - accuracy: 0.5282 - val_loss: 3.6532 - val_accuracy: 0.0652\n",
+      "Epoch 2138/5000\n",
+      "919/919 - 3s - loss: 1.4785 - accuracy: 0.5287 - val_loss: 3.6638 - val_accuracy: 0.0648\n",
+      "Epoch 2139/5000\n",
+      "919/919 - 3s - loss: 1.4585 - accuracy: 0.5320 - val_loss: 3.6637 - val_accuracy: 0.0652\n",
+      "Epoch 2140/5000\n",
+      "919/919 - 3s - loss: 1.4760 - accuracy: 0.5327 - val_loss: 3.6673 - val_accuracy: 0.0652\n",
+      "Epoch 2141/5000\n",
+      "919/919 - 3s - loss: 1.4541 - accuracy: 0.5280 - val_loss: 3.6677 - val_accuracy: 0.0650\n",
+      "Epoch 2142/5000\n",
+      "919/919 - 3s - loss: 1.4699 - accuracy: 0.5263 - val_loss: 3.6604 - val_accuracy: 0.0654\n",
+      "Epoch 2143/5000\n",
+      "919/919 - 3s - loss: 1.4566 - accuracy: 0.5312 - val_loss: 3.6569 - val_accuracy: 0.0647\n",
+      "Epoch 2144/5000\n",
+      "919/919 - 3s - loss: 1.4452 - accuracy: 0.5325 - val_loss: 3.6751 - val_accuracy: 0.0641\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2145/5000\n",
+      "919/919 - 3s - loss: 1.4706 - accuracy: 0.5244 - val_loss: 3.6548 - val_accuracy: 0.0639\n",
+      "Epoch 2146/5000\n",
+      "919/919 - 3s - loss: 1.4675 - accuracy: 0.5255 - val_loss: 3.6400 - val_accuracy: 0.0645\n",
+      "Epoch 2147/5000\n",
+      "919/919 - 3s - loss: 1.4669 - accuracy: 0.5297 - val_loss: 3.6431 - val_accuracy: 0.0645\n",
+      "Epoch 2148/5000\n",
+      "919/919 - 4s - loss: 1.4734 - accuracy: 0.5268 - val_loss: 3.6453 - val_accuracy: 0.0643\n",
+      "Epoch 2149/5000\n",
+      "919/919 - 3s - loss: 1.5082 - accuracy: 0.5281 - val_loss: 3.6459 - val_accuracy: 0.0643\n",
+      "Epoch 2150/5000\n",
+      "919/919 - 3s - loss: 1.4688 - accuracy: 0.5280 - val_loss: 3.6479 - val_accuracy: 0.0646\n",
+      "Epoch 2151/5000\n",
+      "919/919 - 3s - loss: 1.4742 - accuracy: 0.5295 - val_loss: 3.6474 - val_accuracy: 0.0647\n",
+      "Epoch 2152/5000\n",
+      "919/919 - 3s - loss: 1.4528 - accuracy: 0.5357 - val_loss: 3.6532 - val_accuracy: 0.0648\n",
+      "Epoch 2153/5000\n",
+      "919/919 - 3s - loss: 1.4693 - accuracy: 0.5296 - val_loss: 3.6460 - val_accuracy: 0.0649\n",
+      "Epoch 2154/5000\n",
+      "919/919 - 3s - loss: 1.4617 - accuracy: 0.5264 - val_loss: 3.6337 - val_accuracy: 0.0654\n",
+      "Epoch 2155/5000\n",
+      "919/919 - 3s - loss: 1.4780 - accuracy: 0.5266 - val_loss: 3.6352 - val_accuracy: 0.0651\n",
+      "Epoch 2156/5000\n",
+      "919/919 - 3s - loss: 1.4990 - accuracy: 0.5337 - val_loss: 3.6427 - val_accuracy: 0.0657\n",
+      "Epoch 2157/5000\n",
+      "919/919 - 3s - loss: 1.4687 - accuracy: 0.5262 - val_loss: 3.6398 - val_accuracy: 0.0650\n",
+      "Epoch 2158/5000\n",
+      "919/919 - 3s - loss: 1.4689 - accuracy: 0.5305 - val_loss: 3.6481 - val_accuracy: 0.0647\n",
+      "Epoch 2159/5000\n",
+      "919/919 - 3s - loss: 1.5640 - accuracy: 0.5291 - val_loss: 3.6529 - val_accuracy: 0.0648\n",
+      "Epoch 2160/5000\n",
+      "919/919 - 3s - loss: 1.6199 - accuracy: 0.5290 - val_loss: 3.6676 - val_accuracy: 0.0649\n",
+      "Epoch 2161/5000\n",
+      "919/919 - 3s - loss: 1.4641 - accuracy: 0.5287 - val_loss: 3.6623 - val_accuracy: 0.0654\n",
+      "Epoch 2162/5000\n",
+      "919/919 - 3s - loss: 1.4585 - accuracy: 0.5297 - val_loss: 3.6600 - val_accuracy: 0.0652\n",
+      "Epoch 2163/5000\n",
+      "919/919 - 3s - loss: 1.5433 - accuracy: 0.5278 - val_loss: 3.6461 - val_accuracy: 0.0645\n",
+      "Epoch 2164/5000\n",
+      "919/919 - 3s - loss: 1.4713 - accuracy: 0.5263 - val_loss: 3.6518 - val_accuracy: 0.0647\n",
+      "Epoch 2165/5000\n",
+      "919/919 - 3s - loss: 1.4718 - accuracy: 0.5327 - val_loss: 3.6504 - val_accuracy: 0.0641\n",
+      "Epoch 2166/5000\n",
+      "919/919 - 3s - loss: 1.5080 - accuracy: 0.5316 - val_loss: 3.6478 - val_accuracy: 0.0648\n",
+      "Epoch 2167/5000\n",
+      "919/919 - 3s - loss: 1.4929 - accuracy: 0.5307 - val_loss: 3.6574 - val_accuracy: 0.0644\n",
+      "Epoch 2168/5000\n",
+      "919/919 - 3s - loss: 1.4687 - accuracy: 0.5323 - val_loss: 3.6496 - val_accuracy: 0.0647\n",
+      "Epoch 2169/5000\n",
+      "919/919 - 3s - loss: 1.4602 - accuracy: 0.5304 - val_loss: 3.6442 - val_accuracy: 0.0645\n",
+      "Epoch 2170/5000\n",
+      "919/919 - 3s - loss: 1.4558 - accuracy: 0.5305 - val_loss: 3.6362 - val_accuracy: 0.0654\n",
+      "Epoch 2171/5000\n",
+      "919/919 - 3s - loss: 1.4438 - accuracy: 0.5329 - val_loss: 3.6448 - val_accuracy: 0.0647\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2172/5000\n",
+      "919/919 - 3s - loss: 1.4538 - accuracy: 0.5322 - val_loss: 3.6508 - val_accuracy: 0.0641\n",
+      "Epoch 2173/5000\n",
+      "919/919 - 3s - loss: 1.4616 - accuracy: 0.5315 - val_loss: 3.6501 - val_accuracy: 0.0640\n",
+      "Epoch 2174/5000\n",
+      "919/919 - 3s - loss: 1.4747 - accuracy: 0.5262 - val_loss: 3.6563 - val_accuracy: 0.0635\n",
+      "Epoch 2175/5000\n",
+      "919/919 - 3s - loss: 1.4605 - accuracy: 0.5345 - val_loss: 3.6517 - val_accuracy: 0.0646\n",
+      "Epoch 2176/5000\n",
+      "919/919 - 3s - loss: 1.4535 - accuracy: 0.5345 - val_loss: 3.6418 - val_accuracy: 0.0650\n",
+      "Epoch 2177/5000\n",
+      "919/919 - 3s - loss: 1.4663 - accuracy: 0.5316 - val_loss: 3.6475 - val_accuracy: 0.0651\n",
+      "Epoch 2178/5000\n",
+      "919/919 - 3s - loss: 1.4543 - accuracy: 0.5361 - val_loss: 3.6521 - val_accuracy: 0.0654\n",
+      "Epoch 2179/5000\n",
+      "919/919 - 3s - loss: 1.4711 - accuracy: 0.5316 - val_loss: 3.6664 - val_accuracy: 0.0651\n",
+      "Epoch 2180/5000\n",
+      "919/919 - 3s - loss: 1.4663 - accuracy: 0.5314 - val_loss: 3.6515 - val_accuracy: 0.0651\n",
+      "Epoch 2181/5000\n",
+      "919/919 - 3s - loss: 1.4691 - accuracy: 0.5367 - val_loss: 3.6541 - val_accuracy: 0.0650\n",
+      "Epoch 2182/5000\n",
+      "919/919 - 3s - loss: 1.4557 - accuracy: 0.5298 - val_loss: 3.6638 - val_accuracy: 0.0650\n",
+      "Epoch 2183/5000\n",
+      "919/919 - 3s - loss: 1.4735 - accuracy: 0.5264 - val_loss: 3.6609 - val_accuracy: 0.0652\n",
+      "Epoch 2184/5000\n",
+      "919/919 - 3s - loss: 1.4562 - accuracy: 0.5335 - val_loss: 3.6602 - val_accuracy: 0.0655\n",
+      "Epoch 2185/5000\n",
+      "919/919 - 3s - loss: 1.5251 - accuracy: 0.5301 - val_loss: 3.6438 - val_accuracy: 0.0651\n",
+      "Epoch 2186/5000\n",
+      "919/919 - 3s - loss: 1.4564 - accuracy: 0.5316 - val_loss: 3.6506 - val_accuracy: 0.0649\n",
+      "Epoch 2187/5000\n",
+      "919/919 - 3s - loss: 1.4545 - accuracy: 0.5293 - val_loss: 3.6412 - val_accuracy: 0.0654\n",
+      "Epoch 2188/5000\n",
+      "919/919 - 3s - loss: 1.4631 - accuracy: 0.5307 - val_loss: 3.6381 - val_accuracy: 0.0654\n",
+      "Epoch 2189/5000\n",
+      "919/919 - 3s - loss: 1.4511 - accuracy: 0.5315 - val_loss: 3.6366 - val_accuracy: 0.0657\n",
+      "Epoch 2190/5000\n",
+      "919/919 - 3s - loss: 1.4511 - accuracy: 0.5369 - val_loss: 3.6507 - val_accuracy: 0.0655\n",
+      "Epoch 2191/5000\n",
+      "919/919 - 3s - loss: 1.4637 - accuracy: 0.5303 - val_loss: 3.6496 - val_accuracy: 0.0656\n",
+      "Epoch 2192/5000\n",
+      "919/919 - 3s - loss: 1.4323 - accuracy: 0.5395 - val_loss: 3.6620 - val_accuracy: 0.0651\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2193/5000\n",
+      "919/919 - 3s - loss: 1.4448 - accuracy: 0.5331 - val_loss: 3.6634 - val_accuracy: 0.0655\n",
+      "Epoch 2194/5000\n",
+      "919/919 - 3s - loss: 1.4514 - accuracy: 0.5329 - val_loss: 3.6543 - val_accuracy: 0.0656\n",
+      "Epoch 2195/5000\n",
+      "919/919 - 3s - loss: 1.4655 - accuracy: 0.5316 - val_loss: 3.6596 - val_accuracy: 0.0656\n",
+      "Epoch 2196/5000\n",
+      "919/919 - 3s - loss: 1.4535 - accuracy: 0.5327 - val_loss: 3.6505 - val_accuracy: 0.0651\n",
+      "Epoch 2197/5000\n",
+      "919/919 - 3s - loss: 1.4605 - accuracy: 0.5325 - val_loss: 3.6325 - val_accuracy: 0.0653\n",
+      "Epoch 2198/5000\n",
+      "919/919 - 3s - loss: 1.4582 - accuracy: 0.5310 - val_loss: 3.6385 - val_accuracy: 0.0653\n",
+      "Epoch 2199/5000\n",
+      "919/919 - 3s - loss: 1.4509 - accuracy: 0.5382 - val_loss: 3.6517 - val_accuracy: 0.0652\n",
+      "Epoch 2200/5000\n",
+      "919/919 - 3s - loss: 1.4609 - accuracy: 0.5348 - val_loss: 3.6529 - val_accuracy: 0.0656\n",
+      "Epoch 2201/5000\n",
+      "919/919 - 3s - loss: 1.4705 - accuracy: 0.5346 - val_loss: 3.6511 - val_accuracy: 0.0648\n",
+      "Epoch 2202/5000\n",
+      "919/919 - 3s - loss: 1.5793 - accuracy: 0.5312 - val_loss: 3.6393 - val_accuracy: 0.0645\n",
+      "Epoch 2203/5000\n",
+      "919/919 - 3s - loss: 1.4546 - accuracy: 0.5321 - val_loss: 3.6395 - val_accuracy: 0.0651\n",
+      "Epoch 2204/5000\n",
+      "919/919 - 3s - loss: 1.4615 - accuracy: 0.5309 - val_loss: 3.6521 - val_accuracy: 0.0653\n",
+      "Epoch 2205/5000\n",
+      "919/919 - 3s - loss: 1.4542 - accuracy: 0.5348 - val_loss: 3.6451 - val_accuracy: 0.0651\n",
+      "Epoch 2206/5000\n",
+      "919/919 - 3s - loss: 1.4681 - accuracy: 0.5357 - val_loss: 3.6501 - val_accuracy: 0.0656\n",
+      "Epoch 2207/5000\n",
+      "919/919 - 3s - loss: 1.4554 - accuracy: 0.5326 - val_loss: 3.6533 - val_accuracy: 0.0659\n",
+      "Epoch 2208/5000\n",
+      "919/919 - 3s - loss: 1.4366 - accuracy: 0.5371 - val_loss: 3.6719 - val_accuracy: 0.0659\n",
+      "Epoch 2209/5000\n",
+      "919/919 - 3s - loss: 1.4517 - accuracy: 0.5341 - val_loss: 3.6718 - val_accuracy: 0.0659\n",
+      "Epoch 2210/5000\n",
+      "919/919 - 3s - loss: 1.5062 - accuracy: 0.5350 - val_loss: 3.6550 - val_accuracy: 0.0661\n",
+      "Epoch 2211/5000\n",
+      "919/919 - 3s - loss: 1.4901 - accuracy: 0.5362 - val_loss: 3.6674 - val_accuracy: 0.0659\n",
+      "Epoch 2212/5000\n",
+      "919/919 - 3s - loss: 1.4931 - accuracy: 0.5341 - val_loss: 3.6513 - val_accuracy: 0.0660\n",
+      "Epoch 2213/5000\n",
+      "919/919 - 3s - loss: 1.4497 - accuracy: 0.5368 - val_loss: 3.6310 - val_accuracy: 0.0660\n",
+      "Epoch 2214/5000\n",
+      "919/919 - 3s - loss: 1.4470 - accuracy: 0.5364 - val_loss: 3.6403 - val_accuracy: 0.0663\n",
+      "Epoch 2215/5000\n",
+      "919/919 - 3s - loss: 1.4598 - accuracy: 0.5342 - val_loss: 3.6459 - val_accuracy: 0.0662\n",
+      "Epoch 2216/5000\n",
+      "919/919 - 3s - loss: 1.4539 - accuracy: 0.5341 - val_loss: 3.6415 - val_accuracy: 0.0655\n",
+      "Epoch 2217/5000\n",
+      "919/919 - 3s - loss: 1.4781 - accuracy: 0.5337 - val_loss: 3.6434 - val_accuracy: 0.0658\n",
+      "Epoch 2218/5000\n",
+      "919/919 - 3s - loss: 1.4407 - accuracy: 0.5411 - val_loss: 3.6468 - val_accuracy: 0.0657\n",
+      "Epoch 2219/5000\n",
+      "919/919 - 3s - loss: 1.4542 - accuracy: 0.5331 - val_loss: 3.6514 - val_accuracy: 0.0653\n",
+      "Epoch 2220/5000\n",
+      "919/919 - 3s - loss: 1.4719 - accuracy: 0.5339 - val_loss: 3.6499 - val_accuracy: 0.0657\n",
+      "Epoch 2221/5000\n",
+      "919/919 - 3s - loss: 1.5829 - accuracy: 0.5358 - val_loss: 3.6453 - val_accuracy: 0.0652\n",
+      "Epoch 2222/5000\n",
+      "919/919 - 3s - loss: 1.4649 - accuracy: 0.5303 - val_loss: 3.6559 - val_accuracy: 0.0653\n",
+      "Epoch 2223/5000\n",
+      "919/919 - 3s - loss: 1.4481 - accuracy: 0.5332 - val_loss: 3.6509 - val_accuracy: 0.0660\n",
+      "Epoch 2224/5000\n",
+      "919/919 - 3s - loss: 1.4522 - accuracy: 0.5339 - val_loss: 3.6497 - val_accuracy: 0.0658\n",
+      "Epoch 2225/5000\n",
+      "919/919 - 3s - loss: 1.4428 - accuracy: 0.5346 - val_loss: 3.6502 - val_accuracy: 0.0657\n",
+      "Epoch 2226/5000\n",
+      "919/919 - 3s - loss: 1.4439 - accuracy: 0.5371 - val_loss: 3.6497 - val_accuracy: 0.0659\n",
+      "Epoch 2227/5000\n",
+      "919/919 - 3s - loss: 1.4544 - accuracy: 0.5320 - val_loss: 3.6474 - val_accuracy: 0.0659\n",
+      "Epoch 2228/5000\n",
+      "919/919 - 3s - loss: 1.4530 - accuracy: 0.5359 - val_loss: 3.6355 - val_accuracy: 0.0656\n",
+      "Epoch 2229/5000\n",
+      "919/919 - 3s - loss: 1.4438 - accuracy: 0.5369 - val_loss: 3.6350 - val_accuracy: 0.0660\n",
+      "Epoch 2230/5000\n",
+      "919/919 - 3s - loss: 1.4429 - accuracy: 0.5352 - val_loss: 3.6368 - val_accuracy: 0.0658\n",
+      "Epoch 2231/5000\n",
+      "919/919 - 3s - loss: 1.4766 - accuracy: 0.5356 - val_loss: 3.6407 - val_accuracy: 0.0658\n",
+      "Epoch 2232/5000\n",
+      "919/919 - 3s - loss: 1.4643 - accuracy: 0.5280 - val_loss: 3.6361 - val_accuracy: 0.0659\n",
+      "Epoch 2233/5000\n",
+      "919/919 - 3s - loss: 1.4427 - accuracy: 0.5381 - val_loss: 3.6322 - val_accuracy: 0.0659\n",
+      "Epoch 2234/5000\n",
+      "919/919 - 3s - loss: 1.4448 - accuracy: 0.5375 - val_loss: 3.6333 - val_accuracy: 0.0660\n",
+      "Epoch 2235/5000\n",
+      "919/919 - 3s - loss: 1.4456 - accuracy: 0.5314 - val_loss: 3.6451 - val_accuracy: 0.0659\n",
+      "Epoch 2236/5000\n",
+      "919/919 - 3s - loss: 1.4502 - accuracy: 0.5369 - val_loss: 3.6432 - val_accuracy: 0.0661\n",
+      "Epoch 2237/5000\n",
+      "919/919 - 3s - loss: 1.4444 - accuracy: 0.5375 - val_loss: 3.6465 - val_accuracy: 0.0659\n",
+      "Epoch 2238/5000\n",
+      "919/919 - 3s - loss: 1.4321 - accuracy: 0.5392 - val_loss: 3.6453 - val_accuracy: 0.0661\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2239/5000\n",
+      "919/919 - 3s - loss: 1.4588 - accuracy: 0.5344 - val_loss: 3.6535 - val_accuracy: 0.0658\n",
+      "Epoch 2240/5000\n",
+      "919/919 - 3s - loss: 1.4479 - accuracy: 0.5388 - val_loss: 3.6486 - val_accuracy: 0.0661\n",
+      "Epoch 2241/5000\n",
+      "919/919 - 3s - loss: 1.4490 - accuracy: 0.5354 - val_loss: 3.6429 - val_accuracy: 0.0657\n",
+      "Epoch 2242/5000\n",
+      "919/919 - 3s - loss: 1.4716 - accuracy: 0.5341 - val_loss: 3.6310 - val_accuracy: 0.0660\n",
+      "Epoch 2243/5000\n",
+      "919/919 - 3s - loss: 1.4403 - accuracy: 0.5433 - val_loss: 3.6378 - val_accuracy: 0.0667\n",
+      "Epoch 2244/5000\n",
+      "919/919 - 3s - loss: 1.4374 - accuracy: 0.5398 - val_loss: 3.6400 - val_accuracy: 0.0662\n",
+      "Epoch 2245/5000\n",
+      "919/919 - 3s - loss: 1.4503 - accuracy: 0.5367 - val_loss: 3.6370 - val_accuracy: 0.0660\n",
+      "Epoch 2246/5000\n",
+      "919/919 - 3s - loss: 1.4462 - accuracy: 0.5421 - val_loss: 3.6498 - val_accuracy: 0.0663\n",
+      "Epoch 2247/5000\n",
+      "919/919 - 3s - loss: 1.4606 - accuracy: 0.5359 - val_loss: 3.6404 - val_accuracy: 0.0664\n",
+      "Epoch 2248/5000\n",
+      "919/919 - 3s - loss: 1.4393 - accuracy: 0.5369 - val_loss: 3.6495 - val_accuracy: 0.0660\n",
+      "Epoch 2249/5000\n",
+      "919/919 - 3s - loss: 1.4436 - accuracy: 0.5367 - val_loss: 3.6425 - val_accuracy: 0.0659\n",
+      "Epoch 2250/5000\n",
+      "919/919 - 3s - loss: 1.4531 - accuracy: 0.5361 - val_loss: 3.6353 - val_accuracy: 0.0658\n",
+      "Epoch 2251/5000\n",
+      "919/919 - 3s - loss: 1.4492 - accuracy: 0.5363 - val_loss: 3.6403 - val_accuracy: 0.0654\n",
+      "Epoch 2252/5000\n",
+      "919/919 - 3s - loss: 1.4382 - accuracy: 0.5394 - val_loss: 3.6359 - val_accuracy: 0.0659\n",
+      "Epoch 2253/5000\n",
+      "919/919 - 3s - loss: 1.4868 - accuracy: 0.5372 - val_loss: 3.6646 - val_accuracy: 0.0657\n",
+      "Epoch 2254/5000\n",
+      "919/919 - 3s - loss: 1.4286 - accuracy: 0.5397 - val_loss: 3.6560 - val_accuracy: 0.0658\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2255/5000\n",
+      "919/919 - 3s - loss: 1.4393 - accuracy: 0.5390 - val_loss: 3.6465 - val_accuracy: 0.0660\n",
+      "Epoch 2256/5000\n",
+      "919/919 - 3s - loss: 1.4637 - accuracy: 0.5385 - val_loss: 3.6539 - val_accuracy: 0.0663\n",
+      "Epoch 2257/5000\n",
+      "919/919 - 3s - loss: 1.4658 - accuracy: 0.5401 - val_loss: 3.6532 - val_accuracy: 0.0662\n",
+      "Epoch 2258/5000\n",
+      "919/919 - 3s - loss: 1.4679 - accuracy: 0.5372 - val_loss: 3.6550 - val_accuracy: 0.0663\n",
+      "Epoch 2259/5000\n",
+      "919/919 - 3s - loss: 1.4543 - accuracy: 0.5354 - val_loss: 3.6499 - val_accuracy: 0.0667\n",
+      "Epoch 2260/5000\n",
+      "919/919 - 3s - loss: 1.4438 - accuracy: 0.5373 - val_loss: 3.6523 - val_accuracy: 0.0665\n",
+      "Epoch 2261/5000\n",
+      "919/919 - 3s - loss: 1.4630 - accuracy: 0.5363 - val_loss: 3.6516 - val_accuracy: 0.0666\n",
+      "Epoch 2262/5000\n",
+      "919/919 - 3s - loss: 1.4557 - accuracy: 0.5331 - val_loss: 3.6597 - val_accuracy: 0.0666\n",
+      "Epoch 2263/5000\n",
+      "919/919 - 3s - loss: 1.4438 - accuracy: 0.5384 - val_loss: 3.6594 - val_accuracy: 0.0666\n",
+      "Epoch 2264/5000\n",
+      "919/919 - 3s - loss: 1.4397 - accuracy: 0.5414 - val_loss: 3.6609 - val_accuracy: 0.0665\n",
+      "Epoch 2265/5000\n",
+      "919/919 - 3s - loss: 1.5560 - accuracy: 0.5351 - val_loss: 3.6524 - val_accuracy: 0.0663\n",
+      "Epoch 2266/5000\n",
+      "919/919 - 3s - loss: 1.4356 - accuracy: 0.5424 - val_loss: 3.6463 - val_accuracy: 0.0657\n",
+      "Epoch 2267/5000\n",
+      "919/919 - 3s - loss: 1.4428 - accuracy: 0.5395 - val_loss: 3.6417 - val_accuracy: 0.0659\n",
+      "Epoch 2268/5000\n",
+      "919/919 - 3s - loss: 1.4579 - accuracy: 0.5383 - val_loss: 3.6384 - val_accuracy: 0.0659\n",
+      "Epoch 2269/5000\n",
+      "919/919 - 3s - loss: 1.4353 - accuracy: 0.5397 - val_loss: 3.6343 - val_accuracy: 0.0666\n",
+      "Epoch 2270/5000\n",
+      "919/919 - 3s - loss: 1.4372 - accuracy: 0.5397 - val_loss: 3.6364 - val_accuracy: 0.0668\n",
+      "Epoch 2271/5000\n",
+      "919/919 - 3s - loss: 1.4172 - accuracy: 0.5405 - val_loss: 3.6429 - val_accuracy: 0.0668\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2272/5000\n",
+      "919/919 - 3s - loss: 1.4446 - accuracy: 0.5390 - val_loss: 3.6390 - val_accuracy: 0.0666\n",
+      "Epoch 2273/5000\n",
+      "919/919 - 3s - loss: 1.4441 - accuracy: 0.5374 - val_loss: 3.6335 - val_accuracy: 0.0663\n",
+      "Epoch 2274/5000\n",
+      "919/919 - 3s - loss: 1.4215 - accuracy: 0.5403 - val_loss: 3.6518 - val_accuracy: 0.0668\n",
+      "Epoch 2275/5000\n",
+      "919/919 - 3s - loss: 1.4521 - accuracy: 0.5388 - val_loss: 3.6444 - val_accuracy: 0.0665\n",
+      "Epoch 2276/5000\n",
+      "919/919 - 3s - loss: 1.4362 - accuracy: 0.5431 - val_loss: 3.6377 - val_accuracy: 0.0664\n",
+      "Epoch 2277/5000\n",
+      "919/919 - 3s - loss: 1.4390 - accuracy: 0.5345 - val_loss: 3.6365 - val_accuracy: 0.0661\n",
+      "Epoch 2278/5000\n",
+      "919/919 - 3s - loss: 1.4318 - accuracy: 0.5380 - val_loss: 3.6371 - val_accuracy: 0.0659\n",
+      "Epoch 2279/5000\n",
+      "919/919 - 3s - loss: 1.4549 - accuracy: 0.5377 - val_loss: 3.6411 - val_accuracy: 0.0659\n",
+      "Epoch 2280/5000\n",
+      "919/919 - 3s - loss: 1.4527 - accuracy: 0.5375 - val_loss: 3.6505 - val_accuracy: 0.0662\n",
+      "Epoch 2281/5000\n",
+      "919/919 - 3s - loss: 1.4496 - accuracy: 0.5373 - val_loss: 3.6565 - val_accuracy: 0.0664\n",
+      "Epoch 2282/5000\n",
+      "919/919 - 3s - loss: 1.4382 - accuracy: 0.5397 - val_loss: 3.6440 - val_accuracy: 0.0665\n",
+      "Epoch 2283/5000\n",
+      "919/919 - 3s - loss: 1.4387 - accuracy: 0.5367 - val_loss: 3.6561 - val_accuracy: 0.0670\n",
+      "Epoch 2284/5000\n",
+      "919/919 - 3s - loss: 1.4449 - accuracy: 0.5395 - val_loss: 3.6502 - val_accuracy: 0.0670\n",
+      "Epoch 2285/5000\n",
+      "919/919 - 3s - loss: 1.4213 - accuracy: 0.5455 - val_loss: 3.6629 - val_accuracy: 0.0667\n",
+      "Epoch 2286/5000\n",
+      "919/919 - 3s - loss: 1.4484 - accuracy: 0.5380 - val_loss: 3.6566 - val_accuracy: 0.0667\n",
+      "Epoch 2287/5000\n",
+      "919/919 - 3s - loss: 1.4475 - accuracy: 0.5405 - val_loss: 3.6548 - val_accuracy: 0.0660\n",
+      "Epoch 2288/5000\n",
+      "919/919 - 3s - loss: 1.4388 - accuracy: 0.5421 - val_loss: 3.6548 - val_accuracy: 0.0662\n",
+      "Epoch 2289/5000\n",
+      "919/919 - 3s - loss: 1.4605 - accuracy: 0.5392 - val_loss: 3.6608 - val_accuracy: 0.0666\n",
+      "Epoch 2290/5000\n",
+      "919/919 - 3s - loss: 1.4366 - accuracy: 0.5407 - val_loss: 3.6580 - val_accuracy: 0.0668\n",
+      "Epoch 2291/5000\n",
+      "919/919 - 3s - loss: 1.4312 - accuracy: 0.5441 - val_loss: 3.6636 - val_accuracy: 0.0665\n",
+      "Epoch 2292/5000\n",
+      "919/919 - 3s - loss: 1.4531 - accuracy: 0.5421 - val_loss: 3.6583 - val_accuracy: 0.0665\n",
+      "Epoch 2293/5000\n",
+      "919/919 - 3s - loss: 1.4429 - accuracy: 0.5405 - val_loss: 3.6605 - val_accuracy: 0.0663\n",
+      "Epoch 2294/5000\n",
+      "919/919 - 3s - loss: 1.4795 - accuracy: 0.5418 - val_loss: 3.6587 - val_accuracy: 0.0662\n",
+      "Epoch 2295/5000\n",
+      "919/919 - 3s - loss: 1.4304 - accuracy: 0.5392 - val_loss: 3.6605 - val_accuracy: 0.0660\n",
+      "Epoch 2296/5000\n",
+      "919/919 - 3s - loss: 1.4439 - accuracy: 0.5371 - val_loss: 3.6612 - val_accuracy: 0.0665\n",
+      "Epoch 2297/5000\n",
+      "919/919 - 3s - loss: 1.4377 - accuracy: 0.5381 - val_loss: 3.6560 - val_accuracy: 0.0668\n",
+      "Epoch 2298/5000\n",
+      "919/919 - 3s - loss: 1.4436 - accuracy: 0.5397 - val_loss: 3.6462 - val_accuracy: 0.0665\n",
+      "Epoch 2299/5000\n",
+      "919/919 - 3s - loss: 1.4459 - accuracy: 0.5393 - val_loss: 3.6450 - val_accuracy: 0.0667\n",
+      "Epoch 2300/5000\n",
+      "919/919 - 3s - loss: 1.4255 - accuracy: 0.5433 - val_loss: 3.6454 - val_accuracy: 0.0664\n",
+      "Epoch 2301/5000\n",
+      "919/919 - 3s - loss: 1.4771 - accuracy: 0.5430 - val_loss: 3.6562 - val_accuracy: 0.0667\n",
+      "Epoch 2302/5000\n",
+      "919/919 - 3s - loss: 1.4395 - accuracy: 0.5393 - val_loss: 3.6659 - val_accuracy: 0.0651\n",
+      "Epoch 2303/5000\n",
+      "919/919 - 3s - loss: 1.4342 - accuracy: 0.5420 - val_loss: 3.6688 - val_accuracy: 0.0654\n",
+      "Epoch 2304/5000\n",
+      "919/919 - 3s - loss: 1.4193 - accuracy: 0.5452 - val_loss: 3.6634 - val_accuracy: 0.0663\n",
+      "Epoch 2305/5000\n",
+      "919/919 - 3s - loss: 1.4338 - accuracy: 0.5417 - val_loss: 3.6498 - val_accuracy: 0.0667\n",
+      "Epoch 2306/5000\n",
+      "919/919 - 3s - loss: 1.4394 - accuracy: 0.5381 - val_loss: 3.6518 - val_accuracy: 0.0671\n",
+      "Epoch 2307/5000\n",
+      "919/919 - 3s - loss: 1.4528 - accuracy: 0.5393 - val_loss: 3.6517 - val_accuracy: 0.0670\n",
+      "Epoch 2308/5000\n",
+      "919/919 - 3s - loss: 1.4465 - accuracy: 0.5389 - val_loss: 3.6484 - val_accuracy: 0.0670\n",
+      "Epoch 2309/5000\n",
+      "919/919 - 3s - loss: 1.4307 - accuracy: 0.5409 - val_loss: 3.6411 - val_accuracy: 0.0670\n",
+      "Epoch 2310/5000\n",
+      "919/919 - 3s - loss: 1.4418 - accuracy: 0.5404 - val_loss: 3.6471 - val_accuracy: 0.0666\n",
+      "Epoch 2311/5000\n",
+      "919/919 - 3s - loss: 1.4289 - accuracy: 0.5406 - val_loss: 3.6548 - val_accuracy: 0.0668\n",
+      "Epoch 2312/5000\n",
+      "919/919 - 3s - loss: 1.4381 - accuracy: 0.5407 - val_loss: 3.6458 - val_accuracy: 0.0672\n",
+      "Epoch 2313/5000\n",
+      "919/919 - 3s - loss: 1.4327 - accuracy: 0.5422 - val_loss: 3.6525 - val_accuracy: 0.0667\n",
+      "Epoch 2314/5000\n",
+      "919/919 - 3s - loss: 1.4382 - accuracy: 0.5411 - val_loss: 3.6426 - val_accuracy: 0.0667\n",
+      "Epoch 2315/5000\n",
+      "919/919 - 3s - loss: 1.4341 - accuracy: 0.5416 - val_loss: 3.6471 - val_accuracy: 0.0674\n",
+      "Epoch 2316/5000\n",
+      "919/919 - 3s - loss: 1.4407 - accuracy: 0.5412 - val_loss: 3.6470 - val_accuracy: 0.0668\n",
+      "Epoch 2317/5000\n",
+      "919/919 - 3s - loss: 1.4255 - accuracy: 0.5407 - val_loss: 3.6422 - val_accuracy: 0.0668\n",
+      "Epoch 2318/5000\n",
+      "919/919 - 3s - loss: 1.4272 - accuracy: 0.5431 - val_loss: 3.6426 - val_accuracy: 0.0673\n",
+      "Epoch 2319/5000\n",
+      "919/919 - 3s - loss: 1.4438 - accuracy: 0.5404 - val_loss: 3.6538 - val_accuracy: 0.0672\n",
+      "Epoch 2320/5000\n",
+      "919/919 - 3s - loss: 1.4307 - accuracy: 0.5418 - val_loss: 3.6559 - val_accuracy: 0.0668\n",
+      "Epoch 2321/5000\n",
+      "919/919 - 3s - loss: 1.4313 - accuracy: 0.5384 - val_loss: 3.6626 - val_accuracy: 0.0668\n",
+      "Epoch 2322/5000\n",
+      "919/919 - 3s - loss: 1.4401 - accuracy: 0.5424 - val_loss: 3.6543 - val_accuracy: 0.0681\n",
+      "Epoch 2323/5000\n",
+      "919/919 - 3s - loss: 1.4304 - accuracy: 0.5429 - val_loss: 3.6462 - val_accuracy: 0.0678\n",
+      "Epoch 2324/5000\n",
+      "919/919 - 3s - loss: 1.4244 - accuracy: 0.5412 - val_loss: 3.6388 - val_accuracy: 0.0680\n",
+      "Epoch 2325/5000\n",
+      "919/919 - 3s - loss: 1.4287 - accuracy: 0.5431 - val_loss: 3.6584 - val_accuracy: 0.0678\n",
+      "Epoch 2326/5000\n",
+      "919/919 - 3s - loss: 1.4533 - accuracy: 0.5374 - val_loss: 3.6547 - val_accuracy: 0.0678\n",
+      "Epoch 2327/5000\n",
+      "919/919 - 3s - loss: 1.4395 - accuracy: 0.5381 - val_loss: 3.6586 - val_accuracy: 0.0674\n",
+      "Epoch 2328/5000\n",
+      "919/919 - 3s - loss: 1.5873 - accuracy: 0.5437 - val_loss: 3.6547 - val_accuracy: 0.0674\n",
+      "Epoch 2329/5000\n",
+      "919/919 - 3s - loss: 1.4311 - accuracy: 0.5403 - val_loss: 3.6512 - val_accuracy: 0.0672\n",
+      "Epoch 2330/5000\n",
+      "919/919 - 3s - loss: 1.4630 - accuracy: 0.5405 - val_loss: 3.6429 - val_accuracy: 0.0672\n",
+      "Epoch 2331/5000\n",
+      "919/919 - 3s - loss: 1.4299 - accuracy: 0.5447 - val_loss: 3.6520 - val_accuracy: 0.0672\n",
+      "Epoch 2332/5000\n",
+      "919/919 - 3s - loss: 1.4303 - accuracy: 0.5386 - val_loss: 3.6382 - val_accuracy: 0.0672\n",
+      "Epoch 2333/5000\n",
+      "919/919 - 3s - loss: 1.4353 - accuracy: 0.5401 - val_loss: 3.6382 - val_accuracy: 0.0668\n",
+      "Epoch 2334/5000\n",
+      "919/919 - 3s - loss: 1.4252 - accuracy: 0.5446 - val_loss: 3.6433 - val_accuracy: 0.0668\n",
+      "Epoch 2335/5000\n",
+      "919/919 - 3s - loss: 1.4313 - accuracy: 0.5435 - val_loss: 3.6414 - val_accuracy: 0.0675\n",
+      "Epoch 2336/5000\n",
+      "919/919 - 3s - loss: 1.4220 - accuracy: 0.5448 - val_loss: 3.6471 - val_accuracy: 0.0677\n",
+      "Epoch 2337/5000\n",
+      "919/919 - 3s - loss: 1.4283 - accuracy: 0.5447 - val_loss: 3.6551 - val_accuracy: 0.0676\n",
+      "Epoch 2338/5000\n",
+      "919/919 - 3s - loss: 1.4276 - accuracy: 0.5441 - val_loss: 3.6429 - val_accuracy: 0.0679\n",
+      "Epoch 2339/5000\n",
+      "919/919 - 3s - loss: 1.5414 - accuracy: 0.5439 - val_loss: 3.6514 - val_accuracy: 0.0674\n",
+      "Epoch 2340/5000\n",
+      "919/919 - 3s - loss: 1.4504 - accuracy: 0.5414 - val_loss: 3.6556 - val_accuracy: 0.0670\n",
+      "Epoch 2341/5000\n",
+      "919/919 - 3s - loss: 1.4313 - accuracy: 0.5361 - val_loss: 3.6487 - val_accuracy: 0.0672\n",
+      "Epoch 2342/5000\n",
+      "919/919 - 3s - loss: 1.4274 - accuracy: 0.5454 - val_loss: 3.6555 - val_accuracy: 0.0674\n",
+      "Epoch 2343/5000\n",
+      "919/919 - 3s - loss: 1.5282 - accuracy: 0.5455 - val_loss: 3.6511 - val_accuracy: 0.0671\n",
+      "Epoch 2344/5000\n",
+      "919/919 - 3s - loss: 1.4384 - accuracy: 0.5418 - val_loss: 3.6514 - val_accuracy: 0.0674\n",
+      "Epoch 2345/5000\n",
+      "919/919 - 3s - loss: 1.4340 - accuracy: 0.5427 - val_loss: 3.6541 - val_accuracy: 0.0674\n",
+      "Epoch 2346/5000\n",
+      "919/919 - 3s - loss: 1.4265 - accuracy: 0.5458 - val_loss: 3.6446 - val_accuracy: 0.0667\n",
+      "Epoch 2347/5000\n",
+      "919/919 - 3s - loss: 1.4264 - accuracy: 0.5401 - val_loss: 3.6510 - val_accuracy: 0.0661\n",
+      "Epoch 2348/5000\n",
+      "919/919 - 3s - loss: 1.4273 - accuracy: 0.5422 - val_loss: 3.6507 - val_accuracy: 0.0667\n",
+      "Epoch 2349/5000\n",
+      "919/919 - 3s - loss: 1.4331 - accuracy: 0.5430 - val_loss: 3.6573 - val_accuracy: 0.0675\n",
+      "Epoch 2350/5000\n",
+      "919/919 - 3s - loss: 1.4564 - accuracy: 0.5415 - val_loss: 3.6560 - val_accuracy: 0.0668\n",
+      "Epoch 2351/5000\n",
+      "919/919 - 3s - loss: 1.4304 - accuracy: 0.5471 - val_loss: 3.6532 - val_accuracy: 0.0681\n",
+      "Epoch 2352/5000\n",
+      "919/919 - 3s - loss: 1.4310 - accuracy: 0.5427 - val_loss: 3.6576 - val_accuracy: 0.0682\n",
+      "Epoch 2353/5000\n",
+      "919/919 - 3s - loss: 1.4475 - accuracy: 0.5386 - val_loss: 3.6548 - val_accuracy: 0.0680\n",
+      "Epoch 2354/5000\n",
+      "919/919 - 3s - loss: 1.4179 - accuracy: 0.5455 - val_loss: 3.6504 - val_accuracy: 0.0677\n",
+      "Epoch 2355/5000\n",
+      "919/919 - 3s - loss: 1.4288 - accuracy: 0.5446 - val_loss: 3.6433 - val_accuracy: 0.0679\n",
+      "Epoch 2356/5000\n",
+      "919/919 - 3s - loss: 1.4639 - accuracy: 0.5427 - val_loss: 3.6486 - val_accuracy: 0.0680\n",
+      "Epoch 2357/5000\n",
+      "919/919 - 3s - loss: 1.4422 - accuracy: 0.5452 - val_loss: 3.6523 - val_accuracy: 0.0678\n",
+      "Epoch 2358/5000\n",
+      "919/919 - 3s - loss: 1.4403 - accuracy: 0.5418 - val_loss: 3.6544 - val_accuracy: 0.0677\n",
+      "Epoch 2359/5000\n",
+      "919/919 - 3s - loss: 1.4543 - accuracy: 0.5405 - val_loss: 3.6557 - val_accuracy: 0.0673\n",
+      "Epoch 2360/5000\n",
+      "919/919 - 3s - loss: 1.4289 - accuracy: 0.5480 - val_loss: 3.6531 - val_accuracy: 0.0671\n",
+      "Epoch 2361/5000\n",
+      "919/919 - 3s - loss: 1.4742 - accuracy: 0.5454 - val_loss: 3.6497 - val_accuracy: 0.0673\n",
+      "Epoch 2362/5000\n",
+      "919/919 - 3s - loss: 1.4300 - accuracy: 0.5463 - val_loss: 3.6559 - val_accuracy: 0.0679\n",
+      "Epoch 2363/5000\n",
+      "919/919 - 3s - loss: 1.4182 - accuracy: 0.5452 - val_loss: 3.6513 - val_accuracy: 0.0676\n",
+      "Epoch 2364/5000\n",
+      "919/919 - 3s - loss: 1.4310 - accuracy: 0.5476 - val_loss: 3.6486 - val_accuracy: 0.0677\n",
+      "Epoch 2365/5000\n",
+      "919/919 - 3s - loss: 1.4510 - accuracy: 0.5407 - val_loss: 3.6509 - val_accuracy: 0.0674\n",
+      "Epoch 2366/5000\n",
+      "919/919 - 3s - loss: 1.4106 - accuracy: 0.5469 - val_loss: 3.6558 - val_accuracy: 0.0671\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2367/5000\n",
+      "919/919 - 3s - loss: 1.4299 - accuracy: 0.5433 - val_loss: 3.6659 - val_accuracy: 0.0670\n",
+      "Epoch 2368/5000\n",
+      "919/919 - 3s - loss: 1.4325 - accuracy: 0.5429 - val_loss: 3.6663 - val_accuracy: 0.0677\n",
+      "Epoch 2369/5000\n",
+      "919/919 - 3s - loss: 1.4190 - accuracy: 0.5414 - val_loss: 3.6774 - val_accuracy: 0.0664\n",
+      "Epoch 2370/5000\n",
+      "919/919 - 3s - loss: 1.4234 - accuracy: 0.5444 - val_loss: 3.6679 - val_accuracy: 0.0670\n",
+      "Epoch 2371/5000\n",
+      "919/919 - 3s - loss: 1.4274 - accuracy: 0.5431 - val_loss: 3.6608 - val_accuracy: 0.0670\n",
+      "Epoch 2372/5000\n",
+      "919/919 - 3s - loss: 1.4157 - accuracy: 0.5478 - val_loss: 3.6738 - val_accuracy: 0.0667\n",
+      "Epoch 2373/5000\n",
+      "919/919 - 3s - loss: 1.4302 - accuracy: 0.5449 - val_loss: 3.6726 - val_accuracy: 0.0670\n",
+      "Epoch 2374/5000\n",
+      "919/919 - 3s - loss: 1.4172 - accuracy: 0.5419 - val_loss: 3.6806 - val_accuracy: 0.0671\n",
+      "Epoch 2375/5000\n",
+      "919/919 - 3s - loss: 1.4201 - accuracy: 0.5522 - val_loss: 3.6668 - val_accuracy: 0.0681\n",
+      "Epoch 2376/5000\n",
+      "919/919 - 3s - loss: 1.4143 - accuracy: 0.5449 - val_loss: 3.6603 - val_accuracy: 0.0680\n",
+      "Epoch 2377/5000\n",
+      "919/919 - 3s - loss: 1.4226 - accuracy: 0.5430 - val_loss: 3.6621 - val_accuracy: 0.0676\n",
+      "Epoch 2378/5000\n",
+      "919/919 - 3s - loss: 1.4084 - accuracy: 0.5439 - val_loss: 3.6617 - val_accuracy: 0.0678\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2379/5000\n",
+      "919/919 - 3s - loss: 1.4207 - accuracy: 0.5493 - val_loss: 3.6654 - val_accuracy: 0.0677\n",
+      "Epoch 2380/5000\n",
+      "919/919 - 3s - loss: 1.4140 - accuracy: 0.5451 - val_loss: 3.6524 - val_accuracy: 0.0681\n",
+      "Epoch 2381/5000\n",
+      "919/919 - 3s - loss: 1.4187 - accuracy: 0.5472 - val_loss: 3.6530 - val_accuracy: 0.0684\n",
+      "Epoch 2382/5000\n",
+      "919/919 - 3s - loss: 1.4272 - accuracy: 0.5436 - val_loss: 3.6480 - val_accuracy: 0.0686\n",
+      "Epoch 2383/5000\n",
+      "919/919 - 3s - loss: 1.4058 - accuracy: 0.5516 - val_loss: 3.6584 - val_accuracy: 0.0686\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2384/5000\n",
+      "919/919 - 3s - loss: 1.4266 - accuracy: 0.5416 - val_loss: 3.6600 - val_accuracy: 0.0677\n",
+      "Epoch 2385/5000\n",
+      "919/919 - 3s - loss: 1.4592 - accuracy: 0.5426 - val_loss: 3.6618 - val_accuracy: 0.0678\n",
+      "Epoch 2386/5000\n",
+      "919/919 - 3s - loss: 1.4719 - accuracy: 0.5420 - val_loss: 3.6569 - val_accuracy: 0.0677\n",
+      "Epoch 2387/5000\n",
+      "919/919 - 3s - loss: 1.4204 - accuracy: 0.5467 - val_loss: 3.6554 - val_accuracy: 0.0684\n",
+      "Epoch 2388/5000\n",
+      "919/919 - 3s - loss: 1.4268 - accuracy: 0.5456 - val_loss: 3.6562 - val_accuracy: 0.0677\n",
+      "Epoch 2389/5000\n",
+      "919/919 - 3s - loss: 1.5225 - accuracy: 0.5422 - val_loss: 3.6564 - val_accuracy: 0.0676\n",
+      "Epoch 2390/5000\n",
+      "919/919 - 3s - loss: 1.4161 - accuracy: 0.5465 - val_loss: 3.6712 - val_accuracy: 0.0675\n",
+      "Epoch 2391/5000\n",
+      "919/919 - 3s - loss: 1.4939 - accuracy: 0.5450 - val_loss: 3.6769 - val_accuracy: 0.0679\n",
+      "Epoch 2392/5000\n",
+      "919/919 - 3s - loss: 1.4072 - accuracy: 0.5493 - val_loss: 3.6777 - val_accuracy: 0.0683\n",
+      "Epoch 2393/5000\n",
+      "919/919 - 3s - loss: 1.4255 - accuracy: 0.5464 - val_loss: 3.6841 - val_accuracy: 0.0683\n",
+      "Epoch 2394/5000\n",
+      "919/919 - 3s - loss: 1.4324 - accuracy: 0.5456 - val_loss: 3.6794 - val_accuracy: 0.0681\n",
+      "Epoch 2395/5000\n",
+      "919/919 - 3s - loss: 1.5125 - accuracy: 0.5423 - val_loss: 3.6745 - val_accuracy: 0.0685\n",
+      "Epoch 2396/5000\n",
+      "919/919 - 3s - loss: 1.4115 - accuracy: 0.5474 - val_loss: 3.6633 - val_accuracy: 0.0684\n",
+      "Epoch 2397/5000\n",
+      "919/919 - 3s - loss: 1.4232 - accuracy: 0.5465 - val_loss: 3.6655 - val_accuracy: 0.0687\n",
+      "Epoch 2398/5000\n",
+      "919/919 - 3s - loss: 1.4156 - accuracy: 0.5478 - val_loss: 3.6645 - val_accuracy: 0.0683\n",
+      "Epoch 2399/5000\n",
+      "919/919 - 3s - loss: 1.4188 - accuracy: 0.5483 - val_loss: 3.6744 - val_accuracy: 0.0682\n",
+      "Epoch 2400/5000\n",
+      "919/919 - 3s - loss: 1.4139 - accuracy: 0.5460 - val_loss: 3.6695 - val_accuracy: 0.0680\n",
+      "Epoch 2401/5000\n",
+      "919/919 - 3s - loss: 1.4480 - accuracy: 0.5495 - val_loss: 3.6768 - val_accuracy: 0.0684\n",
+      "Epoch 2402/5000\n",
+      "919/919 - 3s - loss: 1.4231 - accuracy: 0.5475 - val_loss: 3.6643 - val_accuracy: 0.0682\n",
+      "Epoch 2403/5000\n",
+      "919/919 - 3s - loss: 1.5147 - accuracy: 0.5476 - val_loss: 3.6776 - val_accuracy: 0.0684\n",
+      "Epoch 2404/5000\n",
+      "919/919 - 3s - loss: 1.4017 - accuracy: 0.5473 - val_loss: 3.6729 - val_accuracy: 0.0683\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2405/5000\n",
+      "919/919 - 3s - loss: 1.4185 - accuracy: 0.5440 - val_loss: 3.6761 - val_accuracy: 0.0680\n",
+      "Epoch 2406/5000\n",
+      "919/919 - 3s - loss: 1.4269 - accuracy: 0.5482 - val_loss: 3.6703 - val_accuracy: 0.0684\n",
+      "Epoch 2407/5000\n",
+      "919/919 - 3s - loss: 1.4203 - accuracy: 0.5459 - val_loss: 3.6765 - val_accuracy: 0.0688\n",
+      "Epoch 2408/5000\n",
+      "919/919 - 3s - loss: 1.4163 - accuracy: 0.5473 - val_loss: 3.6742 - val_accuracy: 0.0690\n",
+      "Epoch 2409/5000\n",
+      "919/919 - 3s - loss: 1.4138 - accuracy: 0.5496 - val_loss: 3.6747 - val_accuracy: 0.0686\n",
+      "Epoch 2410/5000\n",
+      "919/919 - 3s - loss: 1.4203 - accuracy: 0.5441 - val_loss: 3.6698 - val_accuracy: 0.0686\n",
+      "Epoch 2411/5000\n",
+      "919/919 - 3s - loss: 1.4069 - accuracy: 0.5495 - val_loss: 3.6578 - val_accuracy: 0.0686\n",
+      "Epoch 2412/5000\n",
+      "919/919 - 3s - loss: 1.4253 - accuracy: 0.5438 - val_loss: 3.6623 - val_accuracy: 0.0682\n",
+      "Epoch 2413/5000\n",
+      "919/919 - 3s - loss: 1.4161 - accuracy: 0.5451 - val_loss: 3.6644 - val_accuracy: 0.0686\n",
+      "Epoch 2414/5000\n",
+      "919/919 - 3s - loss: 1.4212 - accuracy: 0.5465 - val_loss: 3.6652 - val_accuracy: 0.0687\n",
+      "Epoch 2415/5000\n",
+      "919/919 - 3s - loss: 1.4066 - accuracy: 0.5478 - val_loss: 3.6807 - val_accuracy: 0.0686\n",
+      "Epoch 2416/5000\n",
+      "919/919 - 3s - loss: 1.4175 - accuracy: 0.5476 - val_loss: 3.6948 - val_accuracy: 0.0694\n",
+      "Epoch 2417/5000\n",
+      "919/919 - 3s - loss: 1.4178 - accuracy: 0.5450 - val_loss: 3.6879 - val_accuracy: 0.0696\n",
+      "Epoch 2418/5000\n",
+      "919/919 - 3s - loss: 1.4127 - accuracy: 0.5487 - val_loss: 3.6924 - val_accuracy: 0.0695\n",
+      "Epoch 2419/5000\n",
+      "919/919 - 3s - loss: 1.4250 - accuracy: 0.5469 - val_loss: 3.6697 - val_accuracy: 0.0699\n",
+      "Epoch 2420/5000\n",
+      "919/919 - 3s - loss: 1.3975 - accuracy: 0.5476 - val_loss: 3.6801 - val_accuracy: 0.0694\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2421/5000\n",
+      "919/919 - 3s - loss: 1.4368 - accuracy: 0.5461 - val_loss: 3.6648 - val_accuracy: 0.0687\n",
+      "Epoch 2422/5000\n",
+      "919/919 - 3s - loss: 1.4211 - accuracy: 0.5490 - val_loss: 3.6556 - val_accuracy: 0.0693\n",
+      "Epoch 2423/5000\n",
+      "919/919 - 3s - loss: 1.4801 - accuracy: 0.5468 - val_loss: 3.6592 - val_accuracy: 0.0685\n",
+      "Epoch 2424/5000\n",
+      "919/919 - 3s - loss: 1.4018 - accuracy: 0.5499 - val_loss: 3.6613 - val_accuracy: 0.0687\n",
+      "Epoch 2425/5000\n",
+      "919/919 - 3s - loss: 1.4017 - accuracy: 0.5471 - val_loss: 3.6593 - val_accuracy: 0.0686\n",
+      "Epoch 2426/5000\n",
+      "919/919 - 3s - loss: 1.3962 - accuracy: 0.5528 - val_loss: 3.6624 - val_accuracy: 0.0694\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2427/5000\n",
+      "919/919 - 3s - loss: 1.4053 - accuracy: 0.5511 - val_loss: 3.6661 - val_accuracy: 0.0697\n",
+      "Epoch 2428/5000\n",
+      "919/919 - 3s - loss: 1.4031 - accuracy: 0.5503 - val_loss: 3.6727 - val_accuracy: 0.0688\n",
+      "Epoch 2429/5000\n",
+      "919/919 - 3s - loss: 1.4383 - accuracy: 0.5461 - val_loss: 3.6779 - val_accuracy: 0.0687\n",
+      "Epoch 2430/5000\n",
+      "919/919 - 3s - loss: 1.4638 - accuracy: 0.5464 - val_loss: 3.6860 - val_accuracy: 0.0689\n",
+      "Epoch 2431/5000\n",
+      "919/919 - 3s - loss: 1.4221 - accuracy: 0.5469 - val_loss: 3.6772 - val_accuracy: 0.0695\n",
+      "Epoch 2432/5000\n",
+      "919/919 - 3s - loss: 1.4790 - accuracy: 0.5505 - val_loss: 3.6745 - val_accuracy: 0.0706\n",
+      "Epoch 2433/5000\n",
+      "919/919 - 3s - loss: 1.4113 - accuracy: 0.5463 - val_loss: 3.6705 - val_accuracy: 0.0692\n",
+      "Epoch 2434/5000\n",
+      "919/919 - 3s - loss: 1.5419 - accuracy: 0.5490 - val_loss: 3.6717 - val_accuracy: 0.0693\n",
+      "Epoch 2435/5000\n",
+      "919/919 - 3s - loss: 1.4025 - accuracy: 0.5516 - val_loss: 3.6813 - val_accuracy: 0.0698\n",
+      "Epoch 2436/5000\n",
+      "919/919 - 3s - loss: 1.4014 - accuracy: 0.5528 - val_loss: 3.6891 - val_accuracy: 0.0691\n",
+      "Epoch 2437/5000\n",
+      "919/919 - 3s - loss: 1.4566 - accuracy: 0.5482 - val_loss: 3.6812 - val_accuracy: 0.0691\n",
+      "Epoch 2438/5000\n",
+      "919/919 - 3s - loss: 1.4397 - accuracy: 0.5547 - val_loss: 3.6819 - val_accuracy: 0.0695\n",
+      "Epoch 2439/5000\n",
+      "919/919 - 3s - loss: 1.4079 - accuracy: 0.5510 - val_loss: 3.6858 - val_accuracy: 0.0692\n",
+      "Epoch 2440/5000\n",
+      "919/919 - 3s - loss: 1.4189 - accuracy: 0.5495 - val_loss: 3.6673 - val_accuracy: 0.0699\n",
+      "Epoch 2441/5000\n",
+      "919/919 - 3s - loss: 1.4165 - accuracy: 0.5488 - val_loss: 3.6634 - val_accuracy: 0.0709\n",
+      "Epoch 2442/5000\n",
+      "919/919 - 3s - loss: 1.4236 - accuracy: 0.5494 - val_loss: 3.6606 - val_accuracy: 0.0697\n",
+      "Epoch 2443/5000\n",
+      "919/919 - 3s - loss: 1.4104 - accuracy: 0.5500 - val_loss: 3.6609 - val_accuracy: 0.0702\n",
+      "Epoch 2444/5000\n",
+      "919/919 - 3s - loss: 1.4159 - accuracy: 0.5492 - val_loss: 3.6642 - val_accuracy: 0.0712\n",
+      "Epoch 2445/5000\n",
+      "919/919 - 3s - loss: 1.4114 - accuracy: 0.5470 - val_loss: 3.6665 - val_accuracy: 0.0703\n",
+      "Epoch 2446/5000\n",
+      "919/919 - 3s - loss: 1.3979 - accuracy: 0.5518 - val_loss: 3.6736 - val_accuracy: 0.0697\n",
+      "Epoch 2447/5000\n",
+      "919/919 - 3s - loss: 1.5506 - accuracy: 0.5529 - val_loss: 3.6755 - val_accuracy: 0.0701\n",
+      "Epoch 2448/5000\n",
+      "919/919 - 3s - loss: 1.3981 - accuracy: 0.5539 - val_loss: 3.6793 - val_accuracy: 0.0711\n",
+      "Epoch 2449/5000\n",
+      "919/919 - 3s - loss: 1.4235 - accuracy: 0.5471 - val_loss: 3.6611 - val_accuracy: 0.0712\n",
+      "Epoch 2450/5000\n",
+      "919/919 - 3s - loss: 1.4075 - accuracy: 0.5503 - val_loss: 3.6653 - val_accuracy: 0.0708\n",
+      "Epoch 2451/5000\n",
+      "919/919 - 3s - loss: 1.4080 - accuracy: 0.5496 - val_loss: 3.6775 - val_accuracy: 0.0707\n",
+      "Epoch 2452/5000\n",
+      "919/919 - 3s - loss: 1.4055 - accuracy: 0.5495 - val_loss: 3.6679 - val_accuracy: 0.0713\n",
+      "Epoch 2453/5000\n",
+      "919/919 - 3s - loss: 1.4082 - accuracy: 0.5501 - val_loss: 3.6716 - val_accuracy: 0.0708\n",
+      "Epoch 2454/5000\n",
+      "919/919 - 3s - loss: 1.4120 - accuracy: 0.5523 - val_loss: 3.6710 - val_accuracy: 0.0716\n",
+      "Epoch 2455/5000\n",
+      "919/919 - 3s - loss: 1.5207 - accuracy: 0.5479 - val_loss: 3.6686 - val_accuracy: 0.0706\n",
+      "Epoch 2456/5000\n",
+      "919/919 - 3s - loss: 1.4034 - accuracy: 0.5536 - val_loss: 3.6867 - val_accuracy: 0.0695\n",
+      "Epoch 2457/5000\n",
+      "919/919 - 3s - loss: 1.4095 - accuracy: 0.5522 - val_loss: 3.6782 - val_accuracy: 0.0704\n",
+      "Epoch 2458/5000\n",
+      "919/919 - 3s - loss: 1.4313 - accuracy: 0.5529 - val_loss: 3.6716 - val_accuracy: 0.0704\n",
+      "Epoch 2459/5000\n",
+      "919/919 - 3s - loss: 1.4036 - accuracy: 0.5499 - val_loss: 3.6610 - val_accuracy: 0.0710\n",
+      "Epoch 2460/5000\n",
+      "919/919 - 3s - loss: 1.5146 - accuracy: 0.5507 - val_loss: 3.6708 - val_accuracy: 0.0709\n",
+      "Epoch 2461/5000\n",
+      "919/919 - 3s - loss: 1.4296 - accuracy: 0.5497 - val_loss: 3.6657 - val_accuracy: 0.0695\n",
+      "Epoch 2462/5000\n",
+      "919/919 - 3s - loss: 1.4001 - accuracy: 0.5549 - val_loss: 3.6736 - val_accuracy: 0.0699\n",
+      "Epoch 2463/5000\n",
+      "919/919 - 3s - loss: 1.4253 - accuracy: 0.5502 - val_loss: 3.6795 - val_accuracy: 0.0701\n",
+      "Epoch 2464/5000\n",
+      "919/919 - 3s - loss: 1.4059 - accuracy: 0.5502 - val_loss: 3.6782 - val_accuracy: 0.0711\n",
+      "Epoch 2465/5000\n",
+      "919/919 - 3s - loss: 1.4722 - accuracy: 0.5516 - val_loss: 3.6751 - val_accuracy: 0.0698\n",
+      "Epoch 2466/5000\n",
+      "919/919 - 3s - loss: 1.4215 - accuracy: 0.5511 - val_loss: 3.6652 - val_accuracy: 0.0723\n",
+      "Epoch 2467/5000\n",
+      "919/919 - 3s - loss: 1.3997 - accuracy: 0.5506 - val_loss: 3.6700 - val_accuracy: 0.0711\n",
+      "Epoch 2468/5000\n",
+      "919/919 - 3s - loss: 1.4048 - accuracy: 0.5497 - val_loss: 3.6756 - val_accuracy: 0.0713\n",
+      "Epoch 2469/5000\n",
+      "919/919 - 3s - loss: 1.4188 - accuracy: 0.5472 - val_loss: 3.6732 - val_accuracy: 0.0722\n",
+      "Epoch 2470/5000\n",
+      "919/919 - 3s - loss: 1.4333 - accuracy: 0.5518 - val_loss: 3.6717 - val_accuracy: 0.0722\n",
+      "Epoch 2471/5000\n",
+      "919/919 - 3s - loss: 1.4177 - accuracy: 0.5490 - val_loss: 3.6795 - val_accuracy: 0.0726\n",
+      "Epoch 2472/5000\n",
+      "919/919 - 3s - loss: 1.4009 - accuracy: 0.5548 - val_loss: 3.6717 - val_accuracy: 0.0723\n",
+      "Epoch 2473/5000\n",
+      "919/919 - 3s - loss: 1.4125 - accuracy: 0.5524 - val_loss: 3.6808 - val_accuracy: 0.0722\n",
+      "Epoch 2474/5000\n",
+      "919/919 - 3s - loss: 1.4048 - accuracy: 0.5486 - val_loss: 3.6820 - val_accuracy: 0.0710\n",
+      "Epoch 2475/5000\n",
+      "919/919 - 3s - loss: 1.4250 - accuracy: 0.5516 - val_loss: 3.6815 - val_accuracy: 0.0703\n",
+      "Epoch 2476/5000\n",
+      "919/919 - 3s - loss: 1.4558 - accuracy: 0.5503 - val_loss: 3.6925 - val_accuracy: 0.0712\n",
+      "Epoch 2477/5000\n",
+      "919/919 - 3s - loss: 1.4176 - accuracy: 0.5537 - val_loss: 3.6891 - val_accuracy: 0.0720\n",
+      "Epoch 2478/5000\n",
+      "919/919 - 3s - loss: 1.4119 - accuracy: 0.5497 - val_loss: 3.6919 - val_accuracy: 0.0705\n",
+      "Epoch 2479/5000\n",
+      "919/919 - 3s - loss: 1.4124 - accuracy: 0.5505 - val_loss: 3.6832 - val_accuracy: 0.0705\n",
+      "Epoch 2480/5000\n",
+      "919/919 - 3s - loss: 1.3947 - accuracy: 0.5543 - val_loss: 3.6920 - val_accuracy: 0.0721\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2481/5000\n",
+      "919/919 - 3s - loss: 1.4427 - accuracy: 0.5555 - val_loss: 3.6875 - val_accuracy: 0.0727\n",
+      "Epoch 2482/5000\n",
+      "919/919 - 3s - loss: 1.4464 - accuracy: 0.5493 - val_loss: 3.6782 - val_accuracy: 0.0728\n",
+      "Epoch 2483/5000\n",
+      "919/919 - 3s - loss: 1.3951 - accuracy: 0.5559 - val_loss: 3.6812 - val_accuracy: 0.0722\n",
+      "Epoch 2484/5000\n",
+      "919/919 - 3s - loss: 1.4123 - accuracy: 0.5511 - val_loss: 3.6816 - val_accuracy: 0.0715\n",
+      "Epoch 2485/5000\n",
+      "919/919 - 3s - loss: 1.4273 - accuracy: 0.5556 - val_loss: 3.6820 - val_accuracy: 0.0727\n",
+      "Epoch 2486/5000\n",
+      "919/919 - 3s - loss: 1.4498 - accuracy: 0.5514 - val_loss: 3.6897 - val_accuracy: 0.0714\n",
+      "Epoch 2487/5000\n",
+      "919/919 - 3s - loss: 1.3998 - accuracy: 0.5506 - val_loss: 3.6910 - val_accuracy: 0.0714\n",
+      "Epoch 2488/5000\n",
+      "919/919 - 3s - loss: 1.4055 - accuracy: 0.5514 - val_loss: 3.6810 - val_accuracy: 0.0723\n",
+      "Epoch 2489/5000\n",
+      "919/919 - 3s - loss: 1.4071 - accuracy: 0.5518 - val_loss: 3.6777 - val_accuracy: 0.0729\n",
+      "Epoch 2490/5000\n",
+      "919/919 - 3s - loss: 1.3926 - accuracy: 0.5530 - val_loss: 3.6821 - val_accuracy: 0.0720\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2491/5000\n",
+      "919/919 - 3s - loss: 1.4473 - accuracy: 0.5539 - val_loss: 3.6761 - val_accuracy: 0.0717\n",
+      "Epoch 2492/5000\n",
+      "919/919 - 3s - loss: 1.3965 - accuracy: 0.5544 - val_loss: 3.6865 - val_accuracy: 0.0718\n",
+      "Epoch 2493/5000\n",
+      "919/919 - 3s - loss: 1.4411 - accuracy: 0.5476 - val_loss: 3.6752 - val_accuracy: 0.0717\n",
+      "Epoch 2494/5000\n",
+      "919/919 - 3s - loss: 1.5097 - accuracy: 0.5543 - val_loss: 3.6744 - val_accuracy: 0.0720\n",
+      "Epoch 2495/5000\n",
+      "919/919 - 3s - loss: 1.3957 - accuracy: 0.5550 - val_loss: 3.6842 - val_accuracy: 0.0714\n",
+      "Epoch 2496/5000\n",
+      "919/919 - 3s - loss: 1.4236 - accuracy: 0.5526 - val_loss: 3.6945 - val_accuracy: 0.0706\n",
+      "Epoch 2497/5000\n",
+      "919/919 - 3s - loss: 1.4035 - accuracy: 0.5562 - val_loss: 3.6917 - val_accuracy: 0.0713\n",
+      "Epoch 2498/5000\n",
+      "919/919 - 3s - loss: 1.4070 - accuracy: 0.5531 - val_loss: 3.6966 - val_accuracy: 0.0731\n",
+      "Epoch 2499/5000\n",
+      "919/919 - 3s - loss: 1.4409 - accuracy: 0.5533 - val_loss: 3.6842 - val_accuracy: 0.0722\n",
+      "Epoch 2500/5000\n",
+      "919/919 - 3s - loss: 1.4261 - accuracy: 0.5525 - val_loss: 3.6840 - val_accuracy: 0.0728\n",
+      "Epoch 2501/5000\n",
+      "919/919 - 3s - loss: 1.4122 - accuracy: 0.5503 - val_loss: 3.6860 - val_accuracy: 0.0731\n",
+      "Epoch 2502/5000\n",
+      "919/919 - 3s - loss: 1.3956 - accuracy: 0.5574 - val_loss: 3.6900 - val_accuracy: 0.0733\n",
+      "Epoch 2503/5000\n",
+      "919/919 - 3s - loss: 1.4062 - accuracy: 0.5508 - val_loss: 3.6761 - val_accuracy: 0.0739\n",
+      "Epoch 2504/5000\n",
+      "919/919 - 3s - loss: 1.3812 - accuracy: 0.5550 - val_loss: 3.6813 - val_accuracy: 0.0731\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2505/5000\n",
+      "919/919 - 3s - loss: 1.3917 - accuracy: 0.5566 - val_loss: 3.6898 - val_accuracy: 0.0731\n",
+      "Epoch 2506/5000\n",
+      "919/919 - 3s - loss: 1.4824 - accuracy: 0.5538 - val_loss: 3.6821 - val_accuracy: 0.0733\n",
+      "Epoch 2507/5000\n",
+      "919/919 - 3s - loss: 1.3958 - accuracy: 0.5533 - val_loss: 3.6994 - val_accuracy: 0.0725\n",
+      "Epoch 2508/5000\n",
+      "919/919 - 3s - loss: 1.4184 - accuracy: 0.5569 - val_loss: 3.7025 - val_accuracy: 0.0716\n",
+      "Epoch 2509/5000\n",
+      "919/919 - 3s - loss: 1.3909 - accuracy: 0.5548 - val_loss: 3.6882 - val_accuracy: 0.0722\n",
+      "Epoch 2510/5000\n",
+      "919/919 - 3s - loss: 1.4094 - accuracy: 0.5545 - val_loss: 3.6773 - val_accuracy: 0.0732\n",
+      "Epoch 2511/5000\n",
+      "919/919 - 3s - loss: 1.4082 - accuracy: 0.5518 - val_loss: 3.6807 - val_accuracy: 0.0733\n",
+      "Epoch 2512/5000\n",
+      "919/919 - 3s - loss: 1.4208 - accuracy: 0.5555 - val_loss: 3.6909 - val_accuracy: 0.0748\n",
+      "Epoch 2513/5000\n",
+      "919/919 - 3s - loss: 1.4070 - accuracy: 0.5511 - val_loss: 3.7010 - val_accuracy: 0.0729\n",
+      "Epoch 2514/5000\n",
+      "919/919 - 3s - loss: 1.4054 - accuracy: 0.5524 - val_loss: 3.6897 - val_accuracy: 0.0730\n",
+      "Epoch 2515/5000\n",
+      "919/919 - 3s - loss: 1.3888 - accuracy: 0.5566 - val_loss: 3.6890 - val_accuracy: 0.0731\n",
+      "Epoch 2516/5000\n",
+      "919/919 - 3s - loss: 1.4070 - accuracy: 0.5533 - val_loss: 3.6864 - val_accuracy: 0.0711\n",
+      "Epoch 2517/5000\n",
+      "919/919 - 3s - loss: 1.3990 - accuracy: 0.5541 - val_loss: 3.6959 - val_accuracy: 0.0707\n",
+      "Epoch 2518/5000\n",
+      "919/919 - 3s - loss: 1.3912 - accuracy: 0.5541 - val_loss: 3.6823 - val_accuracy: 0.0706\n",
+      "Epoch 2519/5000\n",
+      "919/919 - 3s - loss: 1.4080 - accuracy: 0.5537 - val_loss: 3.6740 - val_accuracy: 0.0712\n",
+      "Epoch 2520/5000\n",
+      "919/919 - 3s - loss: 1.4002 - accuracy: 0.5549 - val_loss: 3.6835 - val_accuracy: 0.0710\n",
+      "Epoch 2521/5000\n",
+      "919/919 - 3s - loss: 1.3994 - accuracy: 0.5514 - val_loss: 3.6778 - val_accuracy: 0.0710\n",
+      "Epoch 2522/5000\n",
+      "919/919 - 3s - loss: 1.3883 - accuracy: 0.5533 - val_loss: 3.6788 - val_accuracy: 0.0718\n",
+      "Epoch 2523/5000\n",
+      "919/919 - 3s - loss: 1.3995 - accuracy: 0.5545 - val_loss: 3.6832 - val_accuracy: 0.0711\n",
+      "Epoch 2524/5000\n",
+      "919/919 - 3s - loss: 1.5126 - accuracy: 0.5561 - val_loss: 3.6909 - val_accuracy: 0.0705\n",
+      "Epoch 2525/5000\n",
+      "919/919 - 3s - loss: 1.3882 - accuracy: 0.5598 - val_loss: 3.6891 - val_accuracy: 0.0713\n",
+      "Epoch 2526/5000\n",
+      "919/919 - 3s - loss: 1.4468 - accuracy: 0.5512 - val_loss: 3.6818 - val_accuracy: 0.0732\n",
+      "Epoch 2527/5000\n",
+      "919/919 - 3s - loss: 1.4131 - accuracy: 0.5561 - val_loss: 3.6842 - val_accuracy: 0.0713\n",
+      "Epoch 2528/5000\n",
+      "919/919 - 3s - loss: 1.4012 - accuracy: 0.5552 - val_loss: 3.6799 - val_accuracy: 0.0729\n",
+      "Epoch 2529/5000\n",
+      "919/919 - 3s - loss: 1.4009 - accuracy: 0.5516 - val_loss: 3.6762 - val_accuracy: 0.0734\n",
+      "Epoch 2530/5000\n",
+      "919/919 - 3s - loss: 1.3899 - accuracy: 0.5550 - val_loss: 3.6853 - val_accuracy: 0.0731\n",
+      "Epoch 2531/5000\n",
+      "919/919 - 3s - loss: 1.3991 - accuracy: 0.5565 - val_loss: 3.6800 - val_accuracy: 0.0740\n",
+      "Epoch 2532/5000\n",
+      "919/919 - 3s - loss: 1.4007 - accuracy: 0.5542 - val_loss: 3.6888 - val_accuracy: 0.0736\n",
+      "Epoch 2533/5000\n",
+      "919/919 - 3s - loss: 1.3896 - accuracy: 0.5571 - val_loss: 3.6895 - val_accuracy: 0.0739\n",
+      "Epoch 2534/5000\n",
+      "919/919 - 3s - loss: 1.3941 - accuracy: 0.5580 - val_loss: 3.6856 - val_accuracy: 0.0728\n",
+      "Epoch 2535/5000\n",
+      "919/919 - 3s - loss: 1.4201 - accuracy: 0.5523 - val_loss: 3.6832 - val_accuracy: 0.0721\n",
+      "Epoch 2536/5000\n",
+      "919/919 - 3s - loss: 1.4019 - accuracy: 0.5524 - val_loss: 3.6788 - val_accuracy: 0.0732\n",
+      "Epoch 2537/5000\n",
+      "919/919 - 3s - loss: 1.3912 - accuracy: 0.5546 - val_loss: 3.6885 - val_accuracy: 0.0745\n",
+      "Epoch 2538/5000\n",
+      "919/919 - 3s - loss: 1.4683 - accuracy: 0.5531 - val_loss: 3.6820 - val_accuracy: 0.0747\n",
+      "Epoch 2539/5000\n",
+      "919/919 - 3s - loss: 1.3871 - accuracy: 0.5587 - val_loss: 3.6948 - val_accuracy: 0.0734\n",
+      "Epoch 2540/5000\n",
+      "919/919 - 3s - loss: 1.4095 - accuracy: 0.5550 - val_loss: 3.6885 - val_accuracy: 0.0758\n",
+      "Epoch 2541/5000\n",
+      "919/919 - 3s - loss: 1.4507 - accuracy: 0.5530 - val_loss: 3.6897 - val_accuracy: 0.0733\n",
+      "Epoch 2542/5000\n",
+      "919/919 - 3s - loss: 1.4042 - accuracy: 0.5567 - val_loss: 3.6882 - val_accuracy: 0.0733\n",
+      "Epoch 2543/5000\n",
+      "919/919 - 3s - loss: 1.4067 - accuracy: 0.5504 - val_loss: 3.6841 - val_accuracy: 0.0729\n",
+      "Epoch 2544/5000\n",
+      "919/919 - 3s - loss: 1.4013 - accuracy: 0.5554 - val_loss: 3.6964 - val_accuracy: 0.0730\n",
+      "Epoch 2545/5000\n",
+      "919/919 - 3s - loss: 1.3953 - accuracy: 0.5579 - val_loss: 3.6883 - val_accuracy: 0.0714\n",
+      "Epoch 2546/5000\n",
+      "919/919 - 3s - loss: 1.4048 - accuracy: 0.5543 - val_loss: 3.6825 - val_accuracy: 0.0733\n",
+      "Epoch 2547/5000\n",
+      "919/919 - 3s - loss: 1.3977 - accuracy: 0.5547 - val_loss: 3.6831 - val_accuracy: 0.0718\n",
+      "Epoch 2548/5000\n",
+      "919/919 - 3s - loss: 1.3973 - accuracy: 0.5552 - val_loss: 3.6862 - val_accuracy: 0.0731\n",
+      "Epoch 2549/5000\n",
+      "919/919 - 3s - loss: 1.3907 - accuracy: 0.5551 - val_loss: 3.7004 - val_accuracy: 0.0712\n",
+      "Epoch 2550/5000\n",
+      "919/919 - 3s - loss: 1.3872 - accuracy: 0.5555 - val_loss: 3.6940 - val_accuracy: 0.0711\n",
+      "Epoch 2551/5000\n",
+      "919/919 - 3s - loss: 1.3909 - accuracy: 0.5542 - val_loss: 3.6862 - val_accuracy: 0.0722\n",
+      "Epoch 2552/5000\n",
+      "919/919 - 3s - loss: 1.4007 - accuracy: 0.5578 - val_loss: 3.6867 - val_accuracy: 0.0732\n",
+      "Epoch 2553/5000\n",
+      "919/919 - 3s - loss: 1.3906 - accuracy: 0.5576 - val_loss: 3.6976 - val_accuracy: 0.0730\n",
+      "Epoch 2554/5000\n",
+      "919/919 - 3s - loss: 1.4002 - accuracy: 0.5552 - val_loss: 3.6969 - val_accuracy: 0.0737\n",
+      "Epoch 2555/5000\n",
+      "919/919 - 3s - loss: 1.4040 - accuracy: 0.5563 - val_loss: 3.6967 - val_accuracy: 0.0717\n",
+      "Epoch 2556/5000\n",
+      "919/919 - 3s - loss: 1.3894 - accuracy: 0.5603 - val_loss: 3.7026 - val_accuracy: 0.0710\n",
+      "Epoch 2557/5000\n",
+      "919/919 - 3s - loss: 1.3945 - accuracy: 0.5554 - val_loss: 3.6920 - val_accuracy: 0.0729\n",
+      "Epoch 2558/5000\n",
+      "919/919 - 3s - loss: 1.3936 - accuracy: 0.5618 - val_loss: 3.6872 - val_accuracy: 0.0744\n",
+      "Epoch 2559/5000\n",
+      "919/919 - 3s - loss: 1.3898 - accuracy: 0.5588 - val_loss: 3.6940 - val_accuracy: 0.0768\n",
+      "Epoch 2560/5000\n",
+      "919/919 - 3s - loss: 1.4010 - accuracy: 0.5544 - val_loss: 3.7142 - val_accuracy: 0.0749\n",
+      "Epoch 2561/5000\n",
+      "919/919 - 3s - loss: 1.3914 - accuracy: 0.5588 - val_loss: 3.7124 - val_accuracy: 0.0740\n",
+      "Epoch 2562/5000\n",
+      "919/919 - 3s - loss: 1.4172 - accuracy: 0.5553 - val_loss: 3.7202 - val_accuracy: 0.0736\n",
+      "Epoch 2563/5000\n",
+      "919/919 - 3s - loss: 1.3877 - accuracy: 0.5586 - val_loss: 3.7073 - val_accuracy: 0.0738\n",
+      "Epoch 2564/5000\n",
+      "919/919 - 3s - loss: 1.4004 - accuracy: 0.5531 - val_loss: 3.7172 - val_accuracy: 0.0741\n",
+      "Epoch 2565/5000\n",
+      "919/919 - 3s - loss: 1.3864 - accuracy: 0.5575 - val_loss: 3.7141 - val_accuracy: 0.0721\n",
+      "Epoch 2566/5000\n",
+      "919/919 - 3s - loss: 1.3913 - accuracy: 0.5559 - val_loss: 3.7098 - val_accuracy: 0.0725\n",
+      "Epoch 2567/5000\n",
+      "919/919 - 3s - loss: 1.3880 - accuracy: 0.5560 - val_loss: 3.7073 - val_accuracy: 0.0722\n",
+      "Epoch 2568/5000\n",
+      "919/919 - 3s - loss: 1.3839 - accuracy: 0.5597 - val_loss: 3.7067 - val_accuracy: 0.0723\n",
+      "Epoch 2569/5000\n",
+      "919/919 - 3s - loss: 1.4942 - accuracy: 0.5593 - val_loss: 3.7053 - val_accuracy: 0.0724\n",
+      "Epoch 2570/5000\n",
+      "919/919 - 3s - loss: 1.3869 - accuracy: 0.5527 - val_loss: 3.7027 - val_accuracy: 0.0723\n",
+      "Epoch 2571/5000\n",
+      "919/919 - 3s - loss: 1.4107 - accuracy: 0.5573 - val_loss: 3.7134 - val_accuracy: 0.0716\n",
+      "Epoch 2572/5000\n",
+      "919/919 - 3s - loss: 1.3810 - accuracy: 0.5590 - val_loss: 3.7096 - val_accuracy: 0.0731\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2573/5000\n",
+      "919/919 - 3s - loss: 1.3997 - accuracy: 0.5569 - val_loss: 3.7220 - val_accuracy: 0.0724\n",
+      "Epoch 2574/5000\n",
+      "919/919 - 3s - loss: 1.3950 - accuracy: 0.5594 - val_loss: 3.7083 - val_accuracy: 0.0732\n",
+      "Epoch 2575/5000\n",
+      "919/919 - 3s - loss: 1.3860 - accuracy: 0.5593 - val_loss: 3.7147 - val_accuracy: 0.0720\n",
+      "Epoch 2576/5000\n",
+      "919/919 - 3s - loss: 1.3819 - accuracy: 0.5567 - val_loss: 3.7117 - val_accuracy: 0.0718\n",
+      "Epoch 2577/5000\n",
+      "919/919 - 3s - loss: 1.3858 - accuracy: 0.5582 - val_loss: 3.7200 - val_accuracy: 0.0730\n",
+      "Epoch 2578/5000\n",
+      "919/919 - 3s - loss: 1.3877 - accuracy: 0.5539 - val_loss: 3.7190 - val_accuracy: 0.0717\n",
+      "Epoch 2579/5000\n",
+      "919/919 - 3s - loss: 1.3868 - accuracy: 0.5554 - val_loss: 3.7121 - val_accuracy: 0.0729\n",
+      "Epoch 2580/5000\n",
+      "919/919 - 3s - loss: 1.3752 - accuracy: 0.5616 - val_loss: 3.7124 - val_accuracy: 0.0722\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2581/5000\n",
+      "919/919 - 3s - loss: 1.3843 - accuracy: 0.5561 - val_loss: 3.7128 - val_accuracy: 0.0727\n",
+      "Epoch 2582/5000\n",
+      "919/919 - 3s - loss: 1.3900 - accuracy: 0.5638 - val_loss: 3.7197 - val_accuracy: 0.0723\n",
+      "Epoch 2583/5000\n",
+      "919/919 - 3s - loss: 1.3965 - accuracy: 0.5588 - val_loss: 3.7134 - val_accuracy: 0.0725\n",
+      "Epoch 2584/5000\n",
+      "919/919 - 3s - loss: 1.3911 - accuracy: 0.5547 - val_loss: 3.6986 - val_accuracy: 0.0730\n",
+      "Epoch 2585/5000\n",
+      "919/919 - 3s - loss: 1.3824 - accuracy: 0.5590 - val_loss: 3.6880 - val_accuracy: 0.0726\n",
+      "Epoch 2586/5000\n",
+      "919/919 - 3s - loss: 1.4162 - accuracy: 0.5610 - val_loss: 3.7051 - val_accuracy: 0.0738\n",
+      "Epoch 2587/5000\n",
+      "919/919 - 3s - loss: 1.3986 - accuracy: 0.5599 - val_loss: 3.6969 - val_accuracy: 0.0765\n",
+      "Epoch 2588/5000\n",
+      "919/919 - 3s - loss: 1.3985 - accuracy: 0.5546 - val_loss: 3.7054 - val_accuracy: 0.0722\n",
+      "Epoch 2589/5000\n",
+      "919/919 - 3s - loss: 1.3813 - accuracy: 0.5584 - val_loss: 3.7038 - val_accuracy: 0.0729\n",
+      "Epoch 2590/5000\n",
+      "919/919 - 3s - loss: 1.3864 - accuracy: 0.5577 - val_loss: 3.7040 - val_accuracy: 0.0724\n",
+      "Epoch 2591/5000\n",
+      "919/919 - 3s - loss: 1.3968 - accuracy: 0.5561 - val_loss: 3.7261 - val_accuracy: 0.0716\n",
+      "Epoch 2592/5000\n",
+      "919/919 - 3s - loss: 1.3906 - accuracy: 0.5595 - val_loss: 3.7143 - val_accuracy: 0.0725\n",
+      "Epoch 2593/5000\n",
+      "919/919 - 3s - loss: 1.3933 - accuracy: 0.5575 - val_loss: 3.7077 - val_accuracy: 0.0722\n",
+      "Epoch 2594/5000\n",
+      "919/919 - 3s - loss: 1.5072 - accuracy: 0.5575 - val_loss: 3.7097 - val_accuracy: 0.0724\n",
+      "Epoch 2595/5000\n",
+      "919/919 - 3s - loss: 1.3846 - accuracy: 0.5571 - val_loss: 3.7083 - val_accuracy: 0.0722\n",
+      "Epoch 2596/5000\n",
+      "919/919 - 3s - loss: 1.3974 - accuracy: 0.5599 - val_loss: 3.7152 - val_accuracy: 0.0724\n",
+      "Epoch 2597/5000\n",
+      "919/919 - 3s - loss: 1.3892 - accuracy: 0.5572 - val_loss: 3.7095 - val_accuracy: 0.0720\n",
+      "Epoch 2598/5000\n",
+      "919/919 - 3s - loss: 1.4006 - accuracy: 0.5596 - val_loss: 3.7119 - val_accuracy: 0.0722\n",
+      "Epoch 2599/5000\n",
+      "919/919 - 3s - loss: 1.3846 - accuracy: 0.5596 - val_loss: 3.7165 - val_accuracy: 0.0722\n",
+      "Epoch 2600/5000\n",
+      "919/919 - 3s - loss: 1.3818 - accuracy: 0.5594 - val_loss: 3.7243 - val_accuracy: 0.0727\n",
+      "Epoch 2601/5000\n",
+      "919/919 - 3s - loss: 1.4452 - accuracy: 0.5605 - val_loss: 3.7257 - val_accuracy: 0.0731\n",
+      "Epoch 2602/5000\n",
+      "919/919 - 3s - loss: 1.3894 - accuracy: 0.5598 - val_loss: 3.7240 - val_accuracy: 0.0731\n",
+      "Epoch 2603/5000\n",
+      "919/919 - 3s - loss: 1.3812 - accuracy: 0.5623 - val_loss: 3.7100 - val_accuracy: 0.0730\n",
+      "Epoch 2604/5000\n",
+      "919/919 - 3s - loss: 1.3722 - accuracy: 0.5610 - val_loss: 3.7126 - val_accuracy: 0.0731\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2605/5000\n",
+      "919/919 - 3s - loss: 1.3919 - accuracy: 0.5592 - val_loss: 3.7075 - val_accuracy: 0.0726\n",
+      "Epoch 2606/5000\n",
+      "919/919 - 3s - loss: 1.3826 - accuracy: 0.5626 - val_loss: 3.7119 - val_accuracy: 0.0723\n",
+      "Epoch 2607/5000\n",
+      "919/919 - 3s - loss: 1.3825 - accuracy: 0.5539 - val_loss: 3.7250 - val_accuracy: 0.0722\n",
+      "Epoch 2608/5000\n",
+      "919/919 - 3s - loss: 1.3770 - accuracy: 0.5623 - val_loss: 3.7240 - val_accuracy: 0.0727\n",
+      "Epoch 2609/5000\n",
+      "919/919 - 3s - loss: 1.4502 - accuracy: 0.5622 - val_loss: 3.7182 - val_accuracy: 0.0731\n",
+      "Epoch 2610/5000\n",
+      "919/919 - 3s - loss: 1.3688 - accuracy: 0.5635 - val_loss: 3.7177 - val_accuracy: 0.0731\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2611/5000\n",
+      "919/919 - 3s - loss: 1.4023 - accuracy: 0.5561 - val_loss: 3.7178 - val_accuracy: 0.0737\n",
+      "Epoch 2612/5000\n",
+      "919/919 - 3s - loss: 1.3805 - accuracy: 0.5569 - val_loss: 3.7115 - val_accuracy: 0.0731\n",
+      "Epoch 2613/5000\n",
+      "919/919 - 3s - loss: 1.3855 - accuracy: 0.5599 - val_loss: 3.7158 - val_accuracy: 0.0728\n",
+      "Epoch 2614/5000\n",
+      "919/919 - 3s - loss: 1.3788 - accuracy: 0.5607 - val_loss: 3.7147 - val_accuracy: 0.0725\n",
+      "Epoch 2615/5000\n",
+      "919/919 - 3s - loss: 1.3862 - accuracy: 0.5620 - val_loss: 3.7129 - val_accuracy: 0.0724\n",
+      "Epoch 2616/5000\n",
+      "919/919 - 3s - loss: 1.3802 - accuracy: 0.5610 - val_loss: 3.7233 - val_accuracy: 0.0727\n",
+      "Epoch 2617/5000\n",
+      "919/919 - 3s - loss: 1.4046 - accuracy: 0.5599 - val_loss: 3.7310 - val_accuracy: 0.0727\n",
+      "Epoch 2618/5000\n",
+      "919/919 - 3s - loss: 1.3841 - accuracy: 0.5633 - val_loss: 3.7116 - val_accuracy: 0.0747\n",
+      "Epoch 2619/5000\n",
+      "919/919 - 3s - loss: 1.3839 - accuracy: 0.5606 - val_loss: 3.7194 - val_accuracy: 0.0736\n",
+      "Epoch 2620/5000\n",
+      "919/919 - 3s - loss: 1.3871 - accuracy: 0.5654 - val_loss: 3.7166 - val_accuracy: 0.0729\n",
+      "Epoch 2621/5000\n",
+      "919/919 - 3s - loss: 1.3748 - accuracy: 0.5644 - val_loss: 3.7228 - val_accuracy: 0.0728\n",
+      "Epoch 2622/5000\n",
+      "919/919 - 3s - loss: 1.5810 - accuracy: 0.5575 - val_loss: 3.7212 - val_accuracy: 0.0730\n",
+      "Epoch 2623/5000\n",
+      "919/919 - 3s - loss: 1.3866 - accuracy: 0.5621 - val_loss: 3.7203 - val_accuracy: 0.0734\n",
+      "Epoch 2624/5000\n",
+      "919/919 - 3s - loss: 1.3955 - accuracy: 0.5601 - val_loss: 3.7173 - val_accuracy: 0.0731\n",
+      "Epoch 2625/5000\n",
+      "919/919 - 3s - loss: 1.3844 - accuracy: 0.5603 - val_loss: 3.7206 - val_accuracy: 0.0725\n",
+      "Epoch 2626/5000\n",
+      "919/919 - 3s - loss: 1.3845 - accuracy: 0.5610 - val_loss: 3.7261 - val_accuracy: 0.0726\n",
+      "Epoch 2627/5000\n",
+      "919/919 - 3s - loss: 1.3980 - accuracy: 0.5572 - val_loss: 3.7195 - val_accuracy: 0.0725\n",
+      "Epoch 2628/5000\n",
+      "919/919 - 3s - loss: 1.4106 - accuracy: 0.5583 - val_loss: 3.7100 - val_accuracy: 0.0724\n",
+      "Epoch 2629/5000\n",
+      "919/919 - 3s - loss: 1.3675 - accuracy: 0.5656 - val_loss: 3.7136 - val_accuracy: 0.0731\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2630/5000\n",
+      "919/919 - 3s - loss: 1.3682 - accuracy: 0.5633 - val_loss: 3.7235 - val_accuracy: 0.0734\n",
+      "Epoch 2631/5000\n",
+      "919/919 - 3s - loss: 1.3750 - accuracy: 0.5659 - val_loss: 3.7320 - val_accuracy: 0.0738\n",
+      "Epoch 2632/5000\n",
+      "919/919 - 3s - loss: 1.3741 - accuracy: 0.5656 - val_loss: 3.7438 - val_accuracy: 0.0728\n",
+      "Epoch 2633/5000\n",
+      "919/919 - 3s - loss: 1.3753 - accuracy: 0.5663 - val_loss: 3.7320 - val_accuracy: 0.0728\n",
+      "Epoch 2634/5000\n",
+      "919/919 - 3s - loss: 1.3763 - accuracy: 0.5635 - val_loss: 3.7344 - val_accuracy: 0.0726\n",
+      "Epoch 2635/5000\n",
+      "919/919 - 3s - loss: 1.3850 - accuracy: 0.5618 - val_loss: 3.7370 - val_accuracy: 0.0727\n",
+      "Epoch 2636/5000\n",
+      "919/919 - 3s - loss: 1.3810 - accuracy: 0.5607 - val_loss: 3.7293 - val_accuracy: 0.0730\n",
+      "Epoch 2637/5000\n",
+      "919/919 - 3s - loss: 1.3764 - accuracy: 0.5632 - val_loss: 3.7269 - val_accuracy: 0.0732\n",
+      "Epoch 2638/5000\n",
+      "919/919 - 3s - loss: 1.3958 - accuracy: 0.5613 - val_loss: 3.7243 - val_accuracy: 0.0731\n",
+      "Epoch 2639/5000\n",
+      "919/919 - 3s - loss: 1.3999 - accuracy: 0.5564 - val_loss: 3.7277 - val_accuracy: 0.0731\n",
+      "Epoch 2640/5000\n",
+      "919/919 - 3s - loss: 1.3742 - accuracy: 0.5620 - val_loss: 3.7082 - val_accuracy: 0.0745\n",
+      "Epoch 2641/5000\n",
+      "919/919 - 3s - loss: 1.3990 - accuracy: 0.5601 - val_loss: 3.7116 - val_accuracy: 0.0731\n",
+      "Epoch 2642/5000\n",
+      "919/919 - 3s - loss: 1.3822 - accuracy: 0.5592 - val_loss: 3.7154 - val_accuracy: 0.0731\n",
+      "Epoch 2643/5000\n",
+      "919/919 - 3s - loss: 1.3808 - accuracy: 0.5633 - val_loss: 3.7125 - val_accuracy: 0.0729\n",
+      "Epoch 2644/5000\n",
+      "919/919 - 3s - loss: 1.3718 - accuracy: 0.5599 - val_loss: 3.7184 - val_accuracy: 0.0731\n",
+      "Epoch 2645/5000\n",
+      "919/919 - 3s - loss: 1.4020 - accuracy: 0.5622 - val_loss: 3.7195 - val_accuracy: 0.0736\n",
+      "Epoch 2646/5000\n",
+      "919/919 - 3s - loss: 1.3728 - accuracy: 0.5641 - val_loss: 3.7257 - val_accuracy: 0.0734\n",
+      "Epoch 2647/5000\n",
+      "919/919 - 3s - loss: 1.3870 - accuracy: 0.5643 - val_loss: 3.7168 - val_accuracy: 0.0731\n",
+      "Epoch 2648/5000\n",
+      "919/919 - 3s - loss: 1.3724 - accuracy: 0.5636 - val_loss: 3.7185 - val_accuracy: 0.0732\n",
+      "Epoch 2649/5000\n",
+      "919/919 - 3s - loss: 1.3678 - accuracy: 0.5629 - val_loss: 3.7304 - val_accuracy: 0.0734\n",
+      "Epoch 2650/5000\n",
+      "919/919 - 3s - loss: 1.3721 - accuracy: 0.5612 - val_loss: 3.7366 - val_accuracy: 0.0757\n",
+      "Epoch 2651/5000\n",
+      "919/919 - 3s - loss: 1.3836 - accuracy: 0.5603 - val_loss: 3.7289 - val_accuracy: 0.0736\n",
+      "Epoch 2652/5000\n",
+      "919/919 - 3s - loss: 1.3872 - accuracy: 0.5639 - val_loss: 3.7142 - val_accuracy: 0.0740\n",
+      "Epoch 2653/5000\n",
+      "919/919 - 3s - loss: 1.5130 - accuracy: 0.5641 - val_loss: 3.7148 - val_accuracy: 0.0729\n",
+      "Epoch 2654/5000\n",
+      "919/919 - 3s - loss: 1.3804 - accuracy: 0.5613 - val_loss: 3.7282 - val_accuracy: 0.0726\n",
+      "Epoch 2655/5000\n",
+      "919/919 - 3s - loss: 1.3769 - accuracy: 0.5576 - val_loss: 3.7311 - val_accuracy: 0.0729\n",
+      "Epoch 2656/5000\n",
+      "919/919 - 3s - loss: 1.3972 - accuracy: 0.5626 - val_loss: 3.7315 - val_accuracy: 0.0734\n",
+      "Epoch 2657/5000\n",
+      "919/919 - 3s - loss: 1.3799 - accuracy: 0.5622 - val_loss: 3.7215 - val_accuracy: 0.0758\n",
+      "Epoch 2658/5000\n",
+      "919/919 - 3s - loss: 1.4490 - accuracy: 0.5610 - val_loss: 3.7289 - val_accuracy: 0.0745\n",
+      "Epoch 2659/5000\n",
+      "919/919 - 3s - loss: 1.4238 - accuracy: 0.5644 - val_loss: 3.7338 - val_accuracy: 0.0741\n",
+      "Epoch 2660/5000\n",
+      "919/919 - 3s - loss: 1.3850 - accuracy: 0.5623 - val_loss: 3.7309 - val_accuracy: 0.0743\n",
+      "Epoch 2661/5000\n",
+      "919/919 - 3s - loss: 1.3814 - accuracy: 0.5654 - val_loss: 3.7371 - val_accuracy: 0.0737\n",
+      "Epoch 2662/5000\n",
+      "919/919 - 3s - loss: 1.3731 - accuracy: 0.5678 - val_loss: 3.7315 - val_accuracy: 0.0740\n",
+      "Epoch 2663/5000\n",
+      "919/919 - 3s - loss: 1.3746 - accuracy: 0.5631 - val_loss: 3.7223 - val_accuracy: 0.0735\n",
+      "Epoch 2664/5000\n",
+      "919/919 - 3s - loss: 1.3698 - accuracy: 0.5643 - val_loss: 3.7135 - val_accuracy: 0.0737\n",
+      "Epoch 2665/5000\n",
+      "919/919 - 3s - loss: 1.3604 - accuracy: 0.5670 - val_loss: 3.7176 - val_accuracy: 0.0740\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2666/5000\n",
+      "919/919 - 3s - loss: 1.4188 - accuracy: 0.5646 - val_loss: 3.7144 - val_accuracy: 0.0740\n",
+      "Epoch 2667/5000\n",
+      "919/919 - 3s - loss: 1.3957 - accuracy: 0.5693 - val_loss: 3.7252 - val_accuracy: 0.0737\n",
+      "Epoch 2668/5000\n",
+      "919/919 - 3s - loss: 1.3613 - accuracy: 0.5690 - val_loss: 3.7252 - val_accuracy: 0.0745\n",
+      "Epoch 2669/5000\n",
+      "919/919 - 3s - loss: 1.3717 - accuracy: 0.5668 - val_loss: 3.7238 - val_accuracy: 0.0743\n",
+      "Epoch 2670/5000\n",
+      "919/919 - 3s - loss: 1.3900 - accuracy: 0.5566 - val_loss: 3.7248 - val_accuracy: 0.0737\n",
+      "Epoch 2671/5000\n",
+      "919/919 - 3s - loss: 1.3828 - accuracy: 0.5611 - val_loss: 3.7262 - val_accuracy: 0.0756\n",
+      "Epoch 2672/5000\n",
+      "919/919 - 3s - loss: 1.3795 - accuracy: 0.5588 - val_loss: 3.7271 - val_accuracy: 0.0749\n",
+      "Epoch 2673/5000\n",
+      "919/919 - 3s - loss: 1.3802 - accuracy: 0.5623 - val_loss: 3.7373 - val_accuracy: 0.0732\n",
+      "Epoch 2674/5000\n",
+      "919/919 - 3s - loss: 1.3700 - accuracy: 0.5622 - val_loss: 3.7338 - val_accuracy: 0.0730\n",
+      "Epoch 2675/5000\n",
+      "919/919 - 3s - loss: 1.3536 - accuracy: 0.5694 - val_loss: 3.7412 - val_accuracy: 0.0733\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2676/5000\n",
+      "919/919 - 3s - loss: 1.4874 - accuracy: 0.5648 - val_loss: 3.7387 - val_accuracy: 0.0733\n",
+      "Epoch 2677/5000\n",
+      "919/919 - 3s - loss: 1.3725 - accuracy: 0.5609 - val_loss: 3.7430 - val_accuracy: 0.0734\n",
+      "Epoch 2678/5000\n",
+      "919/919 - 3s - loss: 1.3966 - accuracy: 0.5616 - val_loss: 3.7417 - val_accuracy: 0.0737\n",
+      "Epoch 2679/5000\n",
+      "919/919 - 3s - loss: 1.3709 - accuracy: 0.5646 - val_loss: 3.7414 - val_accuracy: 0.0738\n",
+      "Epoch 2680/5000\n",
+      "919/919 - 3s - loss: 1.3719 - accuracy: 0.5638 - val_loss: 3.7367 - val_accuracy: 0.0740\n",
+      "Epoch 2681/5000\n",
+      "919/919 - 3s - loss: 1.4415 - accuracy: 0.5619 - val_loss: 3.7493 - val_accuracy: 0.0736\n",
+      "Epoch 2682/5000\n",
+      "919/919 - 3s - loss: 1.3821 - accuracy: 0.5601 - val_loss: 3.7502 - val_accuracy: 0.0732\n",
+      "Epoch 2683/5000\n",
+      "919/919 - 3s - loss: 1.3879 - accuracy: 0.5640 - val_loss: 3.7473 - val_accuracy: 0.0736\n",
+      "Epoch 2684/5000\n",
+      "919/919 - 3s - loss: 1.4244 - accuracy: 0.5688 - val_loss: 3.7516 - val_accuracy: 0.0733\n",
+      "Epoch 2685/5000\n",
+      "919/919 - 3s - loss: 1.3630 - accuracy: 0.5673 - val_loss: 3.7374 - val_accuracy: 0.0755\n",
+      "Epoch 2686/5000\n",
+      "919/919 - 3s - loss: 1.4222 - accuracy: 0.5669 - val_loss: 3.7382 - val_accuracy: 0.0759\n",
+      "Epoch 2687/5000\n",
+      "919/919 - 3s - loss: 1.3706 - accuracy: 0.5648 - val_loss: 3.7457 - val_accuracy: 0.0759\n",
+      "Epoch 2688/5000\n",
+      "919/919 - 3s - loss: 1.3813 - accuracy: 0.5640 - val_loss: 3.7440 - val_accuracy: 0.0736\n",
+      "Epoch 2689/5000\n",
+      "919/919 - 3s - loss: 1.3738 - accuracy: 0.5592 - val_loss: 3.7450 - val_accuracy: 0.0731\n",
+      "Epoch 2690/5000\n",
+      "919/919 - 3s - loss: 1.4200 - accuracy: 0.5635 - val_loss: 3.7463 - val_accuracy: 0.0736\n",
+      "Epoch 2691/5000\n",
+      "919/919 - 3s - loss: 1.3750 - accuracy: 0.5649 - val_loss: 3.7498 - val_accuracy: 0.0729\n",
+      "Epoch 2692/5000\n",
+      "919/919 - 3s - loss: 1.3658 - accuracy: 0.5655 - val_loss: 3.7524 - val_accuracy: 0.0729\n",
+      "Epoch 2693/5000\n",
+      "919/919 - 3s - loss: 1.4029 - accuracy: 0.5636 - val_loss: 3.7532 - val_accuracy: 0.0730\n",
+      "Epoch 2694/5000\n",
+      "919/919 - 3s - loss: 1.3872 - accuracy: 0.5635 - val_loss: 3.7416 - val_accuracy: 0.0736\n",
+      "Epoch 2695/5000\n",
+      "919/919 - 3s - loss: 1.3791 - accuracy: 0.5635 - val_loss: 3.7314 - val_accuracy: 0.0736\n",
+      "Epoch 2696/5000\n",
+      "919/919 - 3s - loss: 1.3730 - accuracy: 0.5629 - val_loss: 3.7291 - val_accuracy: 0.0730\n",
+      "Epoch 2697/5000\n",
+      "919/919 - 3s - loss: 1.3791 - accuracy: 0.5672 - val_loss: 3.7324 - val_accuracy: 0.0737\n",
+      "Epoch 2698/5000\n",
+      "919/919 - 3s - loss: 1.3736 - accuracy: 0.5668 - val_loss: 3.7428 - val_accuracy: 0.0736\n",
+      "Epoch 2699/5000\n",
+      "919/919 - 3s - loss: 1.3573 - accuracy: 0.5645 - val_loss: 3.7359 - val_accuracy: 0.0740\n",
+      "Epoch 2700/5000\n",
+      "919/919 - 3s - loss: 1.3820 - accuracy: 0.5627 - val_loss: 3.7475 - val_accuracy: 0.0738\n",
+      "Epoch 2701/5000\n",
+      "919/919 - 3s - loss: 1.3589 - accuracy: 0.5676 - val_loss: 3.7606 - val_accuracy: 0.0736\n",
+      "Epoch 2702/5000\n",
+      "919/919 - 3s - loss: 1.3611 - accuracy: 0.5634 - val_loss: 3.7521 - val_accuracy: 0.0739\n",
+      "Epoch 2703/5000\n",
+      "919/919 - 3s - loss: 1.3500 - accuracy: 0.5673 - val_loss: 3.7604 - val_accuracy: 0.0740\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2704/5000\n",
+      "919/919 - 3s - loss: 1.3782 - accuracy: 0.5661 - val_loss: 3.7531 - val_accuracy: 0.0735\n",
+      "Epoch 2705/5000\n",
+      "919/919 - 3s - loss: 1.3675 - accuracy: 0.5662 - val_loss: 3.7468 - val_accuracy: 0.0735\n",
+      "Epoch 2706/5000\n",
+      "919/919 - 3s - loss: 1.3673 - accuracy: 0.5637 - val_loss: 3.7446 - val_accuracy: 0.0738\n",
+      "Epoch 2707/5000\n",
+      "919/919 - 3s - loss: 1.3691 - accuracy: 0.5659 - val_loss: 3.7479 - val_accuracy: 0.0739\n",
+      "Epoch 2708/5000\n",
+      "919/919 - 3s - loss: 1.3684 - accuracy: 0.5637 - val_loss: 3.7497 - val_accuracy: 0.0742\n",
+      "Epoch 2709/5000\n",
+      "919/919 - 3s - loss: 1.3677 - accuracy: 0.5608 - val_loss: 3.7420 - val_accuracy: 0.0744\n",
+      "Epoch 2710/5000\n",
+      "919/919 - 3s - loss: 1.3774 - accuracy: 0.5618 - val_loss: 3.7333 - val_accuracy: 0.0745\n",
+      "Epoch 2711/5000\n",
+      "919/919 - 3s - loss: 1.3620 - accuracy: 0.5647 - val_loss: 3.7393 - val_accuracy: 0.0744\n",
+      "Epoch 2712/5000\n",
+      "919/919 - 3s - loss: 1.3895 - accuracy: 0.5652 - val_loss: 3.7433 - val_accuracy: 0.0747\n",
+      "Epoch 2713/5000\n",
+      "919/919 - 3s - loss: 1.3641 - accuracy: 0.5678 - val_loss: 3.7529 - val_accuracy: 0.0745\n",
+      "Epoch 2714/5000\n",
+      "919/919 - 3s - loss: 1.3712 - accuracy: 0.5670 - val_loss: 3.7419 - val_accuracy: 0.0742\n",
+      "Epoch 2715/5000\n",
+      "919/919 - 3s - loss: 1.3914 - accuracy: 0.5629 - val_loss: 3.7506 - val_accuracy: 0.0735\n",
+      "Epoch 2716/5000\n",
+      "919/919 - 3s - loss: 1.4565 - accuracy: 0.5642 - val_loss: 3.7483 - val_accuracy: 0.0735\n",
+      "Epoch 2717/5000\n",
+      "919/919 - 3s - loss: 1.4403 - accuracy: 0.5630 - val_loss: 3.7550 - val_accuracy: 0.0739\n",
+      "Epoch 2718/5000\n",
+      "919/919 - 3s - loss: 1.3603 - accuracy: 0.5684 - val_loss: 3.7556 - val_accuracy: 0.0740\n",
+      "Epoch 2719/5000\n",
+      "919/919 - 3s - loss: 1.3824 - accuracy: 0.5696 - val_loss: 3.7685 - val_accuracy: 0.0735\n",
+      "Epoch 2720/5000\n",
+      "919/919 - 3s - loss: 1.3584 - accuracy: 0.5657 - val_loss: 3.7643 - val_accuracy: 0.0737\n",
+      "Epoch 2721/5000\n",
+      "919/919 - 3s - loss: 1.3503 - accuracy: 0.5683 - val_loss: 3.7630 - val_accuracy: 0.0741\n",
+      "Epoch 2722/5000\n",
+      "919/919 - 3s - loss: 1.3574 - accuracy: 0.5681 - val_loss: 3.7655 - val_accuracy: 0.0739\n",
+      "Epoch 2723/5000\n",
+      "919/919 - 3s - loss: 1.3762 - accuracy: 0.5638 - val_loss: 3.7638 - val_accuracy: 0.0738\n",
+      "Epoch 2724/5000\n",
+      "919/919 - 3s - loss: 1.3655 - accuracy: 0.5697 - val_loss: 3.7547 - val_accuracy: 0.0740\n",
+      "Epoch 2725/5000\n",
+      "919/919 - 3s - loss: 1.4760 - accuracy: 0.5646 - val_loss: 3.7562 - val_accuracy: 0.0743\n",
+      "Epoch 2726/5000\n",
+      "919/919 - 3s - loss: 1.4253 - accuracy: 0.5669 - val_loss: 3.7660 - val_accuracy: 0.0740\n",
+      "Epoch 2727/5000\n",
+      "919/919 - 3s - loss: 1.3814 - accuracy: 0.5657 - val_loss: 3.7692 - val_accuracy: 0.0735\n",
+      "Epoch 2728/5000\n",
+      "919/919 - 3s - loss: 1.3674 - accuracy: 0.5712 - val_loss: 3.7648 - val_accuracy: 0.0737\n",
+      "Epoch 2729/5000\n",
+      "919/919 - 3s - loss: 1.3697 - accuracy: 0.5707 - val_loss: 3.7618 - val_accuracy: 0.0741\n",
+      "Epoch 2730/5000\n",
+      "919/919 - 3s - loss: 1.3630 - accuracy: 0.5656 - val_loss: 3.7647 - val_accuracy: 0.0739\n",
+      "Epoch 2731/5000\n",
+      "919/919 - 3s - loss: 1.3619 - accuracy: 0.5646 - val_loss: 3.7582 - val_accuracy: 0.0741\n",
+      "Epoch 2732/5000\n",
+      "919/919 - 3s - loss: 1.3637 - accuracy: 0.5694 - val_loss: 3.7565 - val_accuracy: 0.0738\n",
+      "Epoch 2733/5000\n",
+      "919/919 - 3s - loss: 1.3759 - accuracy: 0.5670 - val_loss: 3.7554 - val_accuracy: 0.0740\n",
+      "Epoch 2734/5000\n",
+      "919/919 - 3s - loss: 1.3649 - accuracy: 0.5695 - val_loss: 3.7664 - val_accuracy: 0.0741\n",
+      "Epoch 2735/5000\n",
+      "919/919 - 3s - loss: 1.3734 - accuracy: 0.5696 - val_loss: 3.7687 - val_accuracy: 0.0740\n",
+      "Epoch 2736/5000\n",
+      "919/919 - 3s - loss: 1.3595 - accuracy: 0.5727 - val_loss: 3.7516 - val_accuracy: 0.0736\n",
+      "Epoch 2737/5000\n",
+      "919/919 - 3s - loss: 1.3651 - accuracy: 0.5681 - val_loss: 3.7476 - val_accuracy: 0.0734\n",
+      "Epoch 2738/5000\n",
+      "919/919 - 3s - loss: 1.4215 - accuracy: 0.5667 - val_loss: 3.7608 - val_accuracy: 0.0739\n",
+      "Epoch 2739/5000\n",
+      "919/919 - 3s - loss: 1.3653 - accuracy: 0.5659 - val_loss: 3.7576 - val_accuracy: 0.0737\n",
+      "Epoch 2740/5000\n",
+      "919/919 - 3s - loss: 1.3749 - accuracy: 0.5673 - val_loss: 3.7602 - val_accuracy: 0.0734\n",
+      "Epoch 2741/5000\n",
+      "919/919 - 3s - loss: 1.3732 - accuracy: 0.5674 - val_loss: 3.7609 - val_accuracy: 0.0739\n",
+      "Epoch 2742/5000\n",
+      "919/919 - 3s - loss: 1.4074 - accuracy: 0.5618 - val_loss: 3.7512 - val_accuracy: 0.0737\n",
+      "Epoch 2743/5000\n",
+      "919/919 - 3s - loss: 1.3595 - accuracy: 0.5705 - val_loss: 3.7455 - val_accuracy: 0.0737\n",
+      "Epoch 2744/5000\n",
+      "919/919 - 3s - loss: 1.3573 - accuracy: 0.5695 - val_loss: 3.7487 - val_accuracy: 0.0745\n",
+      "Epoch 2745/5000\n",
+      "919/919 - 3s - loss: 1.3657 - accuracy: 0.5648 - val_loss: 3.7509 - val_accuracy: 0.0744\n",
+      "Epoch 2746/5000\n",
+      "919/919 - 3s - loss: 1.3495 - accuracy: 0.5718 - val_loss: 3.7589 - val_accuracy: 0.0740\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2747/5000\n",
+      "919/919 - 3s - loss: 1.3872 - accuracy: 0.5648 - val_loss: 3.7598 - val_accuracy: 0.0737\n",
+      "Epoch 2748/5000\n",
+      "919/919 - 3s - loss: 1.4481 - accuracy: 0.5672 - val_loss: 3.7599 - val_accuracy: 0.0738\n",
+      "Epoch 2749/5000\n",
+      "919/919 - 3s - loss: 1.3561 - accuracy: 0.5724 - val_loss: 3.7592 - val_accuracy: 0.0742\n",
+      "Epoch 2750/5000\n",
+      "919/919 - 3s - loss: 1.3625 - accuracy: 0.5699 - val_loss: 3.7529 - val_accuracy: 0.0739\n",
+      "Epoch 2751/5000\n",
+      "919/919 - 3s - loss: 1.3777 - accuracy: 0.5651 - val_loss: 3.7473 - val_accuracy: 0.0741\n",
+      "Epoch 2752/5000\n",
+      "919/919 - 3s - loss: 1.3606 - accuracy: 0.5654 - val_loss: 3.7506 - val_accuracy: 0.0741\n",
+      "Epoch 2753/5000\n",
+      "919/919 - 3s - loss: 1.3578 - accuracy: 0.5664 - val_loss: 3.7439 - val_accuracy: 0.0738\n",
+      "Epoch 2754/5000\n",
+      "919/919 - 3s - loss: 1.3577 - accuracy: 0.5663 - val_loss: 3.7570 - val_accuracy: 0.0741\n",
+      "Epoch 2755/5000\n",
+      "919/919 - 3s - loss: 1.3689 - accuracy: 0.5671 - val_loss: 3.7526 - val_accuracy: 0.0740\n",
+      "Epoch 2756/5000\n",
+      "919/919 - 3s - loss: 1.3612 - accuracy: 0.5659 - val_loss: 3.7593 - val_accuracy: 0.0734\n",
+      "Epoch 2757/5000\n",
+      "919/919 - 3s - loss: 1.3465 - accuracy: 0.5739 - val_loss: 3.7561 - val_accuracy: 0.0741\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2758/5000\n",
+      "919/919 - 3s - loss: 1.3878 - accuracy: 0.5722 - val_loss: 3.7529 - val_accuracy: 0.0749\n",
+      "Epoch 2759/5000\n",
+      "919/919 - 3s - loss: 1.3638 - accuracy: 0.5686 - val_loss: 3.7554 - val_accuracy: 0.0744\n",
+      "Epoch 2760/5000\n",
+      "919/919 - 3s - loss: 1.4123 - accuracy: 0.5673 - val_loss: 3.7622 - val_accuracy: 0.0742\n",
+      "Epoch 2761/5000\n",
+      "919/919 - 3s - loss: 1.3572 - accuracy: 0.5676 - val_loss: 3.7615 - val_accuracy: 0.0742\n",
+      "Epoch 2762/5000\n",
+      "919/919 - 3s - loss: 1.4109 - accuracy: 0.5662 - val_loss: 3.7695 - val_accuracy: 0.0744\n",
+      "Epoch 2763/5000\n",
+      "919/919 - 3s - loss: 1.3934 - accuracy: 0.5725 - val_loss: 3.7653 - val_accuracy: 0.0744\n",
+      "Epoch 2764/5000\n",
+      "919/919 - 3s - loss: 1.3667 - accuracy: 0.5661 - val_loss: 3.7536 - val_accuracy: 0.0742\n",
+      "Epoch 2765/5000\n",
+      "919/919 - 3s - loss: 1.3739 - accuracy: 0.5676 - val_loss: 3.7534 - val_accuracy: 0.0741\n",
+      "Epoch 2766/5000\n",
+      "919/919 - 3s - loss: 1.3648 - accuracy: 0.5646 - val_loss: 3.7583 - val_accuracy: 0.0739\n",
+      "Epoch 2767/5000\n",
+      "919/919 - 3s - loss: 1.3492 - accuracy: 0.5704 - val_loss: 3.7525 - val_accuracy: 0.0739\n",
+      "Epoch 2768/5000\n",
+      "919/919 - 3s - loss: 1.3586 - accuracy: 0.5742 - val_loss: 3.7596 - val_accuracy: 0.0742\n",
+      "Epoch 2769/5000\n",
+      "919/919 - 3s - loss: 1.3773 - accuracy: 0.5648 - val_loss: 3.7469 - val_accuracy: 0.0744\n",
+      "Epoch 2770/5000\n",
+      "919/919 - 3s - loss: 1.3620 - accuracy: 0.5701 - val_loss: 3.7477 - val_accuracy: 0.0746\n",
+      "Epoch 2771/5000\n",
+      "919/919 - 3s - loss: 1.3541 - accuracy: 0.5713 - val_loss: 3.7501 - val_accuracy: 0.0749\n",
+      "Epoch 2772/5000\n",
+      "919/919 - 3s - loss: 1.3658 - accuracy: 0.5659 - val_loss: 3.7486 - val_accuracy: 0.0743\n",
+      "Epoch 2773/5000\n",
+      "919/919 - 3s - loss: 1.3668 - accuracy: 0.5660 - val_loss: 3.7488 - val_accuracy: 0.0748\n",
+      "Epoch 2774/5000\n",
+      "919/919 - 3s - loss: 1.3976 - accuracy: 0.5726 - val_loss: 3.7545 - val_accuracy: 0.0746\n",
+      "Epoch 2775/5000\n",
+      "919/919 - 3s - loss: 1.3413 - accuracy: 0.5726 - val_loss: 3.7623 - val_accuracy: 0.0749\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2776/5000\n",
+      "919/919 - 3s - loss: 1.3686 - accuracy: 0.5706 - val_loss: 3.7680 - val_accuracy: 0.0748\n",
+      "Epoch 2777/5000\n",
+      "919/919 - 3s - loss: 1.4358 - accuracy: 0.5679 - val_loss: 3.7737 - val_accuracy: 0.0747\n",
+      "Epoch 2778/5000\n",
+      "919/919 - 3s - loss: 1.3723 - accuracy: 0.5646 - val_loss: 3.7747 - val_accuracy: 0.0742\n",
+      "Epoch 2779/5000\n",
+      "919/919 - 3s - loss: 1.3510 - accuracy: 0.5727 - val_loss: 3.7782 - val_accuracy: 0.0744\n",
+      "Epoch 2780/5000\n",
+      "919/919 - 3s - loss: 1.3775 - accuracy: 0.5673 - val_loss: 3.7783 - val_accuracy: 0.0744\n",
+      "Epoch 2781/5000\n",
+      "919/919 - 3s - loss: 1.3540 - accuracy: 0.5688 - val_loss: 3.7840 - val_accuracy: 0.0741\n",
+      "Epoch 2782/5000\n",
+      "919/919 - 3s - loss: 1.3614 - accuracy: 0.5710 - val_loss: 3.7749 - val_accuracy: 0.0738\n",
+      "Epoch 2783/5000\n",
+      "919/919 - 3s - loss: 1.3607 - accuracy: 0.5714 - val_loss: 3.7737 - val_accuracy: 0.0743\n",
+      "Epoch 2784/5000\n",
+      "919/919 - 3s - loss: 1.3703 - accuracy: 0.5699 - val_loss: 3.7667 - val_accuracy: 0.0751\n",
+      "Epoch 2785/5000\n",
+      "919/919 - 3s - loss: 1.3607 - accuracy: 0.5684 - val_loss: 3.7673 - val_accuracy: 0.0747\n",
+      "Epoch 2786/5000\n",
+      "919/919 - 3s - loss: 1.3582 - accuracy: 0.5720 - val_loss: 3.7709 - val_accuracy: 0.0749\n",
+      "Epoch 2787/5000\n",
+      "919/919 - 3s - loss: 1.3626 - accuracy: 0.5680 - val_loss: 3.7724 - val_accuracy: 0.0748\n",
+      "Epoch 2788/5000\n",
+      "919/919 - 3s - loss: 1.3586 - accuracy: 0.5672 - val_loss: 3.7732 - val_accuracy: 0.0745\n",
+      "Epoch 2789/5000\n",
+      "919/919 - 3s - loss: 1.3516 - accuracy: 0.5725 - val_loss: 3.7695 - val_accuracy: 0.0739\n",
+      "Epoch 2790/5000\n",
+      "919/919 - 3s - loss: 1.3538 - accuracy: 0.5688 - val_loss: 3.7660 - val_accuracy: 0.0747\n",
+      "Epoch 2791/5000\n",
+      "919/919 - 3s - loss: 1.3416 - accuracy: 0.5733 - val_loss: 3.7728 - val_accuracy: 0.0748\n",
+      "Epoch 2792/5000\n",
+      "919/919 - 3s - loss: 1.3447 - accuracy: 0.5695 - val_loss: 3.7698 - val_accuracy: 0.0735\n",
+      "Epoch 2793/5000\n",
+      "919/919 - 3s - loss: 1.3569 - accuracy: 0.5686 - val_loss: 3.7740 - val_accuracy: 0.0739\n",
+      "Epoch 2794/5000\n",
+      "919/919 - 3s - loss: 1.3535 - accuracy: 0.5727 - val_loss: 3.7785 - val_accuracy: 0.0740\n",
+      "Epoch 2795/5000\n",
+      "919/919 - 3s - loss: 1.3483 - accuracy: 0.5680 - val_loss: 3.7740 - val_accuracy: 0.0744\n",
+      "Epoch 2796/5000\n",
+      "919/919 - 3s - loss: 1.3680 - accuracy: 0.5724 - val_loss: 3.7684 - val_accuracy: 0.0749\n",
+      "Epoch 2797/5000\n",
+      "919/919 - 3s - loss: 1.3576 - accuracy: 0.5671 - val_loss: 3.7758 - val_accuracy: 0.0752\n",
+      "Epoch 2798/5000\n",
+      "919/919 - 3s - loss: 1.3525 - accuracy: 0.5724 - val_loss: 3.7715 - val_accuracy: 0.0752\n",
+      "Epoch 2799/5000\n",
+      "919/919 - 3s - loss: 1.3826 - accuracy: 0.5709 - val_loss: 3.7724 - val_accuracy: 0.0747\n",
+      "Epoch 2800/5000\n",
+      "919/919 - 3s - loss: 1.3781 - accuracy: 0.5719 - val_loss: 3.7669 - val_accuracy: 0.0743\n",
+      "Epoch 2801/5000\n",
+      "919/919 - 3s - loss: 1.3628 - accuracy: 0.5685 - val_loss: 3.7579 - val_accuracy: 0.0742\n",
+      "Epoch 2802/5000\n",
+      "919/919 - 3s - loss: 1.3935 - accuracy: 0.5712 - val_loss: 3.7565 - val_accuracy: 0.0745\n",
+      "Epoch 2803/5000\n",
+      "919/919 - 3s - loss: 1.3539 - accuracy: 0.5690 - val_loss: 3.7607 - val_accuracy: 0.0749\n",
+      "Epoch 2804/5000\n",
+      "919/919 - 3s - loss: 1.4394 - accuracy: 0.5737 - val_loss: 3.7625 - val_accuracy: 0.0749\n",
+      "Epoch 2805/5000\n",
+      "919/919 - 3s - loss: 1.3424 - accuracy: 0.5748 - val_loss: 3.7708 - val_accuracy: 0.0749\n",
+      "Epoch 2806/5000\n",
+      "919/919 - 3s - loss: 1.3655 - accuracy: 0.5701 - val_loss: 3.7724 - val_accuracy: 0.0745\n",
+      "Epoch 2807/5000\n",
+      "919/919 - 3s - loss: 1.3531 - accuracy: 0.5724 - val_loss: 3.7596 - val_accuracy: 0.0749\n",
+      "Epoch 2808/5000\n",
+      "919/919 - 3s - loss: 1.4070 - accuracy: 0.5745 - val_loss: 3.7707 - val_accuracy: 0.0749\n",
+      "Epoch 2809/5000\n",
+      "919/919 - 3s - loss: 1.3587 - accuracy: 0.5688 - val_loss: 3.7574 - val_accuracy: 0.0759\n",
+      "Epoch 2810/5000\n",
+      "919/919 - 3s - loss: 1.3596 - accuracy: 0.5684 - val_loss: 3.7511 - val_accuracy: 0.0763\n",
+      "Epoch 2811/5000\n",
+      "919/919 - 3s - loss: 1.3406 - accuracy: 0.5703 - val_loss: 3.7541 - val_accuracy: 0.0758\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2812/5000\n",
+      "919/919 - 3s - loss: 1.3484 - accuracy: 0.5722 - val_loss: 3.7586 - val_accuracy: 0.0749\n",
+      "Epoch 2813/5000\n",
+      "919/919 - 3s - loss: 1.3500 - accuracy: 0.5703 - val_loss: 3.7757 - val_accuracy: 0.0755\n",
+      "Epoch 2814/5000\n",
+      "919/919 - 3s - loss: 1.3517 - accuracy: 0.5739 - val_loss: 3.7709 - val_accuracy: 0.0749\n",
+      "Epoch 2815/5000\n",
+      "919/919 - 3s - loss: 1.3505 - accuracy: 0.5748 - val_loss: 3.7784 - val_accuracy: 0.0762\n",
+      "Epoch 2816/5000\n",
+      "919/919 - 3s - loss: 1.3859 - accuracy: 0.5731 - val_loss: 3.7725 - val_accuracy: 0.0752\n",
+      "Epoch 2817/5000\n",
+      "919/919 - 3s - loss: 1.4259 - accuracy: 0.5719 - val_loss: 3.7737 - val_accuracy: 0.0756\n",
+      "Epoch 2818/5000\n",
+      "919/919 - 3s - loss: 1.3376 - accuracy: 0.5776 - val_loss: 3.7671 - val_accuracy: 0.0758\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2819/5000\n",
+      "919/919 - 3s - loss: 1.3553 - accuracy: 0.5689 - val_loss: 3.7721 - val_accuracy: 0.0757\n",
+      "Epoch 2820/5000\n",
+      "919/919 - 3s - loss: 1.3332 - accuracy: 0.5783 - val_loss: 3.7716 - val_accuracy: 0.0757\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2821/5000\n",
+      "919/919 - 3s - loss: 1.3614 - accuracy: 0.5702 - val_loss: 3.7631 - val_accuracy: 0.0760\n",
+      "Epoch 2822/5000\n",
+      "919/919 - 3s - loss: 1.3569 - accuracy: 0.5761 - val_loss: 3.7629 - val_accuracy: 0.0769\n",
+      "Epoch 2823/5000\n",
+      "919/919 - 3s - loss: 1.3411 - accuracy: 0.5742 - val_loss: 3.7639 - val_accuracy: 0.0767\n",
+      "Epoch 2824/5000\n",
+      "919/919 - 3s - loss: 1.4629 - accuracy: 0.5709 - val_loss: 3.7622 - val_accuracy: 0.0760\n",
+      "Epoch 2825/5000\n",
+      "919/919 - 3s - loss: 1.3576 - accuracy: 0.5709 - val_loss: 3.7633 - val_accuracy: 0.0764\n",
+      "Epoch 2826/5000\n",
+      "919/919 - 3s - loss: 1.3847 - accuracy: 0.5682 - val_loss: 3.7596 - val_accuracy: 0.0758\n",
+      "Epoch 2827/5000\n",
+      "919/919 - 3s - loss: 1.3517 - accuracy: 0.5731 - val_loss: 3.7658 - val_accuracy: 0.0760\n",
+      "Epoch 2828/5000\n",
+      "919/919 - 3s - loss: 1.4553 - accuracy: 0.5780 - val_loss: 3.7597 - val_accuracy: 0.0762\n",
+      "Epoch 2829/5000\n",
+      "919/919 - 3s - loss: 1.3818 - accuracy: 0.5786 - val_loss: 3.7672 - val_accuracy: 0.0760\n",
+      "Epoch 2830/5000\n",
+      "919/919 - 3s - loss: 1.3645 - accuracy: 0.5727 - val_loss: 3.7601 - val_accuracy: 0.0760\n",
+      "Epoch 2831/5000\n",
+      "919/919 - 3s - loss: 1.3464 - accuracy: 0.5723 - val_loss: 3.7623 - val_accuracy: 0.0767\n",
+      "Epoch 2832/5000\n",
+      "919/919 - 3s - loss: 1.3377 - accuracy: 0.5774 - val_loss: 3.7559 - val_accuracy: 0.0760\n",
+      "Epoch 2833/5000\n",
+      "919/919 - 3s - loss: 1.3460 - accuracy: 0.5726 - val_loss: 3.7603 - val_accuracy: 0.0764\n",
+      "Epoch 2834/5000\n",
+      "919/919 - 3s - loss: 1.3839 - accuracy: 0.5769 - val_loss: 3.7665 - val_accuracy: 0.0758\n",
+      "Epoch 2835/5000\n",
+      "919/919 - 3s - loss: 1.3466 - accuracy: 0.5722 - val_loss: 3.7668 - val_accuracy: 0.0756\n",
+      "Epoch 2836/5000\n",
+      "919/919 - 3s - loss: 1.3488 - accuracy: 0.5716 - val_loss: 3.7579 - val_accuracy: 0.0752\n",
+      "Epoch 2837/5000\n",
+      "919/919 - 3s - loss: 1.3519 - accuracy: 0.5729 - val_loss: 3.7422 - val_accuracy: 0.0761\n",
+      "Epoch 2838/5000\n",
+      "919/919 - 3s - loss: 1.3448 - accuracy: 0.5691 - val_loss: 3.7587 - val_accuracy: 0.0761\n",
+      "Epoch 2839/5000\n",
+      "919/919 - 3s - loss: 1.3685 - accuracy: 0.5771 - val_loss: 3.7650 - val_accuracy: 0.0764\n",
+      "Epoch 2840/5000\n",
+      "919/919 - 3s - loss: 1.3538 - accuracy: 0.5738 - val_loss: 3.7742 - val_accuracy: 0.0754\n",
+      "Epoch 2841/5000\n",
+      "919/919 - 3s - loss: 1.3559 - accuracy: 0.5739 - val_loss: 3.7824 - val_accuracy: 0.0755\n",
+      "Epoch 2842/5000\n",
+      "919/919 - 3s - loss: 1.3587 - accuracy: 0.5705 - val_loss: 3.7773 - val_accuracy: 0.0753\n",
+      "Epoch 2843/5000\n",
+      "919/919 - 3s - loss: 1.3465 - accuracy: 0.5717 - val_loss: 3.7732 - val_accuracy: 0.0763\n",
+      "Epoch 2844/5000\n",
+      "919/919 - 3s - loss: 1.3600 - accuracy: 0.5723 - val_loss: 3.7663 - val_accuracy: 0.0760\n",
+      "Epoch 2845/5000\n",
+      "919/919 - 3s - loss: 1.3540 - accuracy: 0.5743 - val_loss: 3.7644 - val_accuracy: 0.0757\n",
+      "Epoch 2846/5000\n",
+      "919/919 - 3s - loss: 1.3443 - accuracy: 0.5718 - val_loss: 3.7742 - val_accuracy: 0.0751\n",
+      "Epoch 2847/5000\n",
+      "919/919 - 3s - loss: 1.3391 - accuracy: 0.5782 - val_loss: 3.7664 - val_accuracy: 0.0756\n",
+      "Epoch 2848/5000\n",
+      "919/919 - 3s - loss: 1.3637 - accuracy: 0.5756 - val_loss: 3.7565 - val_accuracy: 0.0759\n",
+      "Epoch 2849/5000\n",
+      "919/919 - 3s - loss: 1.3302 - accuracy: 0.5769 - val_loss: 3.7635 - val_accuracy: 0.0761\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2850/5000\n",
+      "919/919 - 3s - loss: 1.3454 - accuracy: 0.5716 - val_loss: 3.7714 - val_accuracy: 0.0759\n",
+      "Epoch 2851/5000\n",
+      "919/919 - 3s - loss: 1.3587 - accuracy: 0.5748 - val_loss: 3.7695 - val_accuracy: 0.0756\n",
+      "Epoch 2852/5000\n",
+      "919/919 - 3s - loss: 1.3724 - accuracy: 0.5720 - val_loss: 3.7623 - val_accuracy: 0.0756\n",
+      "Epoch 2853/5000\n",
+      "919/919 - 3s - loss: 1.3613 - accuracy: 0.5761 - val_loss: 3.7555 - val_accuracy: 0.0760\n",
+      "Epoch 2854/5000\n",
+      "919/919 - 3s - loss: 1.3468 - accuracy: 0.5735 - val_loss: 3.7593 - val_accuracy: 0.0752\n",
+      "Epoch 2855/5000\n",
+      "919/919 - 3s - loss: 1.3571 - accuracy: 0.5752 - val_loss: 3.7622 - val_accuracy: 0.0748\n",
+      "Epoch 2856/5000\n",
+      "919/919 - 3s - loss: 1.3799 - accuracy: 0.5719 - val_loss: 3.7678 - val_accuracy: 0.0750\n",
+      "Epoch 2857/5000\n",
+      "919/919 - 3s - loss: 1.3514 - accuracy: 0.5738 - val_loss: 3.7650 - val_accuracy: 0.0748\n",
+      "Epoch 2858/5000\n",
+      "919/919 - 3s - loss: 1.3866 - accuracy: 0.5773 - val_loss: 3.7811 - val_accuracy: 0.0759\n",
+      "Epoch 2859/5000\n",
+      "919/919 - 3s - loss: 1.3785 - accuracy: 0.5755 - val_loss: 3.7704 - val_accuracy: 0.0767\n",
+      "Epoch 2860/5000\n",
+      "919/919 - 3s - loss: 1.3498 - accuracy: 0.5742 - val_loss: 3.7676 - val_accuracy: 0.0758\n",
+      "Epoch 2861/5000\n",
+      "919/919 - 3s - loss: 1.3416 - accuracy: 0.5747 - val_loss: 3.7781 - val_accuracy: 0.0753\n",
+      "Epoch 2862/5000\n",
+      "919/919 - 3s - loss: 1.3426 - accuracy: 0.5788 - val_loss: 3.7801 - val_accuracy: 0.0751\n",
+      "Epoch 2863/5000\n",
+      "919/919 - 3s - loss: 1.3777 - accuracy: 0.5716 - val_loss: 3.7754 - val_accuracy: 0.0749\n",
+      "Epoch 2864/5000\n",
+      "919/919 - 3s - loss: 1.3471 - accuracy: 0.5801 - val_loss: 3.7861 - val_accuracy: 0.0754\n",
+      "Epoch 2865/5000\n",
+      "919/919 - 3s - loss: 1.3602 - accuracy: 0.5741 - val_loss: 3.7890 - val_accuracy: 0.0762\n",
+      "Epoch 2866/5000\n",
+      "919/919 - 3s - loss: 1.3441 - accuracy: 0.5772 - val_loss: 3.7997 - val_accuracy: 0.0758\n",
+      "Epoch 2867/5000\n",
+      "919/919 - 3s - loss: 1.3497 - accuracy: 0.5743 - val_loss: 3.8024 - val_accuracy: 0.0759\n",
+      "Epoch 2868/5000\n",
+      "919/919 - 3s - loss: 1.3570 - accuracy: 0.5721 - val_loss: 3.7824 - val_accuracy: 0.0762\n",
+      "Epoch 2869/5000\n",
+      "919/919 - 3s - loss: 1.3539 - accuracy: 0.5720 - val_loss: 3.7864 - val_accuracy: 0.0762\n",
+      "Epoch 2870/5000\n",
+      "919/919 - 3s - loss: 1.3387 - accuracy: 0.5799 - val_loss: 3.7880 - val_accuracy: 0.0758\n",
+      "Epoch 2871/5000\n",
+      "919/919 - 3s - loss: 1.3449 - accuracy: 0.5759 - val_loss: 3.7860 - val_accuracy: 0.0758\n",
+      "Epoch 2872/5000\n",
+      "919/919 - 3s - loss: 1.3698 - accuracy: 0.5731 - val_loss: 3.7744 - val_accuracy: 0.0759\n",
+      "Epoch 2873/5000\n",
+      "919/919 - 3s - loss: 1.3494 - accuracy: 0.5717 - val_loss: 3.7907 - val_accuracy: 0.0755\n",
+      "Epoch 2874/5000\n",
+      "919/919 - 3s - loss: 1.4838 - accuracy: 0.5756 - val_loss: 3.7672 - val_accuracy: 0.0757\n",
+      "Epoch 2875/5000\n",
+      "919/919 - 3s - loss: 1.3415 - accuracy: 0.5768 - val_loss: 3.7769 - val_accuracy: 0.0758\n",
+      "Epoch 2876/5000\n",
+      "919/919 - 3s - loss: 1.3361 - accuracy: 0.5770 - val_loss: 3.7778 - val_accuracy: 0.0752\n",
+      "Epoch 2877/5000\n",
+      "919/919 - 3s - loss: 1.3303 - accuracy: 0.5748 - val_loss: 3.7894 - val_accuracy: 0.0752\n",
+      "Epoch 2878/5000\n",
+      "919/919 - 3s - loss: 1.3353 - accuracy: 0.5792 - val_loss: 3.7869 - val_accuracy: 0.0758\n",
+      "Epoch 2879/5000\n",
+      "919/919 - 3s - loss: 1.3529 - accuracy: 0.5753 - val_loss: 3.7862 - val_accuracy: 0.0755\n",
+      "Epoch 2880/5000\n",
+      "919/919 - 3s - loss: 1.3332 - accuracy: 0.5759 - val_loss: 3.7926 - val_accuracy: 0.0761\n",
+      "Epoch 2881/5000\n",
+      "919/919 - 3s - loss: 1.3964 - accuracy: 0.5759 - val_loss: 3.7813 - val_accuracy: 0.0756\n",
+      "Epoch 2882/5000\n",
+      "919/919 - 3s - loss: 1.3353 - accuracy: 0.5783 - val_loss: 3.7812 - val_accuracy: 0.0765\n",
+      "Epoch 2883/5000\n",
+      "919/919 - 3s - loss: 1.3530 - accuracy: 0.5774 - val_loss: 3.8058 - val_accuracy: 0.0769\n",
+      "Epoch 2884/5000\n",
+      "919/919 - 3s - loss: 1.3388 - accuracy: 0.5761 - val_loss: 3.7975 - val_accuracy: 0.0765\n",
+      "Epoch 2885/5000\n",
+      "919/919 - 3s - loss: 1.3456 - accuracy: 0.5758 - val_loss: 3.7893 - val_accuracy: 0.0766\n",
+      "Epoch 2886/5000\n",
+      "919/919 - 3s - loss: 1.3275 - accuracy: 0.5771 - val_loss: 3.7868 - val_accuracy: 0.0767\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2887/5000\n",
+      "919/919 - 3s - loss: 1.3393 - accuracy: 0.5779 - val_loss: 3.7873 - val_accuracy: 0.0764\n",
+      "Epoch 2888/5000\n",
+      "919/919 - 3s - loss: 1.3493 - accuracy: 0.5737 - val_loss: 3.7818 - val_accuracy: 0.0762\n",
+      "Epoch 2889/5000\n",
+      "919/919 - 3s - loss: 1.3439 - accuracy: 0.5751 - val_loss: 3.7777 - val_accuracy: 0.0760\n",
+      "Epoch 2890/5000\n",
+      "919/919 - 3s - loss: 1.3461 - accuracy: 0.5775 - val_loss: 3.7868 - val_accuracy: 0.0758\n",
+      "Epoch 2891/5000\n",
+      "919/919 - 3s - loss: 1.3414 - accuracy: 0.5727 - val_loss: 3.7809 - val_accuracy: 0.0753\n",
+      "Epoch 2892/5000\n",
+      "919/919 - 3s - loss: 1.3396 - accuracy: 0.5778 - val_loss: 3.7825 - val_accuracy: 0.0755\n",
+      "Epoch 2893/5000\n",
+      "919/919 - 3s - loss: 1.3383 - accuracy: 0.5748 - val_loss: 3.7891 - val_accuracy: 0.0757\n",
+      "Epoch 2894/5000\n",
+      "919/919 - 3s - loss: 1.3304 - accuracy: 0.5814 - val_loss: 3.7983 - val_accuracy: 0.0759\n",
+      "Epoch 2895/5000\n",
+      "919/919 - 3s - loss: 1.3767 - accuracy: 0.5776 - val_loss: 3.7866 - val_accuracy: 0.0766\n",
+      "Epoch 2896/5000\n",
+      "919/919 - 3s - loss: 1.4445 - accuracy: 0.5746 - val_loss: 3.7892 - val_accuracy: 0.0758\n",
+      "Epoch 2897/5000\n",
+      "919/919 - 3s - loss: 1.3504 - accuracy: 0.5747 - val_loss: 3.7830 - val_accuracy: 0.0759\n",
+      "Epoch 2898/5000\n",
+      "919/919 - 3s - loss: 1.3298 - accuracy: 0.5766 - val_loss: 3.7943 - val_accuracy: 0.0751\n",
+      "Epoch 2899/5000\n",
+      "919/919 - 3s - loss: 1.3462 - accuracy: 0.5728 - val_loss: 3.7871 - val_accuracy: 0.0757\n",
+      "Epoch 2900/5000\n",
+      "919/919 - 3s - loss: 1.3305 - accuracy: 0.5767 - val_loss: 3.7991 - val_accuracy: 0.0756\n",
+      "Epoch 2901/5000\n",
+      "919/919 - 3s - loss: 1.3881 - accuracy: 0.5837 - val_loss: 3.7972 - val_accuracy: 0.0754\n",
+      "Epoch 2902/5000\n",
+      "919/919 - 3s - loss: 1.3382 - accuracy: 0.5744 - val_loss: 3.7925 - val_accuracy: 0.0759\n",
+      "Epoch 2903/5000\n",
+      "919/919 - 3s - loss: 1.3323 - accuracy: 0.5780 - val_loss: 3.7772 - val_accuracy: 0.0759\n",
+      "Epoch 2904/5000\n",
+      "919/919 - 3s - loss: 1.3528 - accuracy: 0.5759 - val_loss: 3.7778 - val_accuracy: 0.0764\n",
+      "Epoch 2905/5000\n",
+      "919/919 - 3s - loss: 1.3301 - accuracy: 0.5776 - val_loss: 3.7890 - val_accuracy: 0.0758\n",
+      "Epoch 2906/5000\n",
+      "919/919 - 3s - loss: 1.3433 - accuracy: 0.5760 - val_loss: 3.7989 - val_accuracy: 0.0761\n",
+      "Epoch 2907/5000\n",
+      "919/919 - 3s - loss: 1.3552 - accuracy: 0.5723 - val_loss: 3.7984 - val_accuracy: 0.0756\n",
+      "Epoch 2908/5000\n",
+      "919/919 - 3s - loss: 1.4472 - accuracy: 0.5716 - val_loss: 3.7934 - val_accuracy: 0.0757\n",
+      "Epoch 2909/5000\n",
+      "919/919 - 3s - loss: 1.3451 - accuracy: 0.5772 - val_loss: 3.7857 - val_accuracy: 0.0759\n",
+      "Epoch 2910/5000\n",
+      "919/919 - 3s - loss: 1.3382 - accuracy: 0.5783 - val_loss: 3.7768 - val_accuracy: 0.0759\n",
+      "Epoch 2911/5000\n",
+      "919/919 - 3s - loss: 1.3335 - accuracy: 0.5795 - val_loss: 3.7781 - val_accuracy: 0.0755\n",
+      "Epoch 2912/5000\n",
+      "919/919 - 3s - loss: 1.3406 - accuracy: 0.5751 - val_loss: 3.7812 - val_accuracy: 0.0758\n",
+      "Epoch 2913/5000\n",
+      "919/919 - 3s - loss: 1.3358 - accuracy: 0.5793 - val_loss: 3.7987 - val_accuracy: 0.0764\n",
+      "Epoch 2914/5000\n",
+      "919/919 - 3s - loss: 1.3310 - accuracy: 0.5814 - val_loss: 3.8023 - val_accuracy: 0.0760\n",
+      "Epoch 2915/5000\n",
+      "919/919 - 3s - loss: 1.3657 - accuracy: 0.5798 - val_loss: 3.8038 - val_accuracy: 0.0765\n",
+      "Epoch 2916/5000\n",
+      "919/919 - 3s - loss: 1.3364 - accuracy: 0.5787 - val_loss: 3.7913 - val_accuracy: 0.0764\n",
+      "Epoch 2917/5000\n",
+      "919/919 - 3s - loss: 1.3373 - accuracy: 0.5756 - val_loss: 3.7884 - val_accuracy: 0.0763\n",
+      "Epoch 2918/5000\n",
+      "919/919 - 3s - loss: 1.3347 - accuracy: 0.5765 - val_loss: 3.7807 - val_accuracy: 0.0768\n",
+      "Epoch 2919/5000\n",
+      "919/919 - 3s - loss: 1.3562 - accuracy: 0.5744 - val_loss: 3.7932 - val_accuracy: 0.0769\n",
+      "Epoch 2920/5000\n",
+      "919/919 - 3s - loss: 1.3707 - accuracy: 0.5757 - val_loss: 3.7903 - val_accuracy: 0.0763\n",
+      "Epoch 2921/5000\n",
+      "919/919 - 3s - loss: 1.3787 - accuracy: 0.5772 - val_loss: 3.7753 - val_accuracy: 0.0767\n",
+      "Epoch 2922/5000\n",
+      "919/919 - 3s - loss: 1.3349 - accuracy: 0.5803 - val_loss: 3.7825 - val_accuracy: 0.0766\n",
+      "Epoch 2923/5000\n",
+      "919/919 - 3s - loss: 1.3360 - accuracy: 0.5766 - val_loss: 3.8015 - val_accuracy: 0.0766\n",
+      "Epoch 2924/5000\n",
+      "919/919 - 3s - loss: 1.4279 - accuracy: 0.5846 - val_loss: 3.7963 - val_accuracy: 0.0767\n",
+      "Epoch 2925/5000\n",
+      "919/919 - 3s - loss: 1.3268 - accuracy: 0.5812 - val_loss: 3.8018 - val_accuracy: 0.0763\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2926/5000\n",
+      "919/919 - 3s - loss: 1.3511 - accuracy: 0.5768 - val_loss: 3.7957 - val_accuracy: 0.0766\n",
+      "Epoch 2927/5000\n",
+      "919/919 - 3s - loss: 1.4771 - accuracy: 0.5776 - val_loss: 3.7909 - val_accuracy: 0.0765\n",
+      "Epoch 2928/5000\n",
+      "919/919 - 3s - loss: 1.3449 - accuracy: 0.5782 - val_loss: 3.7970 - val_accuracy: 0.0764\n",
+      "Epoch 2929/5000\n",
+      "919/919 - 3s - loss: 1.4095 - accuracy: 0.5759 - val_loss: 3.7879 - val_accuracy: 0.0761\n",
+      "Epoch 2930/5000\n",
+      "919/919 - 3s - loss: 1.3248 - accuracy: 0.5759 - val_loss: 3.7770 - val_accuracy: 0.0764\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2931/5000\n",
+      "919/919 - 3s - loss: 1.3254 - accuracy: 0.5810 - val_loss: 3.7822 - val_accuracy: 0.0760\n",
+      "Epoch 2932/5000\n",
+      "919/919 - 3s - loss: 1.3475 - accuracy: 0.5765 - val_loss: 3.7997 - val_accuracy: 0.0758\n",
+      "Epoch 2933/5000\n",
+      "919/919 - 3s - loss: 1.3219 - accuracy: 0.5831 - val_loss: 3.7971 - val_accuracy: 0.0759\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2934/5000\n",
+      "919/919 - 3s - loss: 1.3439 - accuracy: 0.5780 - val_loss: 3.7977 - val_accuracy: 0.0760\n",
+      "Epoch 2935/5000\n",
+      "919/919 - 3s - loss: 1.3537 - accuracy: 0.5805 - val_loss: 3.8083 - val_accuracy: 0.0762\n",
+      "Epoch 2936/5000\n",
+      "919/919 - 3s - loss: 1.3314 - accuracy: 0.5780 - val_loss: 3.8012 - val_accuracy: 0.0753\n",
+      "Epoch 2937/5000\n",
+      "919/919 - 3s - loss: 1.3235 - accuracy: 0.5801 - val_loss: 3.7928 - val_accuracy: 0.0754\n",
+      "Epoch 2938/5000\n",
+      "919/919 - 3s - loss: 1.4270 - accuracy: 0.5761 - val_loss: 3.7966 - val_accuracy: 0.0765\n",
+      "Epoch 2939/5000\n",
+      "919/919 - 3s - loss: 1.3911 - accuracy: 0.5789 - val_loss: 3.7872 - val_accuracy: 0.0764\n",
+      "Epoch 2940/5000\n",
+      "919/919 - 3s - loss: 1.3709 - accuracy: 0.5767 - val_loss: 3.7801 - val_accuracy: 0.0766\n",
+      "Epoch 2941/5000\n",
+      "919/919 - 3s - loss: 1.3428 - accuracy: 0.5770 - val_loss: 3.7728 - val_accuracy: 0.0763\n",
+      "Epoch 2942/5000\n",
+      "919/919 - 3s - loss: 1.3272 - accuracy: 0.5772 - val_loss: 3.7699 - val_accuracy: 0.0762\n",
+      "Epoch 2943/5000\n",
+      "919/919 - 3s - loss: 1.3229 - accuracy: 0.5782 - val_loss: 3.7801 - val_accuracy: 0.0759\n",
+      "Epoch 2944/5000\n",
+      "919/919 - 3s - loss: 1.3464 - accuracy: 0.5766 - val_loss: 3.7793 - val_accuracy: 0.0768\n",
+      "Epoch 2945/5000\n",
+      "919/919 - 3s - loss: 1.3912 - accuracy: 0.5773 - val_loss: 3.7811 - val_accuracy: 0.0768\n",
+      "Epoch 2946/5000\n",
+      "919/919 - 3s - loss: 1.4750 - accuracy: 0.5804 - val_loss: 3.7854 - val_accuracy: 0.0769\n",
+      "Epoch 2947/5000\n",
+      "919/919 - 3s - loss: 1.3448 - accuracy: 0.5787 - val_loss: 3.7918 - val_accuracy: 0.0767\n",
+      "Epoch 2948/5000\n",
+      "919/919 - 3s - loss: 1.3399 - accuracy: 0.5801 - val_loss: 3.7822 - val_accuracy: 0.0763\n",
+      "Epoch 2949/5000\n",
+      "919/919 - 3s - loss: 1.3201 - accuracy: 0.5783 - val_loss: 3.7928 - val_accuracy: 0.0767\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2950/5000\n",
+      "919/919 - 3s - loss: 1.3289 - accuracy: 0.5808 - val_loss: 3.7882 - val_accuracy: 0.0760\n",
+      "Epoch 2951/5000\n",
+      "919/919 - 3s - loss: 1.3775 - accuracy: 0.5756 - val_loss: 3.7859 - val_accuracy: 0.0765\n",
+      "Epoch 2952/5000\n",
+      "919/919 - 3s - loss: 1.3711 - accuracy: 0.5787 - val_loss: 3.7979 - val_accuracy: 0.0756\n",
+      "Epoch 2953/5000\n",
+      "919/919 - 3s - loss: 1.3385 - accuracy: 0.5771 - val_loss: 3.7952 - val_accuracy: 0.0757\n",
+      "Epoch 2954/5000\n",
+      "919/919 - 3s - loss: 1.3749 - accuracy: 0.5829 - val_loss: 3.7932 - val_accuracy: 0.0758\n",
+      "Epoch 2955/5000\n",
+      "919/919 - 3s - loss: 1.3574 - accuracy: 0.5809 - val_loss: 3.7943 - val_accuracy: 0.0760\n",
+      "Epoch 2956/5000\n",
+      "919/919 - 3s - loss: 1.3269 - accuracy: 0.5816 - val_loss: 3.7944 - val_accuracy: 0.0759\n",
+      "Epoch 2957/5000\n",
+      "919/919 - 3s - loss: 1.3379 - accuracy: 0.5798 - val_loss: 3.8060 - val_accuracy: 0.0764\n",
+      "Epoch 2958/5000\n",
+      "919/919 - 3s - loss: 1.3441 - accuracy: 0.5816 - val_loss: 3.8001 - val_accuracy: 0.0767\n",
+      "Epoch 2959/5000\n",
+      "919/919 - 3s - loss: 1.3347 - accuracy: 0.5810 - val_loss: 3.8005 - val_accuracy: 0.0772\n",
+      "Epoch 2960/5000\n",
+      "919/919 - 3s - loss: 1.3214 - accuracy: 0.5773 - val_loss: 3.7947 - val_accuracy: 0.0779\n",
+      "Epoch 2961/5000\n",
+      "919/919 - 3s - loss: 1.3303 - accuracy: 0.5787 - val_loss: 3.7826 - val_accuracy: 0.0772\n",
+      "Epoch 2962/5000\n",
+      "919/919 - 3s - loss: 1.3344 - accuracy: 0.5818 - val_loss: 3.7859 - val_accuracy: 0.0779\n",
+      "Epoch 2963/5000\n",
+      "919/919 - 3s - loss: 1.3305 - accuracy: 0.5810 - val_loss: 3.7987 - val_accuracy: 0.0776\n",
+      "Epoch 2964/5000\n",
+      "919/919 - 3s - loss: 1.3223 - accuracy: 0.5819 - val_loss: 3.8067 - val_accuracy: 0.0770\n",
+      "Epoch 2965/5000\n",
+      "919/919 - 3s - loss: 1.3282 - accuracy: 0.5810 - val_loss: 3.8085 - val_accuracy: 0.0776\n",
+      "Epoch 2966/5000\n",
+      "919/919 - 3s - loss: 1.3276 - accuracy: 0.5799 - val_loss: 3.8134 - val_accuracy: 0.0773\n",
+      "Epoch 2967/5000\n",
+      "919/919 - 3s - loss: 1.3333 - accuracy: 0.5816 - val_loss: 3.8064 - val_accuracy: 0.0763\n",
+      "Epoch 2968/5000\n",
+      "919/919 - 3s - loss: 1.3404 - accuracy: 0.5798 - val_loss: 3.8001 - val_accuracy: 0.0768\n",
+      "Epoch 2969/5000\n",
+      "919/919 - 3s - loss: 1.3453 - accuracy: 0.5775 - val_loss: 3.7955 - val_accuracy: 0.0765\n",
+      "Epoch 2970/5000\n",
+      "919/919 - 3s - loss: 1.3299 - accuracy: 0.5804 - val_loss: 3.8009 - val_accuracy: 0.0763\n",
+      "Epoch 2971/5000\n",
+      "919/919 - 3s - loss: 1.3276 - accuracy: 0.5807 - val_loss: 3.8017 - val_accuracy: 0.0768\n",
+      "Epoch 2972/5000\n",
+      "919/919 - 3s - loss: 1.3318 - accuracy: 0.5814 - val_loss: 3.7939 - val_accuracy: 0.0779\n",
+      "Epoch 2973/5000\n",
+      "919/919 - 3s - loss: 1.3415 - accuracy: 0.5814 - val_loss: 3.7892 - val_accuracy: 0.0777\n",
+      "Epoch 2974/5000\n",
+      "919/919 - 3s - loss: 1.3288 - accuracy: 0.5844 - val_loss: 3.7866 - val_accuracy: 0.0768\n",
+      "Epoch 2975/5000\n",
+      "919/919 - 3s - loss: 1.3297 - accuracy: 0.5820 - val_loss: 3.7911 - val_accuracy: 0.0764\n",
+      "Epoch 2976/5000\n",
+      "919/919 - 3s - loss: 1.3323 - accuracy: 0.5803 - val_loss: 3.7991 - val_accuracy: 0.0764\n",
+      "Epoch 2977/5000\n",
+      "919/919 - 3s - loss: 1.3828 - accuracy: 0.5810 - val_loss: 3.8034 - val_accuracy: 0.0758\n",
+      "Epoch 2978/5000\n",
+      "919/919 - 3s - loss: 1.3290 - accuracy: 0.5799 - val_loss: 3.7972 - val_accuracy: 0.0761\n",
+      "Epoch 2979/5000\n",
+      "919/919 - 3s - loss: 1.3246 - accuracy: 0.5808 - val_loss: 3.8065 - val_accuracy: 0.0767\n",
+      "Epoch 2980/5000\n",
+      "919/919 - 3s - loss: 1.3198 - accuracy: 0.5850 - val_loss: 3.8005 - val_accuracy: 0.0768\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2981/5000\n",
+      "919/919 - 3s - loss: 1.3427 - accuracy: 0.5775 - val_loss: 3.7945 - val_accuracy: 0.0769\n",
+      "Epoch 2982/5000\n",
+      "919/919 - 3s - loss: 1.3312 - accuracy: 0.5844 - val_loss: 3.7965 - val_accuracy: 0.0757\n",
+      "Epoch 2983/5000\n",
+      "919/919 - 3s - loss: 1.3372 - accuracy: 0.5807 - val_loss: 3.7971 - val_accuracy: 0.0758\n",
+      "Epoch 2984/5000\n",
+      "919/919 - 3s - loss: 1.3268 - accuracy: 0.5829 - val_loss: 3.7811 - val_accuracy: 0.0764\n",
+      "Epoch 2985/5000\n",
+      "919/919 - 3s - loss: 1.3265 - accuracy: 0.5795 - val_loss: 3.7843 - val_accuracy: 0.0771\n",
+      "Epoch 2986/5000\n",
+      "919/919 - 3s - loss: 1.3378 - accuracy: 0.5774 - val_loss: 3.7941 - val_accuracy: 0.0773\n",
+      "Epoch 2987/5000\n",
+      "919/919 - 3s - loss: 1.3197 - accuracy: 0.5780 - val_loss: 3.7897 - val_accuracy: 0.0772\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2988/5000\n",
+      "919/919 - 3s - loss: 1.3586 - accuracy: 0.5799 - val_loss: 3.7918 - val_accuracy: 0.0775\n",
+      "Epoch 2989/5000\n",
+      "919/919 - 3s - loss: 1.3276 - accuracy: 0.5835 - val_loss: 3.7857 - val_accuracy: 0.0766\n",
+      "Epoch 2990/5000\n",
+      "919/919 - 3s - loss: 1.3210 - accuracy: 0.5797 - val_loss: 3.7873 - val_accuracy: 0.0771\n",
+      "Epoch 2991/5000\n",
+      "919/919 - 3s - loss: 1.3431 - accuracy: 0.5755 - val_loss: 3.7979 - val_accuracy: 0.0765\n",
+      "Epoch 2992/5000\n",
+      "919/919 - 3s - loss: 1.3113 - accuracy: 0.5871 - val_loss: 3.7958 - val_accuracy: 0.0773\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 2993/5000\n",
+      "919/919 - 3s - loss: 1.3333 - accuracy: 0.5791 - val_loss: 3.8028 - val_accuracy: 0.0766\n",
+      "Epoch 2994/5000\n",
+      "919/919 - 3s - loss: 1.3387 - accuracy: 0.5810 - val_loss: 3.8061 - val_accuracy: 0.0774\n",
+      "Epoch 2995/5000\n",
+      "919/919 - 3s - loss: 1.3259 - accuracy: 0.5823 - val_loss: 3.8158 - val_accuracy: 0.0771\n",
+      "Epoch 2996/5000\n",
+      "919/919 - 3s - loss: 1.3224 - accuracy: 0.5816 - val_loss: 3.8119 - val_accuracy: 0.0775\n",
+      "Epoch 2997/5000\n",
+      "919/919 - 3s - loss: 1.3229 - accuracy: 0.5806 - val_loss: 3.8217 - val_accuracy: 0.0775\n",
+      "Epoch 2998/5000\n",
+      "919/919 - 3s - loss: 1.3214 - accuracy: 0.5825 - val_loss: 3.8218 - val_accuracy: 0.0776\n",
+      "Epoch 2999/5000\n",
+      "919/919 - 3s - loss: 1.3529 - accuracy: 0.5837 - val_loss: 3.8130 - val_accuracy: 0.0784\n",
+      "Epoch 3000/5000\n",
+      "919/919 - 3s - loss: 1.3306 - accuracy: 0.5827 - val_loss: 3.8016 - val_accuracy: 0.0775\n",
+      "Epoch 3001/5000\n",
+      "919/919 - 3s - loss: 1.5140 - accuracy: 0.5810 - val_loss: 3.7975 - val_accuracy: 0.0778\n",
+      "Epoch 3002/5000\n",
+      "919/919 - 3s - loss: 1.3100 - accuracy: 0.5859 - val_loss: 3.8114 - val_accuracy: 0.0775\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3003/5000\n",
+      "919/919 - 3s - loss: 1.3264 - accuracy: 0.5830 - val_loss: 3.8145 - val_accuracy: 0.0774\n",
+      "Epoch 3004/5000\n",
+      "919/919 - 3s - loss: 1.3189 - accuracy: 0.5865 - val_loss: 3.8251 - val_accuracy: 0.0773\n",
+      "Epoch 3005/5000\n",
+      "919/919 - 3s - loss: 1.3205 - accuracy: 0.5846 - val_loss: 3.8116 - val_accuracy: 0.0775\n",
+      "Epoch 3006/5000\n",
+      "919/919 - 3s - loss: 1.3838 - accuracy: 0.5842 - val_loss: 3.7995 - val_accuracy: 0.0773\n",
+      "Epoch 3007/5000\n",
+      "919/919 - 3s - loss: 1.3377 - accuracy: 0.5832 - val_loss: 3.7943 - val_accuracy: 0.0777\n",
+      "Epoch 3008/5000\n",
+      "919/919 - 3s - loss: 1.3277 - accuracy: 0.5770 - val_loss: 3.8068 - val_accuracy: 0.0772\n",
+      "Epoch 3009/5000\n",
+      "919/919 - 3s - loss: 1.3188 - accuracy: 0.5841 - val_loss: 3.8025 - val_accuracy: 0.0771\n",
+      "Epoch 3010/5000\n",
+      "919/919 - 3s - loss: 1.3975 - accuracy: 0.5814 - val_loss: 3.8042 - val_accuracy: 0.0776\n",
+      "Epoch 3011/5000\n",
+      "919/919 - 3s - loss: 1.3287 - accuracy: 0.5821 - val_loss: 3.8041 - val_accuracy: 0.0772\n",
+      "Epoch 3012/5000\n",
+      "919/919 - 3s - loss: 1.3269 - accuracy: 0.5819 - val_loss: 3.7958 - val_accuracy: 0.0773\n",
+      "Epoch 3013/5000\n",
+      "919/919 - 3s - loss: 1.3196 - accuracy: 0.5827 - val_loss: 3.8003 - val_accuracy: 0.0780\n",
+      "Epoch 3014/5000\n",
+      "919/919 - 3s - loss: 1.3175 - accuracy: 0.5863 - val_loss: 3.7965 - val_accuracy: 0.0779\n",
+      "Epoch 3015/5000\n",
+      "919/919 - 3s - loss: 1.3246 - accuracy: 0.5816 - val_loss: 3.7967 - val_accuracy: 0.0784\n",
+      "Epoch 3016/5000\n",
+      "919/919 - 3s - loss: 1.3257 - accuracy: 0.5846 - val_loss: 3.8008 - val_accuracy: 0.0784\n",
+      "Epoch 3017/5000\n",
+      "919/919 - 3s - loss: 1.3288 - accuracy: 0.5838 - val_loss: 3.8015 - val_accuracy: 0.0785\n",
+      "Epoch 3018/5000\n",
+      "919/919 - 3s - loss: 1.3689 - accuracy: 0.5825 - val_loss: 3.7958 - val_accuracy: 0.0781\n",
+      "Epoch 3019/5000\n",
+      "919/919 - 3s - loss: 1.3122 - accuracy: 0.5859 - val_loss: 3.7920 - val_accuracy: 0.0783\n",
+      "Epoch 3020/5000\n",
+      "919/919 - 3s - loss: 1.3263 - accuracy: 0.5864 - val_loss: 3.7997 - val_accuracy: 0.0781\n",
+      "Epoch 3021/5000\n",
+      "919/919 - 3s - loss: 1.3361 - accuracy: 0.5798 - val_loss: 3.8083 - val_accuracy: 0.0782\n",
+      "Epoch 3022/5000\n",
+      "919/919 - 3s - loss: 1.3470 - accuracy: 0.5848 - val_loss: 3.8078 - val_accuracy: 0.0782\n",
+      "Epoch 3023/5000\n",
+      "919/919 - 3s - loss: 1.3188 - accuracy: 0.5847 - val_loss: 3.8024 - val_accuracy: 0.0785\n",
+      "Epoch 3024/5000\n",
+      "919/919 - 3s - loss: 1.3483 - accuracy: 0.5867 - val_loss: 3.7945 - val_accuracy: 0.0785\n",
+      "Epoch 3025/5000\n",
+      "919/919 - 3s - loss: 1.3467 - accuracy: 0.5819 - val_loss: 3.7969 - val_accuracy: 0.0786\n",
+      "Epoch 3026/5000\n",
+      "919/919 - 3s - loss: 1.3353 - accuracy: 0.5820 - val_loss: 3.7949 - val_accuracy: 0.0782\n",
+      "Epoch 3027/5000\n",
+      "919/919 - 3s - loss: 1.3412 - accuracy: 0.5802 - val_loss: 3.7971 - val_accuracy: 0.0781\n",
+      "Epoch 3028/5000\n",
+      "919/919 - 3s - loss: 1.3121 - accuracy: 0.5841 - val_loss: 3.8029 - val_accuracy: 0.0785\n",
+      "Epoch 3029/5000\n",
+      "919/919 - 3s - loss: 1.3241 - accuracy: 0.5800 - val_loss: 3.8044 - val_accuracy: 0.0782\n",
+      "Epoch 3030/5000\n",
+      "919/919 - 3s - loss: 1.3468 - accuracy: 0.5814 - val_loss: 3.8004 - val_accuracy: 0.0777\n",
+      "Epoch 3031/5000\n",
+      "919/919 - 3s - loss: 1.3999 - accuracy: 0.5842 - val_loss: 3.7988 - val_accuracy: 0.0777\n",
+      "Epoch 3032/5000\n",
+      "919/919 - 3s - loss: 1.3599 - accuracy: 0.5828 - val_loss: 3.8101 - val_accuracy: 0.0783\n",
+      "Epoch 3033/5000\n",
+      "919/919 - 3s - loss: 1.3205 - accuracy: 0.5878 - val_loss: 3.7955 - val_accuracy: 0.0786\n",
+      "Epoch 3034/5000\n",
+      "919/919 - 3s - loss: 1.3677 - accuracy: 0.5816 - val_loss: 3.7978 - val_accuracy: 0.0785\n",
+      "Epoch 3035/5000\n",
+      "919/919 - 3s - loss: 1.3228 - accuracy: 0.5846 - val_loss: 3.8149 - val_accuracy: 0.0782\n",
+      "Epoch 3036/5000\n",
+      "919/919 - 3s - loss: 1.3262 - accuracy: 0.5821 - val_loss: 3.8147 - val_accuracy: 0.0781\n",
+      "Epoch 3037/5000\n",
+      "919/919 - 3s - loss: 1.3421 - accuracy: 0.5887 - val_loss: 3.8163 - val_accuracy: 0.0776\n",
+      "Epoch 3038/5000\n",
+      "919/919 - 3s - loss: 1.3163 - accuracy: 0.5853 - val_loss: 3.8141 - val_accuracy: 0.0781\n",
+      "Epoch 3039/5000\n",
+      "919/919 - 3s - loss: 1.3160 - accuracy: 0.5846 - val_loss: 3.8138 - val_accuracy: 0.0782\n",
+      "Epoch 3040/5000\n",
+      "919/919 - 3s - loss: 1.3502 - accuracy: 0.5847 - val_loss: 3.8182 - val_accuracy: 0.0785\n",
+      "Epoch 3041/5000\n",
+      "919/919 - 3s - loss: 1.3259 - accuracy: 0.5867 - val_loss: 3.8192 - val_accuracy: 0.0793\n",
+      "Epoch 3042/5000\n",
+      "919/919 - 3s - loss: 1.3102 - accuracy: 0.5867 - val_loss: 3.8176 - val_accuracy: 0.0788\n",
+      "Epoch 3043/5000\n",
+      "919/919 - 3s - loss: 1.3391 - accuracy: 0.5848 - val_loss: 3.8128 - val_accuracy: 0.0782\n",
+      "Epoch 3044/5000\n",
+      "919/919 - 3s - loss: 1.3558 - accuracy: 0.5840 - val_loss: 3.8034 - val_accuracy: 0.0778\n",
+      "Epoch 3045/5000\n",
+      "919/919 - 3s - loss: 1.3067 - accuracy: 0.5867 - val_loss: 3.7991 - val_accuracy: 0.0776\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3046/5000\n",
+      "919/919 - 3s - loss: 1.3289 - accuracy: 0.5819 - val_loss: 3.8050 - val_accuracy: 0.0783\n",
+      "Epoch 3047/5000\n",
+      "919/919 - 3s - loss: 1.3370 - accuracy: 0.5827 - val_loss: 3.8074 - val_accuracy: 0.0779\n",
+      "Epoch 3048/5000\n",
+      "919/919 - 3s - loss: 1.3679 - accuracy: 0.5846 - val_loss: 3.8087 - val_accuracy: 0.0780\n",
+      "Epoch 3049/5000\n",
+      "919/919 - 3s - loss: 1.3113 - accuracy: 0.5825 - val_loss: 3.8029 - val_accuracy: 0.0787\n",
+      "Epoch 3050/5000\n",
+      "919/919 - 3s - loss: 1.3095 - accuracy: 0.5849 - val_loss: 3.8216 - val_accuracy: 0.0782\n",
+      "Epoch 3051/5000\n",
+      "919/919 - 3s - loss: 1.3337 - accuracy: 0.5831 - val_loss: 3.8164 - val_accuracy: 0.0785\n",
+      "Epoch 3052/5000\n",
+      "919/919 - 3s - loss: 1.3172 - accuracy: 0.5818 - val_loss: 3.8020 - val_accuracy: 0.0785\n",
+      "Epoch 3053/5000\n",
+      "919/919 - 3s - loss: 1.3140 - accuracy: 0.5847 - val_loss: 3.8051 - val_accuracy: 0.0780\n",
+      "Epoch 3054/5000\n",
+      "919/919 - 3s - loss: 1.3192 - accuracy: 0.5829 - val_loss: 3.8180 - val_accuracy: 0.0785\n",
+      "Epoch 3055/5000\n",
+      "919/919 - 3s - loss: 1.3226 - accuracy: 0.5844 - val_loss: 3.8168 - val_accuracy: 0.0783\n",
+      "Epoch 3056/5000\n",
+      "919/919 - 3s - loss: 1.3359 - accuracy: 0.5833 - val_loss: 3.8132 - val_accuracy: 0.0781\n",
+      "Epoch 3057/5000\n",
+      "919/919 - 3s - loss: 1.3197 - accuracy: 0.5839 - val_loss: 3.8201 - val_accuracy: 0.0777\n",
+      "Epoch 3058/5000\n",
+      "919/919 - 3s - loss: 1.3307 - accuracy: 0.5818 - val_loss: 3.8269 - val_accuracy: 0.0777\n",
+      "Epoch 3059/5000\n",
+      "919/919 - 3s - loss: 1.3223 - accuracy: 0.5882 - val_loss: 3.8341 - val_accuracy: 0.0772\n",
+      "Epoch 3060/5000\n",
+      "919/919 - 3s - loss: 1.3654 - accuracy: 0.5846 - val_loss: 3.8271 - val_accuracy: 0.0780\n",
+      "Epoch 3061/5000\n",
+      "919/919 - 3s - loss: 1.3390 - accuracy: 0.5873 - val_loss: 3.8178 - val_accuracy: 0.0784\n",
+      "Epoch 3062/5000\n",
+      "919/919 - 3s - loss: 1.3325 - accuracy: 0.5824 - val_loss: 3.8181 - val_accuracy: 0.0780\n",
+      "Epoch 3063/5000\n",
+      "919/919 - 3s - loss: 1.3441 - accuracy: 0.5816 - val_loss: 3.8123 - val_accuracy: 0.0785\n",
+      "Epoch 3064/5000\n",
+      "919/919 - 3s - loss: 1.3237 - accuracy: 0.5841 - val_loss: 3.8122 - val_accuracy: 0.0785\n",
+      "Epoch 3065/5000\n",
+      "919/919 - 3s - loss: 1.3161 - accuracy: 0.5839 - val_loss: 3.8109 - val_accuracy: 0.0792\n",
+      "Epoch 3066/5000\n",
+      "919/919 - 3s - loss: 1.3585 - accuracy: 0.5835 - val_loss: 3.8102 - val_accuracy: 0.0785\n",
+      "Epoch 3067/5000\n",
+      "919/919 - 3s - loss: 1.3141 - accuracy: 0.5871 - val_loss: 3.8225 - val_accuracy: 0.0785\n",
+      "Epoch 3068/5000\n",
+      "919/919 - 3s - loss: 1.3690 - accuracy: 0.5844 - val_loss: 3.8139 - val_accuracy: 0.0785\n",
+      "Epoch 3069/5000\n",
+      "919/919 - 3s - loss: 1.3272 - accuracy: 0.5861 - val_loss: 3.8139 - val_accuracy: 0.0787\n",
+      "Epoch 3070/5000\n",
+      "919/919 - 3s - loss: 1.3203 - accuracy: 0.5848 - val_loss: 3.8227 - val_accuracy: 0.0783\n",
+      "Epoch 3071/5000\n",
+      "919/919 - 3s - loss: 1.3217 - accuracy: 0.5859 - val_loss: 3.8188 - val_accuracy: 0.0784\n",
+      "Epoch 3072/5000\n",
+      "919/919 - 3s - loss: 1.3097 - accuracy: 0.5831 - val_loss: 3.8166 - val_accuracy: 0.0790\n",
+      "Epoch 3073/5000\n",
+      "919/919 - 3s - loss: 1.3314 - accuracy: 0.5839 - val_loss: 3.8175 - val_accuracy: 0.0793\n",
+      "Epoch 3074/5000\n",
+      "919/919 - 3s - loss: 1.5725 - accuracy: 0.5785 - val_loss: 3.8119 - val_accuracy: 0.0800\n",
+      "Epoch 3075/5000\n",
+      "919/919 - 3s - loss: 1.3142 - accuracy: 0.5860 - val_loss: 3.8193 - val_accuracy: 0.0792\n",
+      "Epoch 3076/5000\n",
+      "919/919 - 3s - loss: 1.4264 - accuracy: 0.5861 - val_loss: 3.8220 - val_accuracy: 0.0798\n",
+      "Epoch 3077/5000\n",
+      "919/919 - 3s - loss: 1.3323 - accuracy: 0.5816 - val_loss: 3.8111 - val_accuracy: 0.0794\n",
+      "Epoch 3078/5000\n",
+      "919/919 - 3s - loss: 1.3122 - accuracy: 0.5887 - val_loss: 3.8191 - val_accuracy: 0.0794\n",
+      "Epoch 3079/5000\n",
+      "919/919 - 3s - loss: 1.3061 - accuracy: 0.5864 - val_loss: 3.8218 - val_accuracy: 0.0793\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3080/5000\n",
+      "919/919 - 3s - loss: 1.3072 - accuracy: 0.5897 - val_loss: 3.8153 - val_accuracy: 0.0792\n",
+      "Epoch 3081/5000\n",
+      "919/919 - 3s - loss: 1.3167 - accuracy: 0.5831 - val_loss: 3.8025 - val_accuracy: 0.0793\n",
+      "Epoch 3082/5000\n",
+      "919/919 - 3s - loss: 1.3301 - accuracy: 0.5895 - val_loss: 3.8057 - val_accuracy: 0.0786\n",
+      "Epoch 3083/5000\n",
+      "919/919 - 3s - loss: 1.2985 - accuracy: 0.5871 - val_loss: 3.8113 - val_accuracy: 0.0788\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3084/5000\n",
+      "919/919 - 3s - loss: 1.3202 - accuracy: 0.5880 - val_loss: 3.8149 - val_accuracy: 0.0782\n",
+      "Epoch 3085/5000\n",
+      "919/919 - 3s - loss: 1.4723 - accuracy: 0.5849 - val_loss: 3.8107 - val_accuracy: 0.0793\n",
+      "Epoch 3086/5000\n",
+      "919/919 - 3s - loss: 1.4215 - accuracy: 0.5847 - val_loss: 3.8090 - val_accuracy: 0.0794\n",
+      "Epoch 3087/5000\n",
+      "919/919 - 3s - loss: 1.2980 - accuracy: 0.5886 - val_loss: 3.8135 - val_accuracy: 0.0800\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3088/5000\n",
+      "919/919 - 3s - loss: 1.3252 - accuracy: 0.5861 - val_loss: 3.8188 - val_accuracy: 0.0794\n",
+      "Epoch 3089/5000\n",
+      "919/919 - 3s - loss: 1.3167 - accuracy: 0.5850 - val_loss: 3.8161 - val_accuracy: 0.0794\n",
+      "Epoch 3090/5000\n",
+      "919/919 - 3s - loss: 1.3150 - accuracy: 0.5865 - val_loss: 3.8202 - val_accuracy: 0.0794\n",
+      "Epoch 3091/5000\n",
+      "919/919 - 3s - loss: 1.3626 - accuracy: 0.5846 - val_loss: 3.8182 - val_accuracy: 0.0794\n",
+      "Epoch 3092/5000\n",
+      "919/919 - 3s - loss: 1.3015 - accuracy: 0.5863 - val_loss: 3.8266 - val_accuracy: 0.0798\n",
+      "Epoch 3093/5000\n",
+      "919/919 - 3s - loss: 1.2962 - accuracy: 0.5918 - val_loss: 3.8324 - val_accuracy: 0.0801\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3094/5000\n",
+      "919/919 - 3s - loss: 1.3281 - accuracy: 0.5839 - val_loss: 3.8181 - val_accuracy: 0.0801\n",
+      "Epoch 3095/5000\n",
+      "919/919 - 3s - loss: 1.3077 - accuracy: 0.5841 - val_loss: 3.8262 - val_accuracy: 0.0803\n",
+      "Epoch 3096/5000\n",
+      "919/919 - 3s - loss: 1.3286 - accuracy: 0.5838 - val_loss: 3.8225 - val_accuracy: 0.0798\n",
+      "Epoch 3097/5000\n",
+      "919/919 - 3s - loss: 1.3240 - accuracy: 0.5884 - val_loss: 3.8215 - val_accuracy: 0.0798\n",
+      "Epoch 3098/5000\n",
+      "919/919 - 3s - loss: 1.3213 - accuracy: 0.5869 - val_loss: 3.8221 - val_accuracy: 0.0798\n",
+      "Epoch 3099/5000\n",
+      "919/919 - 3s - loss: 1.3110 - accuracy: 0.5861 - val_loss: 3.8095 - val_accuracy: 0.0794\n",
+      "Epoch 3100/5000\n",
+      "919/919 - 3s - loss: 1.3008 - accuracy: 0.5884 - val_loss: 3.8081 - val_accuracy: 0.0800\n",
+      "Epoch 3101/5000\n",
+      "919/919 - 3s - loss: 1.4682 - accuracy: 0.5881 - val_loss: 3.8142 - val_accuracy: 0.0791\n",
+      "Epoch 3102/5000\n",
+      "919/919 - 3s - loss: 1.3890 - accuracy: 0.5873 - val_loss: 3.8060 - val_accuracy: 0.0796\n",
+      "Epoch 3103/5000\n",
+      "919/919 - 3s - loss: 1.3066 - accuracy: 0.5858 - val_loss: 3.8214 - val_accuracy: 0.0797\n",
+      "Epoch 3104/5000\n",
+      "919/919 - 3s - loss: 1.3228 - accuracy: 0.5907 - val_loss: 3.8134 - val_accuracy: 0.0801\n",
+      "Epoch 3105/5000\n",
+      "919/919 - 3s - loss: 1.3233 - accuracy: 0.5883 - val_loss: 3.8001 - val_accuracy: 0.0794\n",
+      "Epoch 3106/5000\n",
+      "919/919 - 3s - loss: 1.2987 - accuracy: 0.5865 - val_loss: 3.8062 - val_accuracy: 0.0787\n",
+      "Epoch 3107/5000\n",
+      "919/919 - 3s - loss: 1.3120 - accuracy: 0.5877 - val_loss: 3.8115 - val_accuracy: 0.0803\n",
+      "Epoch 3108/5000\n",
+      "919/919 - 3s - loss: 1.3251 - accuracy: 0.5854 - val_loss: 3.8047 - val_accuracy: 0.0792\n",
+      "Epoch 3109/5000\n",
+      "919/919 - 3s - loss: 1.4006 - accuracy: 0.5860 - val_loss: 3.8042 - val_accuracy: 0.0790\n",
+      "Epoch 3110/5000\n",
+      "919/919 - 3s - loss: 1.5447 - accuracy: 0.5840 - val_loss: 3.8138 - val_accuracy: 0.0791\n",
+      "Epoch 3111/5000\n",
+      "919/919 - 3s - loss: 1.4086 - accuracy: 0.5839 - val_loss: 3.8046 - val_accuracy: 0.0787\n",
+      "Epoch 3112/5000\n",
+      "919/919 - 3s - loss: 1.3124 - accuracy: 0.5846 - val_loss: 3.8004 - val_accuracy: 0.0791\n",
+      "Epoch 3113/5000\n",
+      "919/919 - 3s - loss: 1.3349 - accuracy: 0.5901 - val_loss: 3.8038 - val_accuracy: 0.0797\n",
+      "Epoch 3114/5000\n",
+      "919/919 - 3s - loss: 1.3065 - accuracy: 0.5867 - val_loss: 3.8014 - val_accuracy: 0.0794\n",
+      "Epoch 3115/5000\n",
+      "919/919 - 3s - loss: 1.3062 - accuracy: 0.5886 - val_loss: 3.8098 - val_accuracy: 0.0794\n",
+      "Epoch 3116/5000\n",
+      "919/919 - 3s - loss: 1.3498 - accuracy: 0.5890 - val_loss: 3.8170 - val_accuracy: 0.0795\n",
+      "Epoch 3117/5000\n",
+      "919/919 - 3s - loss: 1.3105 - accuracy: 0.5863 - val_loss: 3.8114 - val_accuracy: 0.0799\n",
+      "Epoch 3118/5000\n",
+      "919/919 - 3s - loss: 1.3141 - accuracy: 0.5863 - val_loss: 3.8146 - val_accuracy: 0.0797\n",
+      "Epoch 3119/5000\n",
+      "919/919 - 3s - loss: 1.2885 - accuracy: 0.5877 - val_loss: 3.8134 - val_accuracy: 0.0797\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3120/5000\n",
+      "919/919 - 3s - loss: 1.3261 - accuracy: 0.5894 - val_loss: 3.8168 - val_accuracy: 0.0797\n",
+      "Epoch 3121/5000\n",
+      "919/919 - 3s - loss: 1.2996 - accuracy: 0.5912 - val_loss: 3.8200 - val_accuracy: 0.0802\n",
+      "Epoch 3122/5000\n",
+      "919/919 - 3s - loss: 1.3618 - accuracy: 0.5877 - val_loss: 3.8245 - val_accuracy: 0.0794\n",
+      "Epoch 3123/5000\n",
+      "919/919 - 3s - loss: 1.3056 - accuracy: 0.5880 - val_loss: 3.8202 - val_accuracy: 0.0793\n",
+      "Epoch 3124/5000\n",
+      "919/919 - 3s - loss: 1.3078 - accuracy: 0.5902 - val_loss: 3.8228 - val_accuracy: 0.0793\n",
+      "Epoch 3125/5000\n",
+      "919/919 - 3s - loss: 1.3082 - accuracy: 0.5883 - val_loss: 3.8213 - val_accuracy: 0.0803\n",
+      "Epoch 3126/5000\n",
+      "919/919 - 3s - loss: 1.3113 - accuracy: 0.5886 - val_loss: 3.8174 - val_accuracy: 0.0804\n",
+      "Epoch 3127/5000\n",
+      "919/919 - 3s - loss: 1.3015 - accuracy: 0.5895 - val_loss: 3.8211 - val_accuracy: 0.0805\n",
+      "Epoch 3128/5000\n",
+      "919/919 - 3s - loss: 1.3097 - accuracy: 0.5883 - val_loss: 3.8172 - val_accuracy: 0.0807\n",
+      "Epoch 3129/5000\n",
+      "919/919 - 3s - loss: 1.3032 - accuracy: 0.5939 - val_loss: 3.8174 - val_accuracy: 0.0803\n",
+      "Epoch 3130/5000\n",
+      "919/919 - 3s - loss: 1.3804 - accuracy: 0.5826 - val_loss: 3.8284 - val_accuracy: 0.0799\n",
+      "Epoch 3131/5000\n",
+      "919/919 - 3s - loss: 1.3405 - accuracy: 0.5905 - val_loss: 3.8206 - val_accuracy: 0.0802\n",
+      "Epoch 3132/5000\n",
+      "919/919 - 3s - loss: 1.3097 - accuracy: 0.5887 - val_loss: 3.8249 - val_accuracy: 0.0800\n",
+      "Epoch 3133/5000\n",
+      "919/919 - 3s - loss: 1.3142 - accuracy: 0.5849 - val_loss: 3.8381 - val_accuracy: 0.0806\n",
+      "Epoch 3134/5000\n",
+      "919/919 - 3s - loss: 1.3598 - accuracy: 0.5909 - val_loss: 3.8360 - val_accuracy: 0.0806\n",
+      "Epoch 3135/5000\n",
+      "919/919 - 3s - loss: 1.3274 - accuracy: 0.5902 - val_loss: 3.8280 - val_accuracy: 0.0798\n",
+      "Epoch 3136/5000\n",
+      "919/919 - 3s - loss: 1.3590 - accuracy: 0.5905 - val_loss: 3.8152 - val_accuracy: 0.0800\n",
+      "Epoch 3137/5000\n",
+      "919/919 - 3s - loss: 1.3116 - accuracy: 0.5859 - val_loss: 3.8176 - val_accuracy: 0.0792\n",
+      "Epoch 3138/5000\n",
+      "919/919 - 3s - loss: 1.2851 - accuracy: 0.5943 - val_loss: 3.8175 - val_accuracy: 0.0801\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3139/5000\n",
+      "919/919 - 3s - loss: 1.3018 - accuracy: 0.5905 - val_loss: 3.8174 - val_accuracy: 0.0799\n",
+      "Epoch 3140/5000\n",
+      "919/919 - 3s - loss: 1.3070 - accuracy: 0.5864 - val_loss: 3.8101 - val_accuracy: 0.0802\n",
+      "Epoch 3141/5000\n",
+      "919/919 - 3s - loss: 1.4030 - accuracy: 0.5881 - val_loss: 3.8167 - val_accuracy: 0.0813\n",
+      "Epoch 3142/5000\n",
+      "919/919 - 3s - loss: 1.3071 - accuracy: 0.5882 - val_loss: 3.8171 - val_accuracy: 0.0809\n",
+      "Epoch 3143/5000\n",
+      "919/919 - 3s - loss: 1.4282 - accuracy: 0.5935 - val_loss: 3.8176 - val_accuracy: 0.0811\n",
+      "Epoch 3144/5000\n",
+      "919/919 - 3s - loss: 1.3257 - accuracy: 0.5906 - val_loss: 3.8145 - val_accuracy: 0.0812\n",
+      "Epoch 3145/5000\n",
+      "919/919 - 3s - loss: 1.3060 - accuracy: 0.5885 - val_loss: 3.8112 - val_accuracy: 0.0808\n",
+      "Epoch 3146/5000\n",
+      "919/919 - 3s - loss: 1.2954 - accuracy: 0.5919 - val_loss: 3.8191 - val_accuracy: 0.0804\n",
+      "Epoch 3147/5000\n",
+      "919/919 - 3s - loss: 1.3146 - accuracy: 0.5895 - val_loss: 3.8181 - val_accuracy: 0.0797\n",
+      "Epoch 3148/5000\n",
+      "919/919 - 3s - loss: 1.2972 - accuracy: 0.5856 - val_loss: 3.8396 - val_accuracy: 0.0800\n",
+      "Epoch 3149/5000\n",
+      "919/919 - 3s - loss: 1.3126 - accuracy: 0.5893 - val_loss: 3.8374 - val_accuracy: 0.0806\n",
+      "Epoch 3150/5000\n",
+      "919/919 - 3s - loss: 1.3119 - accuracy: 0.5863 - val_loss: 3.8348 - val_accuracy: 0.0805\n",
+      "Epoch 3151/5000\n",
+      "919/919 - 3s - loss: 1.3130 - accuracy: 0.5829 - val_loss: 3.8315 - val_accuracy: 0.0803\n",
+      "Epoch 3152/5000\n",
+      "919/919 - 3s - loss: 1.3074 - accuracy: 0.5855 - val_loss: 3.8457 - val_accuracy: 0.0801\n",
+      "Epoch 3153/5000\n",
+      "919/919 - 3s - loss: 1.3954 - accuracy: 0.5861 - val_loss: 3.8403 - val_accuracy: 0.0798\n",
+      "Epoch 3154/5000\n",
+      "919/919 - 3s - loss: 1.3086 - accuracy: 0.5888 - val_loss: 3.8312 - val_accuracy: 0.0803\n",
+      "Epoch 3155/5000\n",
+      "919/919 - 3s - loss: 1.3038 - accuracy: 0.5884 - val_loss: 3.8291 - val_accuracy: 0.0807\n",
+      "Epoch 3156/5000\n",
+      "919/919 - 3s - loss: 1.3054 - accuracy: 0.5908 - val_loss: 3.8365 - val_accuracy: 0.0813\n",
+      "Epoch 3157/5000\n",
+      "919/919 - 3s - loss: 1.2997 - accuracy: 0.5877 - val_loss: 3.8406 - val_accuracy: 0.0812\n",
+      "Epoch 3158/5000\n",
+      "919/919 - 3s - loss: 1.3195 - accuracy: 0.5887 - val_loss: 3.8295 - val_accuracy: 0.0812\n",
+      "Epoch 3159/5000\n",
+      "919/919 - 3s - loss: 1.3080 - accuracy: 0.5891 - val_loss: 3.8254 - val_accuracy: 0.0811\n",
+      "Epoch 3160/5000\n",
+      "919/919 - 3s - loss: 1.2977 - accuracy: 0.5910 - val_loss: 3.8234 - val_accuracy: 0.0816\n",
+      "Epoch 3161/5000\n",
+      "919/919 - 3s - loss: 1.3079 - accuracy: 0.5891 - val_loss: 3.8279 - val_accuracy: 0.0818\n",
+      "Epoch 3162/5000\n",
+      "919/919 - 3s - loss: 1.3049 - accuracy: 0.5896 - val_loss: 3.8399 - val_accuracy: 0.0805\n",
+      "Epoch 3163/5000\n",
+      "919/919 - 3s - loss: 1.3074 - accuracy: 0.5914 - val_loss: 3.8404 - val_accuracy: 0.0812\n",
+      "Epoch 3164/5000\n",
+      "919/919 - 3s - loss: 1.3105 - accuracy: 0.5946 - val_loss: 3.8338 - val_accuracy: 0.0816\n",
+      "Epoch 3165/5000\n",
+      "919/919 - 3s - loss: 1.3023 - accuracy: 0.5926 - val_loss: 3.8386 - val_accuracy: 0.0814\n",
+      "Epoch 3166/5000\n",
+      "919/919 - 3s - loss: 1.2895 - accuracy: 0.5931 - val_loss: 3.8523 - val_accuracy: 0.0812\n",
+      "Epoch 3167/5000\n",
+      "919/919 - 3s - loss: 1.3189 - accuracy: 0.5911 - val_loss: 3.8345 - val_accuracy: 0.0808\n",
+      "Epoch 3168/5000\n",
+      "919/919 - 3s - loss: 1.3210 - accuracy: 0.5944 - val_loss: 3.8398 - val_accuracy: 0.0808\n",
+      "Epoch 3169/5000\n",
+      "919/919 - 3s - loss: 1.3368 - accuracy: 0.5867 - val_loss: 3.8375 - val_accuracy: 0.0808\n",
+      "Epoch 3170/5000\n",
+      "919/919 - 3s - loss: 1.2988 - accuracy: 0.5929 - val_loss: 3.8319 - val_accuracy: 0.0801\n",
+      "Epoch 3171/5000\n",
+      "919/919 - 3s - loss: 1.3042 - accuracy: 0.5894 - val_loss: 3.8186 - val_accuracy: 0.0796\n",
+      "Epoch 3172/5000\n",
+      "919/919 - 3s - loss: 1.3645 - accuracy: 0.5914 - val_loss: 3.8197 - val_accuracy: 0.0800\n",
+      "Epoch 3173/5000\n",
+      "919/919 - 3s - loss: 1.3158 - accuracy: 0.5906 - val_loss: 3.8255 - val_accuracy: 0.0811\n",
+      "Epoch 3174/5000\n",
+      "919/919 - 3s - loss: 1.2986 - accuracy: 0.5933 - val_loss: 3.8311 - val_accuracy: 0.0794\n",
+      "Epoch 3175/5000\n",
+      "919/919 - 3s - loss: 1.3025 - accuracy: 0.5911 - val_loss: 3.8394 - val_accuracy: 0.0795\n",
+      "Epoch 3176/5000\n",
+      "919/919 - 3s - loss: 1.3478 - accuracy: 0.5865 - val_loss: 3.8318 - val_accuracy: 0.0799\n",
+      "Epoch 3177/5000\n",
+      "919/919 - 3s - loss: 1.3289 - accuracy: 0.5886 - val_loss: 3.8214 - val_accuracy: 0.0798\n",
+      "Epoch 3178/5000\n",
+      "919/919 - 3s - loss: 1.3099 - accuracy: 0.5908 - val_loss: 3.8191 - val_accuracy: 0.0796\n",
+      "Epoch 3179/5000\n",
+      "919/919 - 3s - loss: 1.3462 - accuracy: 0.5883 - val_loss: 3.8147 - val_accuracy: 0.0802\n",
+      "Epoch 3180/5000\n",
+      "919/919 - 3s - loss: 1.3377 - accuracy: 0.5864 - val_loss: 3.8205 - val_accuracy: 0.0797\n",
+      "Epoch 3181/5000\n",
+      "919/919 - 3s - loss: 1.2896 - accuracy: 0.5916 - val_loss: 3.8274 - val_accuracy: 0.0805\n",
+      "Epoch 3182/5000\n",
+      "919/919 - 3s - loss: 1.3045 - accuracy: 0.5879 - val_loss: 3.8286 - val_accuracy: 0.0806\n",
+      "Epoch 3183/5000\n",
+      "919/919 - 3s - loss: 1.3209 - accuracy: 0.5891 - val_loss: 3.8306 - val_accuracy: 0.0802\n",
+      "Epoch 3184/5000\n",
+      "919/919 - 3s - loss: 1.3100 - accuracy: 0.5873 - val_loss: 3.8256 - val_accuracy: 0.0808\n",
+      "Epoch 3185/5000\n",
+      "919/919 - 3s - loss: 1.3095 - accuracy: 0.5890 - val_loss: 3.8198 - val_accuracy: 0.0816\n",
+      "Epoch 3186/5000\n",
+      "919/919 - 3s - loss: 1.3010 - accuracy: 0.5918 - val_loss: 3.8297 - val_accuracy: 0.0811\n",
+      "Epoch 3187/5000\n",
+      "919/919 - 3s - loss: 1.4176 - accuracy: 0.5882 - val_loss: 3.8328 - val_accuracy: 0.0809\n",
+      "Epoch 3188/5000\n",
+      "919/919 - 3s - loss: 1.3396 - accuracy: 0.5905 - val_loss: 3.8264 - val_accuracy: 0.0809\n",
+      "Epoch 3189/5000\n",
+      "919/919 - 3s - loss: 1.2958 - accuracy: 0.5922 - val_loss: 3.8207 - val_accuracy: 0.0811\n",
+      "Epoch 3190/5000\n",
+      "919/919 - 3s - loss: 1.3196 - accuracy: 0.5882 - val_loss: 3.8174 - val_accuracy: 0.0812\n",
+      "Epoch 3191/5000\n",
+      "919/919 - 3s - loss: 1.3058 - accuracy: 0.5930 - val_loss: 3.8155 - val_accuracy: 0.0817\n",
+      "Epoch 3192/5000\n",
+      "919/919 - 3s - loss: 1.2860 - accuracy: 0.5935 - val_loss: 3.8234 - val_accuracy: 0.0812\n",
+      "Epoch 3193/5000\n",
+      "919/919 - 3s - loss: 1.3206 - accuracy: 0.5911 - val_loss: 3.8280 - val_accuracy: 0.0814\n",
+      "Epoch 3194/5000\n",
+      "919/919 - 3s - loss: 1.2849 - accuracy: 0.5931 - val_loss: 3.8298 - val_accuracy: 0.0814\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3195/5000\n",
+      "919/919 - 3s - loss: 1.2962 - accuracy: 0.5897 - val_loss: 3.8389 - val_accuracy: 0.0812\n",
+      "Epoch 3196/5000\n",
+      "919/919 - 3s - loss: 1.2862 - accuracy: 0.5939 - val_loss: 3.8363 - val_accuracy: 0.0810\n",
+      "Epoch 3197/5000\n",
+      "919/919 - 3s - loss: 1.5874 - accuracy: 0.5920 - val_loss: 3.8396 - val_accuracy: 0.0817\n",
+      "Epoch 3198/5000\n",
+      "919/919 - 3s - loss: 1.2947 - accuracy: 0.5928 - val_loss: 3.8494 - val_accuracy: 0.0813\n",
+      "Epoch 3199/5000\n",
+      "919/919 - 3s - loss: 1.2881 - accuracy: 0.5937 - val_loss: 3.8354 - val_accuracy: 0.0822\n",
+      "Epoch 3200/5000\n",
+      "919/919 - 3s - loss: 1.4790 - accuracy: 0.5915 - val_loss: 3.8429 - val_accuracy: 0.0818\n",
+      "Epoch 3201/5000\n",
+      "919/919 - 3s - loss: 1.3494 - accuracy: 0.5927 - val_loss: 3.8356 - val_accuracy: 0.0817\n",
+      "Epoch 3202/5000\n",
+      "919/919 - 3s - loss: 1.3400 - accuracy: 0.5901 - val_loss: 3.8233 - val_accuracy: 0.0817\n",
+      "Epoch 3203/5000\n",
+      "919/919 - 3s - loss: 1.3575 - accuracy: 0.5935 - val_loss: 3.8292 - val_accuracy: 0.0821\n",
+      "Epoch 3204/5000\n",
+      "919/919 - 3s - loss: 1.2894 - accuracy: 0.5952 - val_loss: 3.8358 - val_accuracy: 0.0814\n",
+      "Epoch 3205/5000\n",
+      "919/919 - 3s - loss: 1.3033 - accuracy: 0.5931 - val_loss: 3.8336 - val_accuracy: 0.0814\n",
+      "Epoch 3206/5000\n",
+      "919/919 - 3s - loss: 1.2866 - accuracy: 0.5950 - val_loss: 3.8354 - val_accuracy: 0.0819\n",
+      "Epoch 3207/5000\n",
+      "919/919 - 3s - loss: 1.3063 - accuracy: 0.5913 - val_loss: 3.8280 - val_accuracy: 0.0820\n",
+      "Epoch 3208/5000\n",
+      "919/919 - 3s - loss: 1.2948 - accuracy: 0.5911 - val_loss: 3.8241 - val_accuracy: 0.0823\n",
+      "Epoch 3209/5000\n",
+      "919/919 - 3s - loss: 1.3207 - accuracy: 0.5894 - val_loss: 3.8169 - val_accuracy: 0.0819\n",
+      "Epoch 3210/5000\n",
+      "919/919 - 3s - loss: 1.3357 - accuracy: 0.5897 - val_loss: 3.8292 - val_accuracy: 0.0813\n",
+      "Epoch 3211/5000\n",
+      "919/919 - 3s - loss: 1.3142 - accuracy: 0.5934 - val_loss: 3.8182 - val_accuracy: 0.0812\n",
+      "Epoch 3212/5000\n",
+      "919/919 - 3s - loss: 1.3223 - accuracy: 0.5900 - val_loss: 3.8218 - val_accuracy: 0.0813\n",
+      "Epoch 3213/5000\n",
+      "919/919 - 3s - loss: 1.3807 - accuracy: 0.5946 - val_loss: 3.8219 - val_accuracy: 0.0815\n",
+      "Epoch 3214/5000\n",
+      "919/919 - 3s - loss: 1.3130 - accuracy: 0.5944 - val_loss: 3.8258 - val_accuracy: 0.0817\n",
+      "Epoch 3215/5000\n",
+      "919/919 - 3s - loss: 1.3016 - accuracy: 0.5937 - val_loss: 3.8332 - val_accuracy: 0.0813\n",
+      "Epoch 3216/5000\n",
+      "919/919 - 3s - loss: 1.2942 - accuracy: 0.5916 - val_loss: 3.8408 - val_accuracy: 0.0817\n",
+      "Epoch 3217/5000\n",
+      "919/919 - 3s - loss: 1.2977 - accuracy: 0.5927 - val_loss: 3.8286 - val_accuracy: 0.0817\n",
+      "Epoch 3218/5000\n",
+      "919/919 - 3s - loss: 1.2779 - accuracy: 0.5961 - val_loss: 3.8303 - val_accuracy: 0.0810\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3219/5000\n",
+      "919/919 - 3s - loss: 1.2906 - accuracy: 0.5974 - val_loss: 3.8314 - val_accuracy: 0.0814\n",
+      "Epoch 3220/5000\n",
+      "919/919 - 3s - loss: 1.2941 - accuracy: 0.5914 - val_loss: 3.8304 - val_accuracy: 0.0813\n",
+      "Epoch 3221/5000\n",
+      "919/919 - 3s - loss: 1.3006 - accuracy: 0.5930 - val_loss: 3.8275 - val_accuracy: 0.0810\n",
+      "Epoch 3222/5000\n",
+      "919/919 - 3s - loss: 1.2913 - accuracy: 0.5937 - val_loss: 3.8333 - val_accuracy: 0.0818\n",
+      "Epoch 3223/5000\n",
+      "919/919 - 3s - loss: 1.2871 - accuracy: 0.5937 - val_loss: 3.8352 - val_accuracy: 0.0816\n",
+      "Epoch 3224/5000\n",
+      "919/919 - 3s - loss: 1.2995 - accuracy: 0.5872 - val_loss: 3.8313 - val_accuracy: 0.0813\n",
+      "Epoch 3225/5000\n",
+      "919/919 - 3s - loss: 1.2850 - accuracy: 0.5964 - val_loss: 3.8306 - val_accuracy: 0.0816\n",
+      "Epoch 3226/5000\n",
+      "919/919 - 3s - loss: 1.3036 - accuracy: 0.5958 - val_loss: 3.8211 - val_accuracy: 0.0815\n",
+      "Epoch 3227/5000\n",
+      "919/919 - 3s - loss: 1.3018 - accuracy: 0.5956 - val_loss: 3.8148 - val_accuracy: 0.0810\n",
+      "Epoch 3228/5000\n",
+      "919/919 - 3s - loss: 1.3488 - accuracy: 0.5917 - val_loss: 3.8174 - val_accuracy: 0.0808\n",
+      "Epoch 3229/5000\n",
+      "919/919 - 3s - loss: 1.3856 - accuracy: 0.5965 - val_loss: 3.8281 - val_accuracy: 0.0807\n",
+      "Epoch 3230/5000\n",
+      "919/919 - 3s - loss: 1.2906 - accuracy: 0.5941 - val_loss: 3.8269 - val_accuracy: 0.0812\n",
+      "Epoch 3231/5000\n",
+      "919/919 - 3s - loss: 1.3027 - accuracy: 0.5884 - val_loss: 3.8106 - val_accuracy: 0.0812\n",
+      "Epoch 3232/5000\n",
+      "919/919 - 3s - loss: 1.3055 - accuracy: 0.5895 - val_loss: 3.8109 - val_accuracy: 0.0809\n",
+      "Epoch 3233/5000\n",
+      "919/919 - 3s - loss: 1.2914 - accuracy: 0.5871 - val_loss: 3.8176 - val_accuracy: 0.0810\n",
+      "Epoch 3234/5000\n",
+      "919/919 - 3s - loss: 1.2871 - accuracy: 0.5902 - val_loss: 3.8208 - val_accuracy: 0.0815\n",
+      "Epoch 3235/5000\n",
+      "919/919 - 3s - loss: 1.3209 - accuracy: 0.5899 - val_loss: 3.8261 - val_accuracy: 0.0814\n",
+      "Epoch 3236/5000\n",
+      "919/919 - 3s - loss: 1.3176 - accuracy: 0.5904 - val_loss: 3.8258 - val_accuracy: 0.0812\n",
+      "Epoch 3237/5000\n",
+      "919/919 - 3s - loss: 1.2932 - accuracy: 0.5930 - val_loss: 3.8168 - val_accuracy: 0.0814\n",
+      "Epoch 3238/5000\n",
+      "919/919 - 3s - loss: 1.3009 - accuracy: 0.5879 - val_loss: 3.8184 - val_accuracy: 0.0811\n",
+      "Epoch 3239/5000\n",
+      "919/919 - 3s - loss: 1.2994 - accuracy: 0.5939 - val_loss: 3.8205 - val_accuracy: 0.0814\n",
+      "Epoch 3240/5000\n",
+      "919/919 - 3s - loss: 1.2972 - accuracy: 0.5939 - val_loss: 3.8217 - val_accuracy: 0.0818\n",
+      "Epoch 3241/5000\n",
+      "919/919 - 3s - loss: 1.2945 - accuracy: 0.5893 - val_loss: 3.8223 - val_accuracy: 0.0814\n",
+      "Epoch 3242/5000\n",
+      "919/919 - 3s - loss: 1.2947 - accuracy: 0.5926 - val_loss: 3.8244 - val_accuracy: 0.0809\n",
+      "Epoch 3243/5000\n",
+      "919/919 - 3s - loss: 1.4228 - accuracy: 0.5888 - val_loss: 3.8198 - val_accuracy: 0.0809\n",
+      "Epoch 3244/5000\n",
+      "919/919 - 3s - loss: 1.3491 - accuracy: 0.5925 - val_loss: 3.8217 - val_accuracy: 0.0809\n",
+      "Epoch 3245/5000\n",
+      "919/919 - 3s - loss: 1.3582 - accuracy: 0.5926 - val_loss: 3.8226 - val_accuracy: 0.0819\n",
+      "Epoch 3246/5000\n",
+      "919/919 - 3s - loss: 1.2775 - accuracy: 0.5956 - val_loss: 3.8242 - val_accuracy: 0.0821\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3247/5000\n",
+      "919/919 - 3s - loss: 1.3463 - accuracy: 0.5909 - val_loss: 3.8223 - val_accuracy: 0.0827\n",
+      "Epoch 3248/5000\n",
+      "919/919 - 3s - loss: 1.3272 - accuracy: 0.5924 - val_loss: 3.8250 - val_accuracy: 0.0817\n",
+      "Epoch 3249/5000\n",
+      "919/919 - 3s - loss: 1.2852 - accuracy: 0.5919 - val_loss: 3.8313 - val_accuracy: 0.0819\n",
+      "Epoch 3250/5000\n",
+      "919/919 - 3s - loss: 1.2919 - accuracy: 0.5934 - val_loss: 3.8326 - val_accuracy: 0.0820\n",
+      "Epoch 3251/5000\n",
+      "919/919 - 3s - loss: 1.4222 - accuracy: 0.5935 - val_loss: 3.8380 - val_accuracy: 0.0818\n",
+      "Epoch 3252/5000\n",
+      "919/919 - 3s - loss: 1.3333 - accuracy: 0.5941 - val_loss: 3.8376 - val_accuracy: 0.0811\n",
+      "Epoch 3253/5000\n",
+      "919/919 - 3s - loss: 1.2965 - accuracy: 0.5951 - val_loss: 3.8278 - val_accuracy: 0.0822\n",
+      "Epoch 3254/5000\n",
+      "919/919 - 3s - loss: 1.2874 - accuracy: 0.5932 - val_loss: 3.8408 - val_accuracy: 0.0815\n",
+      "Epoch 3255/5000\n",
+      "919/919 - 3s - loss: 1.3646 - accuracy: 0.5914 - val_loss: 3.8335 - val_accuracy: 0.0825\n",
+      "Epoch 3256/5000\n",
+      "919/919 - 3s - loss: 1.2954 - accuracy: 0.5901 - val_loss: 3.8343 - val_accuracy: 0.0822\n",
+      "Epoch 3257/5000\n",
+      "919/919 - 3s - loss: 1.3202 - accuracy: 0.5970 - val_loss: 3.8348 - val_accuracy: 0.0821\n",
+      "Epoch 3258/5000\n",
+      "919/919 - 3s - loss: 1.3144 - accuracy: 0.5949 - val_loss: 3.8351 - val_accuracy: 0.0822\n",
+      "Epoch 3259/5000\n",
+      "919/919 - 3s - loss: 1.2793 - accuracy: 0.5939 - val_loss: 3.8483 - val_accuracy: 0.0820\n",
+      "Epoch 3260/5000\n",
+      "919/919 - 3s - loss: 1.2975 - accuracy: 0.5945 - val_loss: 3.8341 - val_accuracy: 0.0820\n",
+      "Epoch 3261/5000\n",
+      "919/919 - 3s - loss: 1.2900 - accuracy: 0.5969 - val_loss: 3.8316 - val_accuracy: 0.0828\n",
+      "Epoch 3262/5000\n",
+      "919/919 - 3s - loss: 1.2969 - accuracy: 0.6000 - val_loss: 3.8198 - val_accuracy: 0.0816\n",
+      "Epoch 3263/5000\n",
+      "919/919 - 3s - loss: 1.3327 - accuracy: 0.5934 - val_loss: 3.8203 - val_accuracy: 0.0813\n",
+      "Epoch 3264/5000\n",
+      "919/919 - 3s - loss: 1.2985 - accuracy: 0.5904 - val_loss: 3.8212 - val_accuracy: 0.0813\n",
+      "Epoch 3265/5000\n",
+      "919/919 - 3s - loss: 1.3025 - accuracy: 0.5911 - val_loss: 3.8296 - val_accuracy: 0.0816\n",
+      "Epoch 3266/5000\n",
+      "919/919 - 3s - loss: 1.4232 - accuracy: 0.5937 - val_loss: 3.8363 - val_accuracy: 0.0821\n",
+      "Epoch 3267/5000\n",
+      "919/919 - 3s - loss: 1.2891 - accuracy: 0.5944 - val_loss: 3.8325 - val_accuracy: 0.0820\n",
+      "Epoch 3268/5000\n",
+      "919/919 - 3s - loss: 1.2843 - accuracy: 0.5956 - val_loss: 3.8326 - val_accuracy: 0.0810\n",
+      "Epoch 3269/5000\n",
+      "919/919 - 3s - loss: 1.2921 - accuracy: 0.5944 - val_loss: 3.8433 - val_accuracy: 0.0817\n",
+      "Epoch 3270/5000\n",
+      "919/919 - 3s - loss: 1.3022 - accuracy: 0.5923 - val_loss: 3.8321 - val_accuracy: 0.0830\n",
+      "Epoch 3271/5000\n",
+      "919/919 - 3s - loss: 1.2842 - accuracy: 0.5954 - val_loss: 3.8314 - val_accuracy: 0.0829\n",
+      "Epoch 3272/5000\n",
+      "919/919 - 3s - loss: 1.3057 - accuracy: 0.5959 - val_loss: 3.8371 - val_accuracy: 0.0814\n",
+      "Epoch 3273/5000\n",
+      "919/919 - 3s - loss: 1.3002 - accuracy: 0.5933 - val_loss: 3.8344 - val_accuracy: 0.0810\n",
+      "Epoch 3274/5000\n",
+      "919/919 - 3s - loss: 1.3626 - accuracy: 0.5944 - val_loss: 3.8287 - val_accuracy: 0.0820\n",
+      "Epoch 3275/5000\n",
+      "919/919 - 3s - loss: 1.3091 - accuracy: 0.5933 - val_loss: 3.8274 - val_accuracy: 0.0823\n",
+      "Epoch 3276/5000\n",
+      "919/919 - 3s - loss: 1.2869 - accuracy: 0.5946 - val_loss: 3.8347 - val_accuracy: 0.0816\n",
+      "Epoch 3277/5000\n",
+      "919/919 - 3s - loss: 1.3297 - accuracy: 0.5954 - val_loss: 3.8373 - val_accuracy: 0.0807\n",
+      "Epoch 3278/5000\n",
+      "919/919 - 3s - loss: 1.2875 - accuracy: 0.5960 - val_loss: 3.8334 - val_accuracy: 0.0811\n",
+      "Epoch 3279/5000\n",
+      "919/919 - 3s - loss: 1.2993 - accuracy: 0.5955 - val_loss: 3.8390 - val_accuracy: 0.0813\n",
+      "Epoch 3280/5000\n",
+      "919/919 - 3s - loss: 1.2947 - accuracy: 0.5965 - val_loss: 3.8349 - val_accuracy: 0.0812\n",
+      "Epoch 3281/5000\n",
+      "919/919 - 3s - loss: 1.3318 - accuracy: 0.5931 - val_loss: 3.8272 - val_accuracy: 0.0815\n",
+      "Epoch 3282/5000\n",
+      "919/919 - 3s - loss: 1.2984 - accuracy: 0.5935 - val_loss: 3.8316 - val_accuracy: 0.0824\n",
+      "Epoch 3283/5000\n",
+      "919/919 - 3s - loss: 1.2919 - accuracy: 0.5955 - val_loss: 3.8300 - val_accuracy: 0.0821\n",
+      "Epoch 3284/5000\n",
+      "919/919 - 3s - loss: 1.3050 - accuracy: 0.5929 - val_loss: 3.8340 - val_accuracy: 0.0830\n",
+      "Epoch 3285/5000\n",
+      "919/919 - 3s - loss: 1.2890 - accuracy: 0.5961 - val_loss: 3.8398 - val_accuracy: 0.0828\n",
+      "Epoch 3286/5000\n",
+      "919/919 - 3s - loss: 1.3049 - accuracy: 0.5891 - val_loss: 3.8406 - val_accuracy: 0.0820\n",
+      "Epoch 3287/5000\n",
+      "919/919 - 3s - loss: 1.2838 - accuracy: 0.5982 - val_loss: 3.8387 - val_accuracy: 0.0813\n",
+      "Epoch 3288/5000\n",
+      "919/919 - 3s - loss: 1.2945 - accuracy: 0.5924 - val_loss: 3.8295 - val_accuracy: 0.0819\n",
+      "Epoch 3289/5000\n",
+      "919/919 - 3s - loss: 1.2922 - accuracy: 0.5965 - val_loss: 3.8276 - val_accuracy: 0.0819\n",
+      "Epoch 3290/5000\n",
+      "919/919 - 3s - loss: 1.2923 - accuracy: 0.5929 - val_loss: 3.8311 - val_accuracy: 0.0824\n",
+      "Epoch 3291/5000\n",
+      "919/919 - 3s - loss: 1.2926 - accuracy: 0.5984 - val_loss: 3.8472 - val_accuracy: 0.0821\n",
+      "Epoch 3292/5000\n",
+      "919/919 - 3s - loss: 1.2960 - accuracy: 0.5934 - val_loss: 3.8446 - val_accuracy: 0.0821\n",
+      "Epoch 3293/5000\n",
+      "919/919 - 3s - loss: 1.2886 - accuracy: 0.5973 - val_loss: 3.8417 - val_accuracy: 0.0823\n",
+      "Epoch 3294/5000\n",
+      "919/919 - 3s - loss: 1.2849 - accuracy: 0.5948 - val_loss: 3.8476 - val_accuracy: 0.0824\n",
+      "Epoch 3295/5000\n",
+      "919/919 - 3s - loss: 1.2850 - accuracy: 0.5968 - val_loss: 3.8446 - val_accuracy: 0.0828\n",
+      "Epoch 3296/5000\n",
+      "919/919 - 3s - loss: 1.3261 - accuracy: 0.6000 - val_loss: 3.8456 - val_accuracy: 0.0828\n",
+      "Epoch 3297/5000\n",
+      "919/919 - 3s - loss: 1.3066 - accuracy: 0.6014 - val_loss: 3.8540 - val_accuracy: 0.0830\n",
+      "Epoch 3298/5000\n",
+      "919/919 - 3s - loss: 1.2865 - accuracy: 0.5946 - val_loss: 3.8490 - val_accuracy: 0.0840\n",
+      "Epoch 3299/5000\n",
+      "919/919 - 3s - loss: 1.3266 - accuracy: 0.5949 - val_loss: 3.8362 - val_accuracy: 0.0846\n",
+      "Epoch 3300/5000\n",
+      "919/919 - 3s - loss: 1.2860 - accuracy: 0.5975 - val_loss: 3.8466 - val_accuracy: 0.0839\n",
+      "Epoch 3301/5000\n",
+      "919/919 - 3s - loss: 1.2904 - accuracy: 0.5936 - val_loss: 3.8569 - val_accuracy: 0.0836\n",
+      "Epoch 3302/5000\n",
+      "919/919 - 3s - loss: 1.2958 - accuracy: 0.5927 - val_loss: 3.8519 - val_accuracy: 0.0835\n",
+      "Epoch 3303/5000\n",
+      "919/919 - 3s - loss: 1.2892 - accuracy: 0.5920 - val_loss: 3.8531 - val_accuracy: 0.0825\n",
+      "Epoch 3304/5000\n",
+      "919/919 - 3s - loss: 1.3053 - accuracy: 0.5956 - val_loss: 3.8575 - val_accuracy: 0.0826\n",
+      "Epoch 3305/5000\n",
+      "919/919 - 3s - loss: 1.3484 - accuracy: 0.5926 - val_loss: 3.8584 - val_accuracy: 0.0824\n",
+      "Epoch 3306/5000\n",
+      "919/919 - 3s - loss: 1.2957 - accuracy: 0.5910 - val_loss: 3.8533 - val_accuracy: 0.0830\n",
+      "Epoch 3307/5000\n",
+      "919/919 - 3s - loss: 1.3008 - accuracy: 0.5986 - val_loss: 3.8508 - val_accuracy: 0.0830\n",
+      "Epoch 3308/5000\n",
+      "919/919 - 3s - loss: 1.3002 - accuracy: 0.5960 - val_loss: 3.8540 - val_accuracy: 0.0830\n",
+      "Epoch 3309/5000\n",
+      "919/919 - 3s - loss: 1.2778 - accuracy: 0.5967 - val_loss: 3.8557 - val_accuracy: 0.0825\n",
+      "Epoch 3310/5000\n",
+      "919/919 - 3s - loss: 1.2744 - accuracy: 0.5988 - val_loss: 3.8527 - val_accuracy: 0.0827\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3311/5000\n",
+      "919/919 - 3s - loss: 1.3780 - accuracy: 0.5949 - val_loss: 3.8538 - val_accuracy: 0.0816\n",
+      "Epoch 3312/5000\n",
+      "919/919 - 3s - loss: 1.3536 - accuracy: 0.5924 - val_loss: 3.8547 - val_accuracy: 0.0812\n",
+      "Epoch 3313/5000\n",
+      "919/919 - 3s - loss: 1.2943 - accuracy: 0.5921 - val_loss: 3.8541 - val_accuracy: 0.0817\n",
+      "Epoch 3314/5000\n",
+      "919/919 - 3s - loss: 1.2993 - accuracy: 0.5956 - val_loss: 3.8616 - val_accuracy: 0.0820\n",
+      "Epoch 3315/5000\n",
+      "919/919 - 3s - loss: 1.2967 - accuracy: 0.5981 - val_loss: 3.8505 - val_accuracy: 0.0821\n",
+      "Epoch 3316/5000\n",
+      "919/919 - 3s - loss: 1.2849 - accuracy: 0.5975 - val_loss: 3.8490 - val_accuracy: 0.0831\n",
+      "Epoch 3317/5000\n",
+      "919/919 - 3s - loss: 1.2863 - accuracy: 0.5971 - val_loss: 3.8574 - val_accuracy: 0.0829\n",
+      "Epoch 3318/5000\n",
+      "919/919 - 3s - loss: 1.4799 - accuracy: 0.5965 - val_loss: 3.8583 - val_accuracy: 0.0822\n",
+      "Epoch 3319/5000\n",
+      "919/919 - 3s - loss: 1.2949 - accuracy: 0.5944 - val_loss: 3.8458 - val_accuracy: 0.0820\n",
+      "Epoch 3320/5000\n",
+      "919/919 - 3s - loss: 1.2856 - accuracy: 0.5978 - val_loss: 3.8408 - val_accuracy: 0.0817\n",
+      "Epoch 3321/5000\n",
+      "919/919 - 3s - loss: 1.3091 - accuracy: 0.5906 - val_loss: 3.8442 - val_accuracy: 0.0813\n",
+      "Epoch 3322/5000\n",
+      "919/919 - 3s - loss: 1.3077 - accuracy: 0.5952 - val_loss: 3.8473 - val_accuracy: 0.0821\n",
+      "Epoch 3323/5000\n",
+      "919/919 - 3s - loss: 1.2919 - accuracy: 0.5958 - val_loss: 3.8387 - val_accuracy: 0.0812\n",
+      "Epoch 3324/5000\n",
+      "919/919 - 3s - loss: 1.2686 - accuracy: 0.5990 - val_loss: 3.8385 - val_accuracy: 0.0818\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3325/5000\n",
+      "919/919 - 3s - loss: 1.2832 - accuracy: 0.5959 - val_loss: 3.8519 - val_accuracy: 0.0818\n",
+      "Epoch 3326/5000\n",
+      "919/919 - 3s - loss: 1.2855 - accuracy: 0.6005 - val_loss: 3.8479 - val_accuracy: 0.0824\n",
+      "Epoch 3327/5000\n",
+      "919/919 - 3s - loss: 1.2913 - accuracy: 0.5960 - val_loss: 3.8398 - val_accuracy: 0.0822\n",
+      "Epoch 3328/5000\n",
+      "919/919 - 3s - loss: 1.3466 - accuracy: 0.5989 - val_loss: 3.8490 - val_accuracy: 0.0824\n",
+      "Epoch 3329/5000\n",
+      "919/919 - 3s - loss: 1.2877 - accuracy: 0.5934 - val_loss: 3.8464 - val_accuracy: 0.0824\n",
+      "Epoch 3330/5000\n",
+      "919/919 - 3s - loss: 1.2752 - accuracy: 0.5981 - val_loss: 3.8537 - val_accuracy: 0.0820\n",
+      "Epoch 3331/5000\n",
+      "919/919 - 3s - loss: 1.2886 - accuracy: 0.5980 - val_loss: 3.8605 - val_accuracy: 0.0819\n",
+      "Epoch 3332/5000\n",
+      "919/919 - 3s - loss: 1.2893 - accuracy: 0.5958 - val_loss: 3.8590 - val_accuracy: 0.0814\n",
+      "Epoch 3333/5000\n",
+      "919/919 - 3s - loss: 1.2861 - accuracy: 0.5927 - val_loss: 3.8385 - val_accuracy: 0.0824\n",
+      "Epoch 3334/5000\n",
+      "919/919 - 3s - loss: 1.2921 - accuracy: 0.5944 - val_loss: 3.8384 - val_accuracy: 0.0821\n",
+      "Epoch 3335/5000\n",
+      "919/919 - 3s - loss: 1.3286 - accuracy: 0.5964 - val_loss: 3.8290 - val_accuracy: 0.0822\n",
+      "Epoch 3336/5000\n",
+      "919/919 - 3s - loss: 1.3947 - accuracy: 0.5977 - val_loss: 3.8344 - val_accuracy: 0.0820\n",
+      "Epoch 3337/5000\n",
+      "919/919 - 3s - loss: 1.2649 - accuracy: 0.5999 - val_loss: 3.8480 - val_accuracy: 0.0817\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3338/5000\n",
+      "919/919 - 3s - loss: 1.2797 - accuracy: 0.5987 - val_loss: 3.8602 - val_accuracy: 0.0817\n",
+      "Epoch 3339/5000\n",
+      "919/919 - 3s - loss: 1.3002 - accuracy: 0.5972 - val_loss: 3.8416 - val_accuracy: 0.0823\n",
+      "Epoch 3340/5000\n",
+      "919/919 - 3s - loss: 1.3631 - accuracy: 0.5990 - val_loss: 3.8532 - val_accuracy: 0.0817\n",
+      "Epoch 3341/5000\n",
+      "919/919 - 3s - loss: 1.2999 - accuracy: 0.5936 - val_loss: 3.8478 - val_accuracy: 0.0815\n",
+      "Epoch 3342/5000\n",
+      "919/919 - 3s - loss: 1.2860 - accuracy: 0.5973 - val_loss: 3.8378 - val_accuracy: 0.0810\n",
+      "Epoch 3343/5000\n",
+      "919/919 - 3s - loss: 1.2915 - accuracy: 0.5965 - val_loss: 3.8365 - val_accuracy: 0.0817\n",
+      "Epoch 3344/5000\n",
+      "919/919 - 3s - loss: 1.2888 - accuracy: 0.5977 - val_loss: 3.8462 - val_accuracy: 0.0821\n",
+      "Epoch 3345/5000\n",
+      "919/919 - 3s - loss: 1.3363 - accuracy: 0.5924 - val_loss: 3.8527 - val_accuracy: 0.0820\n",
+      "Epoch 3346/5000\n",
+      "919/919 - 3s - loss: 1.3100 - accuracy: 0.5973 - val_loss: 3.8623 - val_accuracy: 0.0820\n",
+      "Epoch 3347/5000\n",
+      "919/919 - 3s - loss: 1.2998 - accuracy: 0.5982 - val_loss: 3.8493 - val_accuracy: 0.0818\n",
+      "Epoch 3348/5000\n",
+      "919/919 - 3s - loss: 1.2839 - accuracy: 0.5962 - val_loss: 3.8523 - val_accuracy: 0.0821\n",
+      "Epoch 3349/5000\n",
+      "919/919 - 3s - loss: 1.2724 - accuracy: 0.6024 - val_loss: 3.8535 - val_accuracy: 0.0824\n",
+      "Epoch 3350/5000\n",
+      "919/919 - 3s - loss: 1.2983 - accuracy: 0.5948 - val_loss: 3.8582 - val_accuracy: 0.0821\n",
+      "Epoch 3351/5000\n",
+      "919/919 - 3s - loss: 1.2742 - accuracy: 0.5986 - val_loss: 3.8689 - val_accuracy: 0.0825\n",
+      "Epoch 3352/5000\n",
+      "919/919 - 3s - loss: 1.2764 - accuracy: 0.5987 - val_loss: 3.8651 - val_accuracy: 0.0824\n",
+      "Epoch 3353/5000\n",
+      "919/919 - 3s - loss: 1.2989 - accuracy: 0.5961 - val_loss: 3.8629 - val_accuracy: 0.0820\n",
+      "Epoch 3354/5000\n",
+      "919/919 - 3s - loss: 1.3629 - accuracy: 0.5953 - val_loss: 3.8571 - val_accuracy: 0.0822\n",
+      "Epoch 3355/5000\n",
+      "919/919 - 3s - loss: 1.2739 - accuracy: 0.5962 - val_loss: 3.8630 - val_accuracy: 0.0815\n",
+      "Epoch 3356/5000\n",
+      "919/919 - 3s - loss: 1.3361 - accuracy: 0.5990 - val_loss: 3.8533 - val_accuracy: 0.0822\n",
+      "Epoch 3357/5000\n",
+      "919/919 - 3s - loss: 1.2830 - accuracy: 0.5966 - val_loss: 3.8556 - val_accuracy: 0.0819\n",
+      "Epoch 3358/5000\n",
+      "919/919 - 3s - loss: 1.2700 - accuracy: 0.6011 - val_loss: 3.8596 - val_accuracy: 0.0806\n",
+      "Epoch 3359/5000\n",
+      "919/919 - 3s - loss: 1.3065 - accuracy: 0.5950 - val_loss: 3.8439 - val_accuracy: 0.0814\n",
+      "Epoch 3360/5000\n",
+      "919/919 - 3s - loss: 1.2646 - accuracy: 0.6011 - val_loss: 3.8498 - val_accuracy: 0.0812\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3361/5000\n",
+      "919/919 - 3s - loss: 1.2773 - accuracy: 0.6004 - val_loss: 3.8592 - val_accuracy: 0.0815\n",
+      "Epoch 3362/5000\n",
+      "919/919 - 3s - loss: 1.2999 - accuracy: 0.5977 - val_loss: 3.8665 - val_accuracy: 0.0816\n",
+      "Epoch 3363/5000\n",
+      "919/919 - 3s - loss: 1.2649 - accuracy: 0.5984 - val_loss: 3.8674 - val_accuracy: 0.0829\n",
+      "Epoch 3364/5000\n",
+      "919/919 - 3s - loss: 1.2975 - accuracy: 0.6015 - val_loss: 3.8794 - val_accuracy: 0.0830\n",
+      "Epoch 3365/5000\n",
+      "919/919 - 3s - loss: 1.2783 - accuracy: 0.6020 - val_loss: 3.8702 - val_accuracy: 0.0830\n",
+      "Epoch 3366/5000\n",
+      "919/919 - 3s - loss: 1.2828 - accuracy: 0.5986 - val_loss: 3.8745 - val_accuracy: 0.0820\n",
+      "Epoch 3367/5000\n",
+      "919/919 - 3s - loss: 1.3340 - accuracy: 0.5972 - val_loss: 3.8743 - val_accuracy: 0.0820\n",
+      "Epoch 3368/5000\n",
+      "919/919 - 3s - loss: 1.3950 - accuracy: 0.5981 - val_loss: 3.8628 - val_accuracy: 0.0827\n",
+      "Epoch 3369/5000\n",
+      "919/919 - 3s - loss: 1.2793 - accuracy: 0.5969 - val_loss: 3.8631 - val_accuracy: 0.0818\n",
+      "Epoch 3370/5000\n",
+      "919/919 - 3s - loss: 1.2844 - accuracy: 0.5990 - val_loss: 3.8645 - val_accuracy: 0.0823\n",
+      "Epoch 3371/5000\n",
+      "919/919 - 3s - loss: 1.2618 - accuracy: 0.5995 - val_loss: 3.8592 - val_accuracy: 0.0830\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3372/5000\n",
+      "919/919 - 3s - loss: 1.2643 - accuracy: 0.6024 - val_loss: 3.8804 - val_accuracy: 0.0820\n",
+      "Epoch 3373/5000\n",
+      "919/919 - 3s - loss: 1.2765 - accuracy: 0.5976 - val_loss: 3.8645 - val_accuracy: 0.0821\n",
+      "Epoch 3374/5000\n",
+      "919/919 - 3s - loss: 1.2961 - accuracy: 0.5938 - val_loss: 3.8730 - val_accuracy: 0.0820\n",
+      "Epoch 3375/5000\n",
+      "919/919 - 3s - loss: 1.3128 - accuracy: 0.5972 - val_loss: 3.8639 - val_accuracy: 0.0817\n",
+      "Epoch 3376/5000\n",
+      "919/919 - 3s - loss: 1.2904 - accuracy: 0.5985 - val_loss: 3.8637 - val_accuracy: 0.0823\n",
+      "Epoch 3377/5000\n",
+      "919/919 - 3s - loss: 1.2992 - accuracy: 0.5913 - val_loss: 3.8653 - val_accuracy: 0.0813\n",
+      "Epoch 3378/5000\n",
+      "919/919 - 3s - loss: 1.2750 - accuracy: 0.5995 - val_loss: 3.8767 - val_accuracy: 0.0813\n",
+      "Epoch 3379/5000\n",
+      "919/919 - 3s - loss: 1.2677 - accuracy: 0.5996 - val_loss: 3.8748 - val_accuracy: 0.0818\n",
+      "Epoch 3380/5000\n",
+      "919/919 - 3s - loss: 1.2956 - accuracy: 0.5965 - val_loss: 3.8756 - val_accuracy: 0.0806\n",
+      "Epoch 3381/5000\n",
+      "919/919 - 3s - loss: 1.2934 - accuracy: 0.5979 - val_loss: 3.8730 - val_accuracy: 0.0814\n",
+      "Epoch 3382/5000\n",
+      "919/919 - 3s - loss: 1.2840 - accuracy: 0.5917 - val_loss: 3.8794 - val_accuracy: 0.0808\n",
+      "Epoch 3383/5000\n",
+      "919/919 - 3s - loss: 1.3191 - accuracy: 0.5969 - val_loss: 3.8746 - val_accuracy: 0.0813\n",
+      "Epoch 3384/5000\n",
+      "919/919 - 3s - loss: 1.2765 - accuracy: 0.5990 - val_loss: 3.8688 - val_accuracy: 0.0819\n",
+      "Epoch 3385/5000\n",
+      "919/919 - 3s - loss: 1.3363 - accuracy: 0.5961 - val_loss: 3.8630 - val_accuracy: 0.0816\n",
+      "Epoch 3386/5000\n",
+      "919/919 - 3s - loss: 1.3869 - accuracy: 0.5974 - val_loss: 3.8684 - val_accuracy: 0.0821\n",
+      "Epoch 3387/5000\n",
+      "919/919 - 3s - loss: 1.2762 - accuracy: 0.5952 - val_loss: 3.8675 - val_accuracy: 0.0819\n",
+      "Epoch 3388/5000\n",
+      "919/919 - 3s - loss: 1.2771 - accuracy: 0.5983 - val_loss: 3.8632 - val_accuracy: 0.0821\n",
+      "Epoch 3389/5000\n",
+      "919/919 - 3s - loss: 1.2704 - accuracy: 0.5958 - val_loss: 3.8708 - val_accuracy: 0.0823\n",
+      "Epoch 3390/5000\n",
+      "919/919 - 3s - loss: 1.3310 - accuracy: 0.5977 - val_loss: 3.8677 - val_accuracy: 0.0821\n",
+      "Epoch 3391/5000\n",
+      "919/919 - 3s - loss: 1.2787 - accuracy: 0.6021 - val_loss: 3.8638 - val_accuracy: 0.0823\n",
+      "Epoch 3392/5000\n",
+      "919/919 - 3s - loss: 1.2840 - accuracy: 0.5962 - val_loss: 3.8602 - val_accuracy: 0.0820\n",
+      "Epoch 3393/5000\n",
+      "919/919 - 3s - loss: 1.2904 - accuracy: 0.6010 - val_loss: 3.8604 - val_accuracy: 0.0819\n",
+      "Epoch 3394/5000\n",
+      "919/919 - 3s - loss: 1.2825 - accuracy: 0.5976 - val_loss: 3.8654 - val_accuracy: 0.0815\n",
+      "Epoch 3395/5000\n",
+      "919/919 - 3s - loss: 1.2670 - accuracy: 0.5994 - val_loss: 3.8723 - val_accuracy: 0.0821\n",
+      "Epoch 3396/5000\n",
+      "919/919 - 3s - loss: 1.3369 - accuracy: 0.6037 - val_loss: 3.8801 - val_accuracy: 0.0814\n",
+      "Epoch 3397/5000\n",
+      "919/919 - 3s - loss: 1.3160 - accuracy: 0.5969 - val_loss: 3.8821 - val_accuracy: 0.0822\n",
+      "Epoch 3398/5000\n",
+      "919/919 - 3s - loss: 1.2810 - accuracy: 0.5980 - val_loss: 3.8903 - val_accuracy: 0.0817\n",
+      "Epoch 3399/5000\n",
+      "919/919 - 3s - loss: 1.2754 - accuracy: 0.5991 - val_loss: 3.8766 - val_accuracy: 0.0825\n",
+      "Epoch 3400/5000\n",
+      "919/919 - 3s - loss: 1.2695 - accuracy: 0.6014 - val_loss: 3.8723 - val_accuracy: 0.0818\n",
+      "Epoch 3401/5000\n",
+      "919/919 - 3s - loss: 1.2750 - accuracy: 0.5997 - val_loss: 3.8839 - val_accuracy: 0.0819\n",
+      "Epoch 3402/5000\n",
+      "919/919 - 3s - loss: 1.2791 - accuracy: 0.6018 - val_loss: 3.8831 - val_accuracy: 0.0824\n",
+      "Epoch 3403/5000\n",
+      "919/919 - 3s - loss: 1.2714 - accuracy: 0.5990 - val_loss: 3.8860 - val_accuracy: 0.0824\n",
+      "Epoch 3404/5000\n",
+      "919/919 - 3s - loss: 1.2764 - accuracy: 0.6018 - val_loss: 3.8946 - val_accuracy: 0.0827\n",
+      "Epoch 3405/5000\n",
+      "919/919 - 3s - loss: 1.4330 - accuracy: 0.5988 - val_loss: 3.8818 - val_accuracy: 0.0823\n",
+      "Epoch 3406/5000\n",
+      "919/919 - 3s - loss: 1.2826 - accuracy: 0.6019 - val_loss: 3.8663 - val_accuracy: 0.0822\n",
+      "Epoch 3407/5000\n",
+      "919/919 - 3s - loss: 1.2792 - accuracy: 0.5967 - val_loss: 3.8654 - val_accuracy: 0.0828\n",
+      "Epoch 3408/5000\n",
+      "919/919 - 3s - loss: 1.2699 - accuracy: 0.6032 - val_loss: 3.8720 - val_accuracy: 0.0821\n",
+      "Epoch 3409/5000\n",
+      "919/919 - 3s - loss: 1.2687 - accuracy: 0.6003 - val_loss: 3.8745 - val_accuracy: 0.0828\n",
+      "Epoch 3410/5000\n",
+      "919/919 - 3s - loss: 1.2712 - accuracy: 0.6014 - val_loss: 3.8705 - val_accuracy: 0.0828\n",
+      "Epoch 3411/5000\n",
+      "919/919 - 3s - loss: 1.2647 - accuracy: 0.6018 - val_loss: 3.8774 - val_accuracy: 0.0827\n",
+      "Epoch 3412/5000\n",
+      "919/919 - 3s - loss: 1.2894 - accuracy: 0.6016 - val_loss: 3.8677 - val_accuracy: 0.0823\n",
+      "Epoch 3413/5000\n",
+      "919/919 - 3s - loss: 1.2967 - accuracy: 0.6001 - val_loss: 3.8836 - val_accuracy: 0.0827\n",
+      "Epoch 3414/5000\n",
+      "919/919 - 3s - loss: 1.3452 - accuracy: 0.5985 - val_loss: 3.8764 - val_accuracy: 0.0822\n",
+      "Epoch 3415/5000\n",
+      "919/919 - 3s - loss: 1.2811 - accuracy: 0.5990 - val_loss: 3.8771 - val_accuracy: 0.0829\n",
+      "Epoch 3416/5000\n",
+      "919/919 - 3s - loss: 1.2756 - accuracy: 0.6005 - val_loss: 3.8774 - val_accuracy: 0.0823\n",
+      "Epoch 3417/5000\n",
+      "919/919 - 3s - loss: 1.2606 - accuracy: 0.6022 - val_loss: 3.8840 - val_accuracy: 0.0824\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3418/5000\n",
+      "919/919 - 3s - loss: 1.2802 - accuracy: 0.5999 - val_loss: 3.8813 - val_accuracy: 0.0820\n",
+      "Epoch 3419/5000\n",
+      "919/919 - 3s - loss: 1.3115 - accuracy: 0.6000 - val_loss: 3.8811 - val_accuracy: 0.0820\n",
+      "Epoch 3420/5000\n",
+      "919/919 - 3s - loss: 1.2680 - accuracy: 0.5979 - val_loss: 3.8789 - val_accuracy: 0.0815\n",
+      "Epoch 3421/5000\n",
+      "919/919 - 3s - loss: 1.3064 - accuracy: 0.5990 - val_loss: 3.8787 - val_accuracy: 0.0821\n",
+      "Epoch 3422/5000\n",
+      "919/919 - 3s - loss: 1.3048 - accuracy: 0.6024 - val_loss: 3.8911 - val_accuracy: 0.0827\n",
+      "Epoch 3423/5000\n",
+      "919/919 - 3s - loss: 1.3047 - accuracy: 0.6020 - val_loss: 3.8876 - val_accuracy: 0.0820\n",
+      "Epoch 3424/5000\n",
+      "919/919 - 3s - loss: 1.2732 - accuracy: 0.6017 - val_loss: 3.8981 - val_accuracy: 0.0813\n",
+      "Epoch 3425/5000\n",
+      "919/919 - 3s - loss: 1.2815 - accuracy: 0.6003 - val_loss: 3.8768 - val_accuracy: 0.0812\n",
+      "Epoch 3426/5000\n",
+      "919/919 - 3s - loss: 1.2698 - accuracy: 0.5999 - val_loss: 3.8899 - val_accuracy: 0.0819\n",
+      "Epoch 3427/5000\n",
+      "919/919 - 3s - loss: 1.2773 - accuracy: 0.5952 - val_loss: 3.8943 - val_accuracy: 0.0812\n",
+      "Epoch 3428/5000\n",
+      "919/919 - 3s - loss: 1.2715 - accuracy: 0.5987 - val_loss: 3.8878 - val_accuracy: 0.0819\n",
+      "Epoch 3429/5000\n",
+      "919/919 - 3s - loss: 1.2759 - accuracy: 0.5985 - val_loss: 3.8812 - val_accuracy: 0.0821\n",
+      "Epoch 3430/5000\n",
+      "919/919 - 3s - loss: 1.2706 - accuracy: 0.5986 - val_loss: 3.8822 - val_accuracy: 0.0829\n",
+      "Epoch 3431/5000\n",
+      "919/919 - 3s - loss: 1.2759 - accuracy: 0.5963 - val_loss: 3.8836 - val_accuracy: 0.0829\n",
+      "Epoch 3432/5000\n",
+      "919/919 - 3s - loss: 1.2870 - accuracy: 0.5961 - val_loss: 3.8737 - val_accuracy: 0.0828\n",
+      "Epoch 3433/5000\n",
+      "919/919 - 3s - loss: 1.2644 - accuracy: 0.6035 - val_loss: 3.8696 - val_accuracy: 0.0830\n",
+      "Epoch 3434/5000\n",
+      "919/919 - 3s - loss: 1.2702 - accuracy: 0.6029 - val_loss: 3.8724 - val_accuracy: 0.0823\n",
+      "Epoch 3435/5000\n",
+      "919/919 - 3s - loss: 1.3705 - accuracy: 0.6029 - val_loss: 3.8818 - val_accuracy: 0.0827\n",
+      "Epoch 3436/5000\n",
+      "919/919 - 3s - loss: 1.2580 - accuracy: 0.6038 - val_loss: 3.8767 - val_accuracy: 0.0825\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3437/5000\n",
+      "919/919 - 3s - loss: 1.2646 - accuracy: 0.6018 - val_loss: 3.8782 - val_accuracy: 0.0827\n",
+      "Epoch 3438/5000\n",
+      "919/919 - 3s - loss: 1.2672 - accuracy: 0.5992 - val_loss: 3.8822 - val_accuracy: 0.0827\n",
+      "Epoch 3439/5000\n",
+      "919/919 - 3s - loss: 1.2799 - accuracy: 0.6003 - val_loss: 3.9066 - val_accuracy: 0.0817\n",
+      "Epoch 3440/5000\n",
+      "919/919 - 3s - loss: 1.2825 - accuracy: 0.6014 - val_loss: 3.9061 - val_accuracy: 0.0817\n",
+      "Epoch 3441/5000\n",
+      "919/919 - 3s - loss: 1.2670 - accuracy: 0.5993 - val_loss: 3.8919 - val_accuracy: 0.0821\n",
+      "Epoch 3442/5000\n",
+      "919/919 - 3s - loss: 1.2808 - accuracy: 0.6031 - val_loss: 3.9100 - val_accuracy: 0.0815\n",
+      "Epoch 3443/5000\n",
+      "919/919 - 3s - loss: 1.2739 - accuracy: 0.5980 - val_loss: 3.9055 - val_accuracy: 0.0820\n",
+      "Epoch 3444/5000\n",
+      "919/919 - 3s - loss: 1.2747 - accuracy: 0.5986 - val_loss: 3.8981 - val_accuracy: 0.0824\n",
+      "Epoch 3445/5000\n",
+      "919/919 - 3s - loss: 1.2757 - accuracy: 0.5990 - val_loss: 3.9005 - val_accuracy: 0.0820\n",
+      "Epoch 3446/5000\n",
+      "919/919 - 3s - loss: 1.2690 - accuracy: 0.5996 - val_loss: 3.9023 - val_accuracy: 0.0814\n",
+      "Epoch 3447/5000\n",
+      "919/919 - 3s - loss: 1.2924 - accuracy: 0.6022 - val_loss: 3.9083 - val_accuracy: 0.0816\n",
+      "Epoch 3448/5000\n",
+      "919/919 - 3s - loss: 1.2586 - accuracy: 0.6012 - val_loss: 3.9125 - val_accuracy: 0.0822\n",
+      "Epoch 3449/5000\n",
+      "919/919 - 3s - loss: 1.3601 - accuracy: 0.6010 - val_loss: 3.9010 - val_accuracy: 0.0828\n",
+      "Epoch 3450/5000\n",
+      "919/919 - 3s - loss: 1.2676 - accuracy: 0.6030 - val_loss: 3.9050 - val_accuracy: 0.0830\n",
+      "Epoch 3451/5000\n",
+      "919/919 - 3s - loss: 1.3517 - accuracy: 0.6007 - val_loss: 3.8953 - val_accuracy: 0.0834\n",
+      "Epoch 3452/5000\n",
+      "919/919 - 3s - loss: 1.2679 - accuracy: 0.5978 - val_loss: 3.8910 - val_accuracy: 0.0832\n",
+      "Epoch 3453/5000\n",
+      "919/919 - 3s - loss: 1.2626 - accuracy: 0.6015 - val_loss: 3.8934 - val_accuracy: 0.0829\n",
+      "Epoch 3454/5000\n",
+      "919/919 - 3s - loss: 1.2759 - accuracy: 0.5992 - val_loss: 3.8979 - val_accuracy: 0.0830\n",
+      "Epoch 3455/5000\n",
+      "919/919 - 3s - loss: 1.2799 - accuracy: 0.6016 - val_loss: 3.9065 - val_accuracy: 0.0830\n",
+      "Epoch 3456/5000\n",
+      "919/919 - 3s - loss: 1.2709 - accuracy: 0.5976 - val_loss: 3.9118 - val_accuracy: 0.0819\n",
+      "Epoch 3457/5000\n",
+      "919/919 - 3s - loss: 1.2804 - accuracy: 0.6016 - val_loss: 3.8975 - val_accuracy: 0.0825\n",
+      "Epoch 3458/5000\n",
+      "919/919 - 3s - loss: 1.2758 - accuracy: 0.6032 - val_loss: 3.8949 - val_accuracy: 0.0820\n",
+      "Epoch 3459/5000\n",
+      "919/919 - 3s - loss: 1.2688 - accuracy: 0.6050 - val_loss: 3.8991 - val_accuracy: 0.0826\n",
+      "Epoch 3460/5000\n",
+      "919/919 - 3s - loss: 1.2822 - accuracy: 0.6012 - val_loss: 3.8906 - val_accuracy: 0.0824\n",
+      "Epoch 3461/5000\n",
+      "919/919 - 3s - loss: 1.2697 - accuracy: 0.6027 - val_loss: 3.8831 - val_accuracy: 0.0832\n",
+      "Epoch 3462/5000\n",
+      "919/919 - 3s - loss: 1.2796 - accuracy: 0.6022 - val_loss: 3.8835 - val_accuracy: 0.0837\n",
+      "Epoch 3463/5000\n",
+      "919/919 - 3s - loss: 1.2884 - accuracy: 0.6031 - val_loss: 3.8752 - val_accuracy: 0.0840\n",
+      "Epoch 3464/5000\n",
+      "919/919 - 3s - loss: 1.2711 - accuracy: 0.6026 - val_loss: 3.8834 - val_accuracy: 0.0840\n",
+      "Epoch 3465/5000\n",
+      "919/919 - 3s - loss: 1.2752 - accuracy: 0.6027 - val_loss: 3.8855 - val_accuracy: 0.0830\n",
+      "Epoch 3466/5000\n",
+      "919/919 - 3s - loss: 1.2984 - accuracy: 0.5999 - val_loss: 3.8819 - val_accuracy: 0.0837\n",
+      "Epoch 3467/5000\n",
+      "919/919 - 3s - loss: 1.2758 - accuracy: 0.6020 - val_loss: 3.8857 - val_accuracy: 0.0833\n",
+      "Epoch 3468/5000\n",
+      "919/919 - 3s - loss: 1.2840 - accuracy: 0.5989 - val_loss: 3.8868 - val_accuracy: 0.0831\n",
+      "Epoch 3469/5000\n",
+      "919/919 - 3s - loss: 1.3701 - accuracy: 0.6054 - val_loss: 3.8924 - val_accuracy: 0.0828\n",
+      "Epoch 3470/5000\n",
+      "919/919 - 3s - loss: 1.2772 - accuracy: 0.5993 - val_loss: 3.8873 - val_accuracy: 0.0841\n",
+      "Epoch 3471/5000\n",
+      "919/919 - 3s - loss: 1.2718 - accuracy: 0.6051 - val_loss: 3.9006 - val_accuracy: 0.0835\n",
+      "Epoch 3472/5000\n",
+      "919/919 - 3s - loss: 1.3375 - accuracy: 0.6013 - val_loss: 3.9024 - val_accuracy: 0.0830\n",
+      "Epoch 3473/5000\n",
+      "919/919 - 3s - loss: 1.2677 - accuracy: 0.6003 - val_loss: 3.9128 - val_accuracy: 0.0837\n",
+      "Epoch 3474/5000\n",
+      "919/919 - 3s - loss: 1.2614 - accuracy: 0.6034 - val_loss: 3.9154 - val_accuracy: 0.0836\n",
+      "Epoch 3475/5000\n",
+      "919/919 - 3s - loss: 1.3180 - accuracy: 0.6031 - val_loss: 3.9056 - val_accuracy: 0.0830\n",
+      "Epoch 3476/5000\n",
+      "919/919 - 3s - loss: 1.2781 - accuracy: 0.6022 - val_loss: 3.9012 - val_accuracy: 0.0829\n",
+      "Epoch 3477/5000\n",
+      "919/919 - 3s - loss: 1.2738 - accuracy: 0.5985 - val_loss: 3.8976 - val_accuracy: 0.0834\n",
+      "Epoch 3478/5000\n",
+      "919/919 - 3s - loss: 1.3011 - accuracy: 0.6015 - val_loss: 3.8897 - val_accuracy: 0.0839\n",
+      "Epoch 3479/5000\n",
+      "919/919 - 3s - loss: 1.2875 - accuracy: 0.6056 - val_loss: 3.8919 - val_accuracy: 0.0834\n",
+      "Epoch 3480/5000\n",
+      "919/919 - 3s - loss: 1.2685 - accuracy: 0.6043 - val_loss: 3.8920 - val_accuracy: 0.0840\n",
+      "Epoch 3481/5000\n",
+      "919/919 - 3s - loss: 1.2853 - accuracy: 0.6018 - val_loss: 3.9009 - val_accuracy: 0.0841\n",
+      "Epoch 3482/5000\n",
+      "919/919 - 3s - loss: 1.2726 - accuracy: 0.5989 - val_loss: 3.8966 - val_accuracy: 0.0837\n",
+      "Epoch 3483/5000\n",
+      "919/919 - 3s - loss: 1.2708 - accuracy: 0.5995 - val_loss: 3.9073 - val_accuracy: 0.0837\n",
+      "Epoch 3484/5000\n",
+      "919/919 - 3s - loss: 1.2595 - accuracy: 0.6029 - val_loss: 3.8988 - val_accuracy: 0.0847\n",
+      "Epoch 3485/5000\n",
+      "919/919 - 3s - loss: 1.2768 - accuracy: 0.6045 - val_loss: 3.8969 - val_accuracy: 0.0834\n",
+      "Epoch 3486/5000\n",
+      "919/919 - 3s - loss: 1.2624 - accuracy: 0.6007 - val_loss: 3.8991 - val_accuracy: 0.0833\n",
+      "Epoch 3487/5000\n",
+      "919/919 - 3s - loss: 1.2869 - accuracy: 0.6001 - val_loss: 3.9067 - val_accuracy: 0.0833\n",
+      "Epoch 3488/5000\n",
+      "919/919 - 3s - loss: 1.2585 - accuracy: 0.6016 - val_loss: 3.9223 - val_accuracy: 0.0828\n",
+      "Epoch 3489/5000\n",
+      "919/919 - 3s - loss: 1.2868 - accuracy: 0.6010 - val_loss: 3.9162 - val_accuracy: 0.0830\n",
+      "Epoch 3490/5000\n",
+      "919/919 - 3s - loss: 1.2911 - accuracy: 0.6054 - val_loss: 3.9065 - val_accuracy: 0.0833\n",
+      "Epoch 3491/5000\n",
+      "919/919 - 3s - loss: 1.2700 - accuracy: 0.6018 - val_loss: 3.9170 - val_accuracy: 0.0829\n",
+      "Epoch 3492/5000\n",
+      "919/919 - 3s - loss: 1.2644 - accuracy: 0.6053 - val_loss: 3.9170 - val_accuracy: 0.0830\n",
+      "Epoch 3493/5000\n",
+      "919/919 - 3s - loss: 1.3039 - accuracy: 0.6007 - val_loss: 3.8983 - val_accuracy: 0.0840\n",
+      "Epoch 3494/5000\n",
+      "919/919 - 3s - loss: 1.2865 - accuracy: 0.6002 - val_loss: 3.8977 - val_accuracy: 0.0839\n",
+      "Epoch 3495/5000\n",
+      "919/919 - 3s - loss: 1.3318 - accuracy: 0.6030 - val_loss: 3.8921 - val_accuracy: 0.0838\n",
+      "Epoch 3496/5000\n",
+      "919/919 - 3s - loss: 1.2493 - accuracy: 0.6026 - val_loss: 3.8827 - val_accuracy: 0.0833\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3497/5000\n",
+      "919/919 - 3s - loss: 1.2485 - accuracy: 0.6048 - val_loss: 3.9114 - val_accuracy: 0.0837\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3498/5000\n",
+      "919/919 - 3s - loss: 1.2662 - accuracy: 0.6010 - val_loss: 3.9130 - val_accuracy: 0.0833\n",
+      "Epoch 3499/5000\n",
+      "919/919 - 3s - loss: 1.2940 - accuracy: 0.6005 - val_loss: 3.9048 - val_accuracy: 0.0831\n",
+      "Epoch 3500/5000\n",
+      "919/919 - 3s - loss: 1.2729 - accuracy: 0.6012 - val_loss: 3.9049 - val_accuracy: 0.0827\n",
+      "Epoch 3501/5000\n",
+      "919/919 - 3s - loss: 1.2506 - accuracy: 0.6022 - val_loss: 3.9099 - val_accuracy: 0.0833\n",
+      "Epoch 3502/5000\n",
+      "919/919 - 3s - loss: 1.2822 - accuracy: 0.6041 - val_loss: 3.9157 - val_accuracy: 0.0838\n",
+      "Epoch 3503/5000\n",
+      "919/919 - 3s - loss: 1.2529 - accuracy: 0.6029 - val_loss: 3.9210 - val_accuracy: 0.0833\n",
+      "Epoch 3504/5000\n",
+      "919/919 - 3s - loss: 1.2935 - accuracy: 0.5989 - val_loss: 3.9034 - val_accuracy: 0.0830\n",
+      "Epoch 3505/5000\n",
+      "919/919 - 3s - loss: 1.2524 - accuracy: 0.6057 - val_loss: 3.9041 - val_accuracy: 0.0834\n",
+      "Epoch 3506/5000\n",
+      "919/919 - 3s - loss: 1.2618 - accuracy: 0.6057 - val_loss: 3.9022 - val_accuracy: 0.0838\n",
+      "Epoch 3507/5000\n",
+      "919/919 - 3s - loss: 1.2744 - accuracy: 0.6001 - val_loss: 3.9090 - val_accuracy: 0.0838\n",
+      "Epoch 3508/5000\n",
+      "919/919 - 3s - loss: 1.2445 - accuracy: 0.6071 - val_loss: 3.9246 - val_accuracy: 0.0832\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3509/5000\n",
+      "919/919 - 3s - loss: 1.2661 - accuracy: 0.6027 - val_loss: 3.9168 - val_accuracy: 0.0836\n",
+      "Epoch 3510/5000\n",
+      "919/919 - 3s - loss: 1.2841 - accuracy: 0.6005 - val_loss: 3.9112 - val_accuracy: 0.0831\n",
+      "Epoch 3511/5000\n",
+      "919/919 - 3s - loss: 1.2710 - accuracy: 0.6005 - val_loss: 3.9215 - val_accuracy: 0.0839\n",
+      "Epoch 3512/5000\n",
+      "919/919 - 3s - loss: 1.2704 - accuracy: 0.6031 - val_loss: 3.9267 - val_accuracy: 0.0837\n",
+      "Epoch 3513/5000\n",
+      "919/919 - 3s - loss: 1.4890 - accuracy: 0.6014 - val_loss: 3.9291 - val_accuracy: 0.0834\n",
+      "Epoch 3514/5000\n",
+      "919/919 - 3s - loss: 1.3441 - accuracy: 0.6049 - val_loss: 3.9255 - val_accuracy: 0.0841\n",
+      "Epoch 3515/5000\n",
+      "919/919 - 3s - loss: 1.2552 - accuracy: 0.6081 - val_loss: 3.9267 - val_accuracy: 0.0834\n",
+      "Epoch 3516/5000\n",
+      "919/919 - 3s - loss: 1.2642 - accuracy: 0.6055 - val_loss: 3.9218 - val_accuracy: 0.0838\n",
+      "Epoch 3517/5000\n",
+      "919/919 - 3s - loss: 1.2536 - accuracy: 0.6044 - val_loss: 3.9222 - val_accuracy: 0.0833\n",
+      "Epoch 3518/5000\n",
+      "919/919 - 3s - loss: 1.2736 - accuracy: 0.6050 - val_loss: 3.9311 - val_accuracy: 0.0838\n",
+      "Epoch 3519/5000\n",
+      "919/919 - 3s - loss: 1.2635 - accuracy: 0.6043 - val_loss: 3.9170 - val_accuracy: 0.0844\n",
+      "Epoch 3520/5000\n",
+      "919/919 - 3s - loss: 1.2836 - accuracy: 0.5979 - val_loss: 3.9224 - val_accuracy: 0.0837\n",
+      "Epoch 3521/5000\n",
+      "919/919 - 3s - loss: 1.2667 - accuracy: 0.6048 - val_loss: 3.9226 - val_accuracy: 0.0838\n",
+      "Epoch 3522/5000\n",
+      "919/919 - 3s - loss: 1.2547 - accuracy: 0.6048 - val_loss: 3.9187 - val_accuracy: 0.0837\n",
+      "Epoch 3523/5000\n",
+      "919/919 - 3s - loss: 1.2536 - accuracy: 0.6056 - val_loss: 3.9190 - val_accuracy: 0.0835\n",
+      "Epoch 3524/5000\n",
+      "919/919 - 3s - loss: 1.2596 - accuracy: 0.6052 - val_loss: 3.9128 - val_accuracy: 0.0835\n",
+      "Epoch 3525/5000\n",
+      "919/919 - 3s - loss: 1.2585 - accuracy: 0.6011 - val_loss: 3.9061 - val_accuracy: 0.0834\n",
+      "Epoch 3526/5000\n",
+      "919/919 - 3s - loss: 1.2561 - accuracy: 0.6067 - val_loss: 3.9138 - val_accuracy: 0.0833\n",
+      "Epoch 3527/5000\n",
+      "919/919 - 3s - loss: 1.2682 - accuracy: 0.6047 - val_loss: 3.9117 - val_accuracy: 0.0839\n",
+      "Epoch 3528/5000\n",
+      "919/919 - 3s - loss: 1.2679 - accuracy: 0.6012 - val_loss: 3.8968 - val_accuracy: 0.0825\n",
+      "Epoch 3529/5000\n",
+      "919/919 - 3s - loss: 1.3304 - accuracy: 0.6024 - val_loss: 3.9089 - val_accuracy: 0.0837\n",
+      "Epoch 3530/5000\n",
+      "919/919 - 3s - loss: 1.2575 - accuracy: 0.6048 - val_loss: 3.9197 - val_accuracy: 0.0837\n",
+      "Epoch 3531/5000\n",
+      "919/919 - 3s - loss: 1.2578 - accuracy: 0.6072 - val_loss: 3.9228 - val_accuracy: 0.0839\n",
+      "Epoch 3532/5000\n",
+      "919/919 - 3s - loss: 1.2518 - accuracy: 0.6063 - val_loss: 3.9178 - val_accuracy: 0.0831\n",
+      "Epoch 3533/5000\n",
+      "919/919 - 3s - loss: 1.3073 - accuracy: 0.6031 - val_loss: 3.9113 - val_accuracy: 0.0831\n",
+      "Epoch 3534/5000\n",
+      "919/919 - 3s - loss: 1.2945 - accuracy: 0.6032 - val_loss: 3.9060 - val_accuracy: 0.0833\n",
+      "Epoch 3535/5000\n",
+      "919/919 - 3s - loss: 1.2847 - accuracy: 0.6059 - val_loss: 3.9137 - val_accuracy: 0.0837\n",
+      "Epoch 3536/5000\n",
+      "919/919 - 3s - loss: 1.2668 - accuracy: 0.6071 - val_loss: 3.9128 - val_accuracy: 0.0842\n",
+      "Epoch 3537/5000\n",
+      "919/919 - 3s - loss: 1.2570 - accuracy: 0.6084 - val_loss: 3.9029 - val_accuracy: 0.0840\n",
+      "Epoch 3538/5000\n",
+      "919/919 - 3s - loss: 1.2604 - accuracy: 0.6039 - val_loss: 3.8995 - val_accuracy: 0.0840\n",
+      "Epoch 3539/5000\n",
+      "919/919 - 3s - loss: 1.2641 - accuracy: 0.6013 - val_loss: 3.9065 - val_accuracy: 0.0844\n",
+      "Epoch 3540/5000\n",
+      "919/919 - 3s - loss: 1.2886 - accuracy: 0.5997 - val_loss: 3.9090 - val_accuracy: 0.0836\n",
+      "Epoch 3541/5000\n",
+      "919/919 - 3s - loss: 1.2972 - accuracy: 0.6024 - val_loss: 3.9052 - val_accuracy: 0.0839\n",
+      "Epoch 3542/5000\n",
+      "919/919 - 3s - loss: 1.2481 - accuracy: 0.6091 - val_loss: 3.9297 - val_accuracy: 0.0834\n",
+      "Epoch 3543/5000\n",
+      "919/919 - 3s - loss: 1.2624 - accuracy: 0.6032 - val_loss: 3.9198 - val_accuracy: 0.0834\n",
+      "Epoch 3544/5000\n",
+      "919/919 - 3s - loss: 1.2630 - accuracy: 0.6039 - val_loss: 3.9183 - val_accuracy: 0.0837\n",
+      "Epoch 3545/5000\n",
+      "919/919 - 3s - loss: 1.2681 - accuracy: 0.6054 - val_loss: 3.9221 - val_accuracy: 0.0841\n",
+      "Epoch 3546/5000\n",
+      "919/919 - 3s - loss: 1.2595 - accuracy: 0.6093 - val_loss: 3.9382 - val_accuracy: 0.0835\n",
+      "Epoch 3547/5000\n",
+      "919/919 - 3s - loss: 1.2965 - accuracy: 0.6043 - val_loss: 3.9310 - val_accuracy: 0.0834\n",
+      "Epoch 3548/5000\n",
+      "919/919 - 3s - loss: 1.2551 - accuracy: 0.6046 - val_loss: 3.9151 - val_accuracy: 0.0837\n",
+      "Epoch 3549/5000\n",
+      "919/919 - 3s - loss: 1.2822 - accuracy: 0.6011 - val_loss: 3.9090 - val_accuracy: 0.0830\n",
+      "Epoch 3550/5000\n",
+      "919/919 - 3s - loss: 1.2814 - accuracy: 0.6007 - val_loss: 3.8971 - val_accuracy: 0.0835\n",
+      "Epoch 3551/5000\n",
+      "919/919 - 3s - loss: 1.3019 - accuracy: 0.6014 - val_loss: 3.9079 - val_accuracy: 0.0839\n",
+      "Epoch 3552/5000\n",
+      "919/919 - 3s - loss: 1.2657 - accuracy: 0.6035 - val_loss: 3.9213 - val_accuracy: 0.0840\n",
+      "Epoch 3553/5000\n",
+      "919/919 - 3s - loss: 1.2550 - accuracy: 0.6091 - val_loss: 3.9199 - val_accuracy: 0.0839\n",
+      "Epoch 3554/5000\n",
+      "919/919 - 3s - loss: 1.2903 - accuracy: 0.6028 - val_loss: 3.9169 - val_accuracy: 0.0839\n",
+      "Epoch 3555/5000\n",
+      "919/919 - 3s - loss: 1.2617 - accuracy: 0.6027 - val_loss: 3.9116 - val_accuracy: 0.0835\n",
+      "Epoch 3556/5000\n",
+      "919/919 - 3s - loss: 1.2891 - accuracy: 0.6061 - val_loss: 3.9151 - val_accuracy: 0.0830\n",
+      "Epoch 3557/5000\n",
+      "919/919 - 3s - loss: 1.2703 - accuracy: 0.6050 - val_loss: 3.9169 - val_accuracy: 0.0838\n",
+      "Epoch 3558/5000\n",
+      "919/919 - 3s - loss: 1.2611 - accuracy: 0.6035 - val_loss: 3.9244 - val_accuracy: 0.0839\n",
+      "Epoch 3559/5000\n",
+      "919/919 - 3s - loss: 1.2540 - accuracy: 0.6091 - val_loss: 3.9367 - val_accuracy: 0.0838\n",
+      "Epoch 3560/5000\n",
+      "919/919 - 3s - loss: 1.2524 - accuracy: 0.6069 - val_loss: 3.9340 - val_accuracy: 0.0841\n",
+      "Epoch 3561/5000\n",
+      "919/919 - 3s - loss: 1.2668 - accuracy: 0.6017 - val_loss: 3.9319 - val_accuracy: 0.0843\n",
+      "Epoch 3562/5000\n",
+      "919/919 - 3s - loss: 1.2803 - accuracy: 0.6069 - val_loss: 3.9241 - val_accuracy: 0.0845\n",
+      "Epoch 3563/5000\n",
+      "919/919 - 3s - loss: 1.2915 - accuracy: 0.6063 - val_loss: 3.9238 - val_accuracy: 0.0842\n",
+      "Epoch 3564/5000\n",
+      "919/919 - 3s - loss: 1.2657 - accuracy: 0.6031 - val_loss: 3.9132 - val_accuracy: 0.0839\n",
+      "Epoch 3565/5000\n",
+      "919/919 - 3s - loss: 1.2593 - accuracy: 0.6039 - val_loss: 3.9065 - val_accuracy: 0.0839\n",
+      "Epoch 3566/5000\n",
+      "919/919 - 3s - loss: 1.3266 - accuracy: 0.6035 - val_loss: 3.9031 - val_accuracy: 0.0833\n",
+      "Epoch 3567/5000\n",
+      "919/919 - 3s - loss: 1.2663 - accuracy: 0.6040 - val_loss: 3.9041 - val_accuracy: 0.0831\n",
+      "Epoch 3568/5000\n",
+      "919/919 - 3s - loss: 1.2729 - accuracy: 0.6024 - val_loss: 3.9046 - val_accuracy: 0.0831\n",
+      "Epoch 3569/5000\n",
+      "919/919 - 3s - loss: 1.2511 - accuracy: 0.6039 - val_loss: 3.9085 - val_accuracy: 0.0834\n",
+      "Epoch 3570/5000\n",
+      "919/919 - 3s - loss: 1.2547 - accuracy: 0.6046 - val_loss: 3.9215 - val_accuracy: 0.0840\n",
+      "Epoch 3571/5000\n",
+      "919/919 - 3s - loss: 1.2621 - accuracy: 0.6037 - val_loss: 3.9223 - val_accuracy: 0.0829\n",
+      "Epoch 3572/5000\n",
+      "919/919 - 3s - loss: 1.3411 - accuracy: 0.6067 - val_loss: 3.9191 - val_accuracy: 0.0837\n",
+      "Epoch 3573/5000\n",
+      "919/919 - 3s - loss: 1.2713 - accuracy: 0.6016 - val_loss: 3.9203 - val_accuracy: 0.0836\n",
+      "Epoch 3574/5000\n",
+      "919/919 - 3s - loss: 1.2384 - accuracy: 0.6098 - val_loss: 3.9249 - val_accuracy: 0.0837\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3575/5000\n",
+      "919/919 - 3s - loss: 1.3401 - accuracy: 0.6059 - val_loss: 3.9264 - val_accuracy: 0.0838\n",
+      "Epoch 3576/5000\n",
+      "919/919 - 3s - loss: 1.2738 - accuracy: 0.6035 - val_loss: 3.9196 - val_accuracy: 0.0841\n",
+      "Epoch 3577/5000\n",
+      "919/919 - 3s - loss: 1.3484 - accuracy: 0.6018 - val_loss: 3.9288 - val_accuracy: 0.0839\n",
+      "Epoch 3578/5000\n",
+      "919/919 - 3s - loss: 1.2576 - accuracy: 0.6051 - val_loss: 3.9277 - val_accuracy: 0.0837\n",
+      "Epoch 3579/5000\n",
+      "919/919 - 3s - loss: 1.2692 - accuracy: 0.5999 - val_loss: 3.9264 - val_accuracy: 0.0834\n",
+      "Epoch 3580/5000\n",
+      "919/919 - 3s - loss: 1.2350 - accuracy: 0.6066 - val_loss: 3.9292 - val_accuracy: 0.0829\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3581/5000\n",
+      "919/919 - 3s - loss: 1.2576 - accuracy: 0.6069 - val_loss: 3.9193 - val_accuracy: 0.0839\n",
+      "Epoch 3582/5000\n",
+      "919/919 - 3s - loss: 1.2592 - accuracy: 0.6090 - val_loss: 3.9247 - val_accuracy: 0.0839\n",
+      "Epoch 3583/5000\n",
+      "919/919 - 3s - loss: 1.2561 - accuracy: 0.6081 - val_loss: 3.9200 - val_accuracy: 0.0841\n",
+      "Epoch 3584/5000\n",
+      "919/919 - 3s - loss: 1.2650 - accuracy: 0.6058 - val_loss: 3.9175 - val_accuracy: 0.0842\n",
+      "Epoch 3585/5000\n",
+      "919/919 - 3s - loss: 1.2587 - accuracy: 0.6048 - val_loss: 3.9240 - val_accuracy: 0.0836\n",
+      "Epoch 3586/5000\n",
+      "919/919 - 3s - loss: 1.2606 - accuracy: 0.6086 - val_loss: 3.9172 - val_accuracy: 0.0836\n",
+      "Epoch 3587/5000\n",
+      "919/919 - 3s - loss: 1.2525 - accuracy: 0.6050 - val_loss: 3.9167 - val_accuracy: 0.0839\n",
+      "Epoch 3588/5000\n",
+      "919/919 - 3s - loss: 1.3225 - accuracy: 0.6012 - val_loss: 3.9102 - val_accuracy: 0.0839\n",
+      "Epoch 3589/5000\n",
+      "919/919 - 3s - loss: 1.2643 - accuracy: 0.6021 - val_loss: 3.9064 - val_accuracy: 0.0839\n",
+      "Epoch 3590/5000\n",
+      "919/919 - 3s - loss: 1.2587 - accuracy: 0.6028 - val_loss: 3.9158 - val_accuracy: 0.0840\n",
+      "Epoch 3591/5000\n",
+      "919/919 - 3s - loss: 1.3042 - accuracy: 0.6063 - val_loss: 3.9168 - val_accuracy: 0.0851\n",
+      "Epoch 3592/5000\n",
+      "919/919 - 3s - loss: 1.2567 - accuracy: 0.6050 - val_loss: 3.9199 - val_accuracy: 0.0840\n",
+      "Epoch 3593/5000\n",
+      "919/919 - 3s - loss: 1.3599 - accuracy: 0.6056 - val_loss: 3.9341 - val_accuracy: 0.0832\n",
+      "Epoch 3594/5000\n",
+      "919/919 - 3s - loss: 1.2589 - accuracy: 0.6033 - val_loss: 3.9253 - val_accuracy: 0.0843\n",
+      "Epoch 3595/5000\n",
+      "919/919 - 3s - loss: 1.2785 - accuracy: 0.6053 - val_loss: 3.9211 - val_accuracy: 0.0844\n",
+      "Epoch 3596/5000\n",
+      "919/919 - 3s - loss: 1.2782 - accuracy: 0.6007 - val_loss: 3.9362 - val_accuracy: 0.0842\n",
+      "Epoch 3597/5000\n",
+      "919/919 - 3s - loss: 1.2981 - accuracy: 0.6029 - val_loss: 3.9256 - val_accuracy: 0.0846\n",
+      "Epoch 3598/5000\n",
+      "919/919 - 3s - loss: 1.4629 - accuracy: 0.6041 - val_loss: 3.9265 - val_accuracy: 0.0842\n",
+      "Epoch 3599/5000\n",
+      "919/919 - 3s - loss: 1.2496 - accuracy: 0.6065 - val_loss: 3.9328 - val_accuracy: 0.0846\n",
+      "Epoch 3600/5000\n",
+      "919/919 - 3s - loss: 1.3102 - accuracy: 0.6016 - val_loss: 3.9311 - val_accuracy: 0.0848\n",
+      "Epoch 3601/5000\n",
+      "919/919 - 3s - loss: 1.3098 - accuracy: 0.5984 - val_loss: 3.9237 - val_accuracy: 0.0839\n",
+      "Epoch 3602/5000\n",
+      "919/919 - 3s - loss: 1.2516 - accuracy: 0.6071 - val_loss: 3.9332 - val_accuracy: 0.0839\n",
+      "Epoch 3603/5000\n",
+      "919/919 - 3s - loss: 1.2540 - accuracy: 0.6045 - val_loss: 3.9319 - val_accuracy: 0.0841\n",
+      "Epoch 3604/5000\n",
+      "919/919 - 3s - loss: 1.2903 - accuracy: 0.6093 - val_loss: 3.9295 - val_accuracy: 0.0840\n",
+      "Epoch 3605/5000\n",
+      "919/919 - 3s - loss: 1.2874 - accuracy: 0.6047 - val_loss: 3.9299 - val_accuracy: 0.0843\n",
+      "Epoch 3606/5000\n",
+      "919/919 - 3s - loss: 1.2493 - accuracy: 0.6043 - val_loss: 3.9381 - val_accuracy: 0.0840\n",
+      "Epoch 3607/5000\n",
+      "919/919 - 3s - loss: 1.2483 - accuracy: 0.6097 - val_loss: 3.9265 - val_accuracy: 0.0841\n",
+      "Epoch 3608/5000\n",
+      "919/919 - 3s - loss: 1.2423 - accuracy: 0.6067 - val_loss: 3.9196 - val_accuracy: 0.0846\n",
+      "Epoch 3609/5000\n",
+      "919/919 - 3s - loss: 1.2518 - accuracy: 0.6054 - val_loss: 3.9343 - val_accuracy: 0.0841\n",
+      "Epoch 3610/5000\n",
+      "919/919 - 3s - loss: 1.2591 - accuracy: 0.6058 - val_loss: 3.9325 - val_accuracy: 0.0843\n",
+      "Epoch 3611/5000\n",
+      "919/919 - 3s - loss: 1.2992 - accuracy: 0.6042 - val_loss: 3.9234 - val_accuracy: 0.0838\n",
+      "Epoch 3612/5000\n",
+      "919/919 - 3s - loss: 1.2712 - accuracy: 0.6016 - val_loss: 3.9214 - val_accuracy: 0.0835\n",
+      "Epoch 3613/5000\n",
+      "919/919 - 3s - loss: 1.2898 - accuracy: 0.6034 - val_loss: 3.9224 - val_accuracy: 0.0833\n",
+      "Epoch 3614/5000\n",
+      "919/919 - 3s - loss: 1.2663 - accuracy: 0.6010 - val_loss: 3.9246 - val_accuracy: 0.0833\n",
+      "Epoch 3615/5000\n",
+      "919/919 - 3s - loss: 1.2607 - accuracy: 0.6039 - val_loss: 3.9282 - val_accuracy: 0.0840\n",
+      "Epoch 3616/5000\n",
+      "919/919 - 3s - loss: 1.2602 - accuracy: 0.6061 - val_loss: 3.9331 - val_accuracy: 0.0833\n",
+      "Epoch 3617/5000\n",
+      "919/919 - 3s - loss: 1.2663 - accuracy: 0.6001 - val_loss: 3.9370 - val_accuracy: 0.0835\n",
+      "Epoch 3618/5000\n",
+      "919/919 - 3s - loss: 1.2793 - accuracy: 0.6029 - val_loss: 3.9291 - val_accuracy: 0.0828\n",
+      "Epoch 3619/5000\n",
+      "919/919 - 3s - loss: 1.3071 - accuracy: 0.6071 - val_loss: 3.9460 - val_accuracy: 0.0834\n",
+      "Epoch 3620/5000\n",
+      "919/919 - 3s - loss: 1.2410 - accuracy: 0.6085 - val_loss: 3.9478 - val_accuracy: 0.0830\n",
+      "Epoch 3621/5000\n",
+      "919/919 - 3s - loss: 1.3362 - accuracy: 0.6052 - val_loss: 3.9467 - val_accuracy: 0.0833\n",
+      "Epoch 3622/5000\n",
+      "919/919 - 3s - loss: 1.2623 - accuracy: 0.6032 - val_loss: 3.9568 - val_accuracy: 0.0839\n",
+      "Epoch 3623/5000\n",
+      "919/919 - 3s - loss: 1.2518 - accuracy: 0.6069 - val_loss: 3.9601 - val_accuracy: 0.0844\n",
+      "Epoch 3624/5000\n",
+      "919/919 - 3s - loss: 1.2488 - accuracy: 0.6087 - val_loss: 3.9535 - val_accuracy: 0.0837\n",
+      "Epoch 3625/5000\n",
+      "919/919 - 3s - loss: 1.2711 - accuracy: 0.6026 - val_loss: 3.9416 - val_accuracy: 0.0836\n",
+      "Epoch 3626/5000\n",
+      "919/919 - 3s - loss: 1.2510 - accuracy: 0.6081 - val_loss: 3.9436 - val_accuracy: 0.0838\n",
+      "Epoch 3627/5000\n",
+      "919/919 - 3s - loss: 1.2587 - accuracy: 0.6022 - val_loss: 3.9424 - val_accuracy: 0.0842\n",
+      "Epoch 3628/5000\n",
+      "919/919 - 3s - loss: 1.2402 - accuracy: 0.6101 - val_loss: 3.9331 - val_accuracy: 0.0844\n",
+      "Epoch 3629/5000\n",
+      "919/919 - 3s - loss: 1.2573 - accuracy: 0.6047 - val_loss: 3.9531 - val_accuracy: 0.0842\n",
+      "Epoch 3630/5000\n",
+      "919/919 - 3s - loss: 1.2987 - accuracy: 0.6107 - val_loss: 3.9579 - val_accuracy: 0.0847\n",
+      "Epoch 3631/5000\n",
+      "919/919 - 3s - loss: 1.2488 - accuracy: 0.6056 - val_loss: 3.9601 - val_accuracy: 0.0849\n",
+      "Epoch 3632/5000\n",
+      "919/919 - 3s - loss: 1.2727 - accuracy: 0.6050 - val_loss: 3.9582 - val_accuracy: 0.0836\n",
+      "Epoch 3633/5000\n",
+      "919/919 - 3s - loss: 1.2545 - accuracy: 0.6064 - val_loss: 3.9683 - val_accuracy: 0.0843\n",
+      "Epoch 3634/5000\n",
+      "919/919 - 3s - loss: 1.2459 - accuracy: 0.6086 - val_loss: 3.9571 - val_accuracy: 0.0838\n",
+      "Epoch 3635/5000\n",
+      "919/919 - 3s - loss: 1.2485 - accuracy: 0.6060 - val_loss: 3.9620 - val_accuracy: 0.0839\n",
+      "Epoch 3636/5000\n",
+      "919/919 - 3s - loss: 1.2414 - accuracy: 0.6108 - val_loss: 3.9562 - val_accuracy: 0.0839\n",
+      "Epoch 3637/5000\n",
+      "919/919 - 3s - loss: 1.2486 - accuracy: 0.6048 - val_loss: 3.9623 - val_accuracy: 0.0838\n",
+      "Epoch 3638/5000\n",
+      "919/919 - 3s - loss: 1.2602 - accuracy: 0.6056 - val_loss: 3.9540 - val_accuracy: 0.0842\n",
+      "Epoch 3639/5000\n",
+      "919/919 - 3s - loss: 1.2456 - accuracy: 0.6080 - val_loss: 3.9536 - val_accuracy: 0.0843\n",
+      "Epoch 3640/5000\n",
+      "919/919 - 3s - loss: 1.3167 - accuracy: 0.6059 - val_loss: 3.9602 - val_accuracy: 0.0840\n",
+      "Epoch 3641/5000\n",
+      "919/919 - 3s - loss: 1.2889 - accuracy: 0.6118 - val_loss: 3.9596 - val_accuracy: 0.0848\n",
+      "Epoch 3642/5000\n",
+      "919/919 - 3s - loss: 1.2618 - accuracy: 0.6076 - val_loss: 3.9613 - val_accuracy: 0.0848\n",
+      "Epoch 3643/5000\n",
+      "919/919 - 3s - loss: 1.2666 - accuracy: 0.6047 - val_loss: 3.9489 - val_accuracy: 0.0839\n",
+      "Epoch 3644/5000\n",
+      "919/919 - 3s - loss: 1.2618 - accuracy: 0.6034 - val_loss: 3.9612 - val_accuracy: 0.0844\n",
+      "Epoch 3645/5000\n",
+      "919/919 - 3s - loss: 1.2561 - accuracy: 0.6052 - val_loss: 3.9492 - val_accuracy: 0.0839\n",
+      "Epoch 3646/5000\n",
+      "919/919 - 3s - loss: 1.2548 - accuracy: 0.6093 - val_loss: 3.9534 - val_accuracy: 0.0843\n",
+      "Epoch 3647/5000\n",
+      "919/919 - 3s - loss: 1.2555 - accuracy: 0.6069 - val_loss: 3.9548 - val_accuracy: 0.0846\n",
+      "Epoch 3648/5000\n",
+      "919/919 - 3s - loss: 1.2378 - accuracy: 0.6083 - val_loss: 3.9489 - val_accuracy: 0.0844\n",
+      "Epoch 3649/5000\n",
+      "919/919 - 3s - loss: 1.2725 - accuracy: 0.6025 - val_loss: 3.9451 - val_accuracy: 0.0842\n",
+      "Epoch 3650/5000\n",
+      "919/919 - 3s - loss: 1.2350 - accuracy: 0.6105 - val_loss: 3.9707 - val_accuracy: 0.0847\n",
+      "Epoch 3651/5000\n",
+      "919/919 - 3s - loss: 1.2425 - accuracy: 0.6082 - val_loss: 3.9568 - val_accuracy: 0.0851\n",
+      "Epoch 3652/5000\n",
+      "919/919 - 3s - loss: 1.2689 - accuracy: 0.6087 - val_loss: 3.9398 - val_accuracy: 0.0847\n",
+      "Epoch 3653/5000\n",
+      "919/919 - 3s - loss: 1.2586 - accuracy: 0.6086 - val_loss: 3.9437 - val_accuracy: 0.0846\n",
+      "Epoch 3654/5000\n",
+      "919/919 - 3s - loss: 1.2540 - accuracy: 0.6071 - val_loss: 3.9523 - val_accuracy: 0.0843\n",
+      "Epoch 3655/5000\n",
+      "919/919 - 3s - loss: 1.3700 - accuracy: 0.6087 - val_loss: 3.9510 - val_accuracy: 0.0850\n",
+      "Epoch 3656/5000\n",
+      "919/919 - 3s - loss: 1.3108 - accuracy: 0.6115 - val_loss: 3.9651 - val_accuracy: 0.0848\n",
+      "Epoch 3657/5000\n",
+      "919/919 - 3s - loss: 1.3021 - accuracy: 0.6059 - val_loss: 3.9458 - val_accuracy: 0.0846\n",
+      "Epoch 3658/5000\n",
+      "919/919 - 3s - loss: 1.2355 - accuracy: 0.6090 - val_loss: 3.9482 - val_accuracy: 0.0843\n",
+      "Epoch 3659/5000\n",
+      "919/919 - 3s - loss: 1.2405 - accuracy: 0.6084 - val_loss: 3.9415 - val_accuracy: 0.0843\n",
+      "Epoch 3660/5000\n",
+      "919/919 - 3s - loss: 1.2477 - accuracy: 0.6085 - val_loss: 3.9477 - val_accuracy: 0.0844\n",
+      "Epoch 3661/5000\n",
+      "919/919 - 3s - loss: 1.2689 - accuracy: 0.6088 - val_loss: 3.9474 - val_accuracy: 0.0840\n",
+      "Epoch 3662/5000\n",
+      "919/919 - 3s - loss: 1.2518 - accuracy: 0.6055 - val_loss: 3.9524 - val_accuracy: 0.0842\n",
+      "Epoch 3663/5000\n",
+      "919/919 - 3s - loss: 1.2430 - accuracy: 0.6108 - val_loss: 3.9550 - val_accuracy: 0.0844\n",
+      "Epoch 3664/5000\n",
+      "919/919 - 3s - loss: 1.2411 - accuracy: 0.6078 - val_loss: 3.9687 - val_accuracy: 0.0847\n",
+      "Epoch 3665/5000\n",
+      "919/919 - 3s - loss: 1.2542 - accuracy: 0.6071 - val_loss: 3.9678 - val_accuracy: 0.0848\n",
+      "Epoch 3666/5000\n",
+      "919/919 - 3s - loss: 1.2649 - accuracy: 0.6055 - val_loss: 3.9676 - val_accuracy: 0.0849\n",
+      "Epoch 3667/5000\n",
+      "919/919 - 3s - loss: 1.2521 - accuracy: 0.6073 - val_loss: 3.9584 - val_accuracy: 0.0844\n",
+      "Epoch 3668/5000\n",
+      "919/919 - 3s - loss: 1.2525 - accuracy: 0.6016 - val_loss: 3.9663 - val_accuracy: 0.0846\n",
+      "Epoch 3669/5000\n",
+      "919/919 - 3s - loss: 1.4816 - accuracy: 0.6103 - val_loss: 3.9628 - val_accuracy: 0.0850\n",
+      "Epoch 3670/5000\n",
+      "919/919 - 3s - loss: 1.2447 - accuracy: 0.6114 - val_loss: 3.9643 - val_accuracy: 0.0857\n",
+      "Epoch 3671/5000\n",
+      "919/919 - 3s - loss: 1.2461 - accuracy: 0.6114 - val_loss: 3.9708 - val_accuracy: 0.0854\n",
+      "Epoch 3672/5000\n",
+      "919/919 - 3s - loss: 1.3146 - accuracy: 0.6067 - val_loss: 3.9664 - val_accuracy: 0.0850\n",
+      "Epoch 3673/5000\n",
+      "919/919 - 3s - loss: 1.2663 - accuracy: 0.6106 - val_loss: 3.9654 - val_accuracy: 0.0853\n",
+      "Epoch 3674/5000\n",
+      "919/919 - 3s - loss: 1.2518 - accuracy: 0.6085 - val_loss: 3.9585 - val_accuracy: 0.0849\n",
+      "Epoch 3675/5000\n",
+      "919/919 - 3s - loss: 1.2580 - accuracy: 0.6051 - val_loss: 3.9531 - val_accuracy: 0.0848\n",
+      "Epoch 3676/5000\n",
+      "919/919 - 3s - loss: 1.2608 - accuracy: 0.6056 - val_loss: 3.9631 - val_accuracy: 0.0849\n",
+      "Epoch 3677/5000\n",
+      "919/919 - 3s - loss: 1.2607 - accuracy: 0.6080 - val_loss: 3.9671 - val_accuracy: 0.0852\n",
+      "Epoch 3678/5000\n",
+      "919/919 - 3s - loss: 1.2497 - accuracy: 0.6104 - val_loss: 3.9687 - val_accuracy: 0.0849\n",
+      "Epoch 3679/5000\n",
+      "919/919 - 3s - loss: 1.2474 - accuracy: 0.6036 - val_loss: 3.9803 - val_accuracy: 0.0847\n",
+      "Epoch 3680/5000\n",
+      "919/919 - 3s - loss: 1.2417 - accuracy: 0.6078 - val_loss: 3.9921 - val_accuracy: 0.0847\n",
+      "Epoch 3681/5000\n",
+      "919/919 - 3s - loss: 1.2455 - accuracy: 0.6079 - val_loss: 3.9878 - val_accuracy: 0.0847\n",
+      "Epoch 3682/5000\n",
+      "919/919 - 3s - loss: 1.2498 - accuracy: 0.6092 - val_loss: 3.9814 - val_accuracy: 0.0843\n",
+      "Epoch 3683/5000\n",
+      "919/919 - 3s - loss: 1.2570 - accuracy: 0.6071 - val_loss: 3.9797 - val_accuracy: 0.0840\n",
+      "Epoch 3684/5000\n",
+      "919/919 - 3s - loss: 1.3134 - accuracy: 0.6084 - val_loss: 3.9793 - val_accuracy: 0.0844\n",
+      "Epoch 3685/5000\n",
+      "919/919 - 3s - loss: 1.2431 - accuracy: 0.6109 - val_loss: 3.9797 - val_accuracy: 0.0844\n",
+      "Epoch 3686/5000\n",
+      "919/919 - 3s - loss: 1.2573 - accuracy: 0.6073 - val_loss: 3.9689 - val_accuracy: 0.0840\n",
+      "Epoch 3687/5000\n",
+      "919/919 - 3s - loss: 1.2544 - accuracy: 0.6063 - val_loss: 3.9804 - val_accuracy: 0.0844\n",
+      "Epoch 3688/5000\n",
+      "919/919 - 3s - loss: 1.2825 - accuracy: 0.6029 - val_loss: 3.9752 - val_accuracy: 0.0842\n",
+      "Epoch 3689/5000\n",
+      "919/919 - 3s - loss: 1.2438 - accuracy: 0.6031 - val_loss: 3.9814 - val_accuracy: 0.0844\n",
+      "Epoch 3690/5000\n",
+      "919/919 - 3s - loss: 1.2804 - accuracy: 0.6066 - val_loss: 3.9737 - val_accuracy: 0.0845\n",
+      "Epoch 3691/5000\n",
+      "919/919 - 3s - loss: 1.2548 - accuracy: 0.6078 - val_loss: 3.9789 - val_accuracy: 0.0842\n",
+      "Epoch 3692/5000\n",
+      "919/919 - 3s - loss: 1.2510 - accuracy: 0.6100 - val_loss: 3.9633 - val_accuracy: 0.0850\n",
+      "Epoch 3693/5000\n",
+      "919/919 - 3s - loss: 1.2518 - accuracy: 0.6038 - val_loss: 3.9596 - val_accuracy: 0.0850\n",
+      "Epoch 3694/5000\n",
+      "919/919 - 3s - loss: 1.2477 - accuracy: 0.6077 - val_loss: 3.9687 - val_accuracy: 0.0842\n",
+      "Epoch 3695/5000\n",
+      "919/919 - 3s - loss: 1.2461 - accuracy: 0.6116 - val_loss: 3.9757 - val_accuracy: 0.0842\n",
+      "Epoch 3696/5000\n",
+      "919/919 - 3s - loss: 1.2470 - accuracy: 0.6082 - val_loss: 3.9789 - val_accuracy: 0.0845\n",
+      "Epoch 3697/5000\n",
+      "919/919 - 3s - loss: 1.2392 - accuracy: 0.6097 - val_loss: 3.9964 - val_accuracy: 0.0842\n",
+      "Epoch 3698/5000\n",
+      "919/919 - 3s - loss: 1.2519 - accuracy: 0.6065 - val_loss: 3.9882 - val_accuracy: 0.0841\n",
+      "Epoch 3699/5000\n",
+      "919/919 - 3s - loss: 1.3029 - accuracy: 0.6133 - val_loss: 3.9861 - val_accuracy: 0.0855\n",
+      "Epoch 3700/5000\n",
+      "919/919 - 3s - loss: 1.2563 - accuracy: 0.6080 - val_loss: 3.9769 - val_accuracy: 0.0848\n",
+      "Epoch 3701/5000\n",
+      "919/919 - 3s - loss: 1.2447 - accuracy: 0.6072 - val_loss: 3.9761 - val_accuracy: 0.0851\n",
+      "Epoch 3702/5000\n",
+      "919/919 - 3s - loss: 1.2520 - accuracy: 0.6067 - val_loss: 3.9786 - val_accuracy: 0.0841\n",
+      "Epoch 3703/5000\n",
+      "919/919 - 3s - loss: 1.2691 - accuracy: 0.6059 - val_loss: 3.9729 - val_accuracy: 0.0841\n",
+      "Epoch 3704/5000\n",
+      "919/919 - 3s - loss: 1.2648 - accuracy: 0.6052 - val_loss: 3.9651 - val_accuracy: 0.0838\n",
+      "Epoch 3705/5000\n",
+      "919/919 - 3s - loss: 1.2514 - accuracy: 0.6086 - val_loss: 3.9724 - val_accuracy: 0.0846\n",
+      "Epoch 3706/5000\n",
+      "919/919 - 3s - loss: 1.2406 - accuracy: 0.6101 - val_loss: 3.9704 - val_accuracy: 0.0851\n",
+      "Epoch 3707/5000\n",
+      "919/919 - 3s - loss: 1.2404 - accuracy: 0.6097 - val_loss: 3.9703 - val_accuracy: 0.0848\n",
+      "Epoch 3708/5000\n",
+      "919/919 - 3s - loss: 1.2562 - accuracy: 0.6091 - val_loss: 3.9682 - val_accuracy: 0.0851\n",
+      "Epoch 3709/5000\n",
+      "919/919 - 3s - loss: 1.2567 - accuracy: 0.6079 - val_loss: 3.9633 - val_accuracy: 0.0846\n",
+      "Epoch 3710/5000\n",
+      "919/919 - 3s - loss: 1.3668 - accuracy: 0.6090 - val_loss: 3.9667 - val_accuracy: 0.0846\n",
+      "Epoch 3711/5000\n",
+      "919/919 - 3s - loss: 1.2364 - accuracy: 0.6081 - val_loss: 3.9719 - val_accuracy: 0.0847\n",
+      "Epoch 3712/5000\n",
+      "919/919 - 3s - loss: 1.2370 - accuracy: 0.6117 - val_loss: 3.9689 - val_accuracy: 0.0845\n",
+      "Epoch 3713/5000\n",
+      "919/919 - 3s - loss: 1.2449 - accuracy: 0.6084 - val_loss: 3.9634 - val_accuracy: 0.0845\n",
+      "Epoch 3714/5000\n",
+      "919/919 - 3s - loss: 1.2821 - accuracy: 0.6111 - val_loss: 3.9821 - val_accuracy: 0.0846\n",
+      "Epoch 3715/5000\n",
+      "919/919 - 3s - loss: 1.2432 - accuracy: 0.6105 - val_loss: 3.9788 - val_accuracy: 0.0848\n",
+      "Epoch 3716/5000\n",
+      "919/919 - 3s - loss: 1.2412 - accuracy: 0.6096 - val_loss: 3.9770 - val_accuracy: 0.0853\n",
+      "Epoch 3717/5000\n",
+      "919/919 - 3s - loss: 1.2467 - accuracy: 0.6082 - val_loss: 3.9818 - val_accuracy: 0.0853\n",
+      "Epoch 3718/5000\n",
+      "919/919 - 3s - loss: 1.2389 - accuracy: 0.6132 - val_loss: 3.9865 - val_accuracy: 0.0855\n",
+      "Epoch 3719/5000\n",
+      "919/919 - 3s - loss: 1.3052 - accuracy: 0.6093 - val_loss: 3.9798 - val_accuracy: 0.0852\n",
+      "Epoch 3720/5000\n",
+      "919/919 - 3s - loss: 1.2414 - accuracy: 0.6093 - val_loss: 3.9829 - val_accuracy: 0.0848\n",
+      "Epoch 3721/5000\n",
+      "919/919 - 3s - loss: 1.2419 - accuracy: 0.6104 - val_loss: 3.9849 - val_accuracy: 0.0846\n",
+      "Epoch 3722/5000\n",
+      "919/919 - 3s - loss: 1.2503 - accuracy: 0.6101 - val_loss: 3.9872 - val_accuracy: 0.0850\n",
+      "Epoch 3723/5000\n",
+      "919/919 - 3s - loss: 1.2433 - accuracy: 0.6156 - val_loss: 3.9936 - val_accuracy: 0.0852\n",
+      "Epoch 3724/5000\n",
+      "919/919 - 3s - loss: 1.2445 - accuracy: 0.6086 - val_loss: 3.9896 - val_accuracy: 0.0848\n",
+      "Epoch 3725/5000\n",
+      "919/919 - 3s - loss: 1.2498 - accuracy: 0.6099 - val_loss: 3.9745 - val_accuracy: 0.0854\n",
+      "Epoch 3726/5000\n",
+      "919/919 - 3s - loss: 1.2470 - accuracy: 0.6103 - val_loss: 3.9766 - val_accuracy: 0.0855\n",
+      "Epoch 3727/5000\n",
+      "919/919 - 3s - loss: 1.2530 - accuracy: 0.6122 - val_loss: 3.9706 - val_accuracy: 0.0853\n",
+      "Epoch 3728/5000\n",
+      "919/919 - 3s - loss: 1.2306 - accuracy: 0.6117 - val_loss: 3.9684 - val_accuracy: 0.0855\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3729/5000\n",
+      "919/919 - 3s - loss: 1.2455 - accuracy: 0.6079 - val_loss: 3.9793 - val_accuracy: 0.0849\n",
+      "Epoch 3730/5000\n",
+      "919/919 - 3s - loss: 1.2736 - accuracy: 0.6075 - val_loss: 3.9804 - val_accuracy: 0.0860\n",
+      "Epoch 3731/5000\n",
+      "919/919 - 3s - loss: 1.2515 - accuracy: 0.6089 - val_loss: 3.9813 - val_accuracy: 0.0859\n",
+      "Epoch 3732/5000\n",
+      "919/919 - 3s - loss: 1.2462 - accuracy: 0.6085 - val_loss: 3.9867 - val_accuracy: 0.0857\n",
+      "Epoch 3733/5000\n",
+      "919/919 - 3s - loss: 1.2348 - accuracy: 0.6114 - val_loss: 3.9789 - val_accuracy: 0.0848\n",
+      "Epoch 3734/5000\n",
+      "919/919 - 3s - loss: 1.2449 - accuracy: 0.6104 - val_loss: 3.9727 - val_accuracy: 0.0846\n",
+      "Epoch 3735/5000\n",
+      "919/919 - 3s - loss: 1.2472 - accuracy: 0.6094 - val_loss: 3.9685 - val_accuracy: 0.0847\n",
+      "Epoch 3736/5000\n",
+      "919/919 - 3s - loss: 1.2601 - accuracy: 0.6059 - val_loss: 3.9754 - val_accuracy: 0.0852\n",
+      "Epoch 3737/5000\n",
+      "919/919 - 3s - loss: 1.2503 - accuracy: 0.6052 - val_loss: 3.9834 - val_accuracy: 0.0847\n",
+      "Epoch 3738/5000\n",
+      "919/919 - 3s - loss: 1.2362 - accuracy: 0.6103 - val_loss: 3.9843 - val_accuracy: 0.0848\n",
+      "Epoch 3739/5000\n",
+      "919/919 - 3s - loss: 1.2429 - accuracy: 0.6076 - val_loss: 3.9735 - val_accuracy: 0.0849\n",
+      "Epoch 3740/5000\n",
+      "919/919 - 3s - loss: 1.2395 - accuracy: 0.6122 - val_loss: 3.9900 - val_accuracy: 0.0847\n",
+      "Epoch 3741/5000\n",
+      "919/919 - 3s - loss: 1.2685 - accuracy: 0.6058 - val_loss: 3.9915 - val_accuracy: 0.0845\n",
+      "Epoch 3742/5000\n",
+      "919/919 - 3s - loss: 1.2513 - accuracy: 0.6097 - val_loss: 3.9786 - val_accuracy: 0.0845\n",
+      "Epoch 3743/5000\n",
+      "919/919 - 3s - loss: 1.2391 - accuracy: 0.6102 - val_loss: 3.9780 - val_accuracy: 0.0848\n",
+      "Epoch 3744/5000\n",
+      "919/919 - 3s - loss: 1.2402 - accuracy: 0.6127 - val_loss: 3.9800 - val_accuracy: 0.0851\n",
+      "Epoch 3745/5000\n",
+      "919/919 - 3s - loss: 1.2977 - accuracy: 0.6103 - val_loss: 3.9844 - val_accuracy: 0.0851\n",
+      "Epoch 3746/5000\n",
+      "919/919 - 3s - loss: 1.2589 - accuracy: 0.6088 - val_loss: 3.9876 - val_accuracy: 0.0846\n",
+      "Epoch 3747/5000\n",
+      "919/919 - 3s - loss: 1.2752 - accuracy: 0.6059 - val_loss: 3.9918 - val_accuracy: 0.0841\n",
+      "Epoch 3748/5000\n",
+      "919/919 - 3s - loss: 1.3681 - accuracy: 0.6093 - val_loss: 3.9940 - val_accuracy: 0.0846\n",
+      "Epoch 3749/5000\n",
+      "919/919 - 3s - loss: 1.2695 - accuracy: 0.6073 - val_loss: 3.9865 - val_accuracy: 0.0842\n",
+      "Epoch 3750/5000\n",
+      "919/919 - 3s - loss: 1.2339 - accuracy: 0.6118 - val_loss: 3.9783 - val_accuracy: 0.0848\n",
+      "Epoch 3751/5000\n",
+      "919/919 - 3s - loss: 1.2436 - accuracy: 0.6085 - val_loss: 3.9805 - val_accuracy: 0.0849\n",
+      "Epoch 3752/5000\n",
+      "919/919 - 3s - loss: 1.2641 - accuracy: 0.6073 - val_loss: 3.9832 - val_accuracy: 0.0848\n",
+      "Epoch 3753/5000\n",
+      "919/919 - 3s - loss: 1.2527 - accuracy: 0.6066 - val_loss: 3.9899 - val_accuracy: 0.0844\n",
+      "Epoch 3754/5000\n",
+      "919/919 - 3s - loss: 1.2264 - accuracy: 0.6135 - val_loss: 3.9896 - val_accuracy: 0.0847\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3755/5000\n",
+      "919/919 - 3s - loss: 1.2466 - accuracy: 0.6117 - val_loss: 3.9830 - val_accuracy: 0.0845\n",
+      "Epoch 3756/5000\n",
+      "919/919 - 3s - loss: 1.2493 - accuracy: 0.6074 - val_loss: 3.9685 - val_accuracy: 0.0848\n",
+      "Epoch 3757/5000\n",
+      "919/919 - 3s - loss: 1.2368 - accuracy: 0.6123 - val_loss: 3.9852 - val_accuracy: 0.0849\n",
+      "Epoch 3758/5000\n",
+      "919/919 - 3s - loss: 1.2488 - accuracy: 0.6071 - val_loss: 3.9764 - val_accuracy: 0.0848\n",
+      "Epoch 3759/5000\n",
+      "919/919 - 3s - loss: 1.2514 - accuracy: 0.6031 - val_loss: 3.9749 - val_accuracy: 0.0844\n",
+      "Epoch 3760/5000\n",
+      "919/919 - 3s - loss: 1.2861 - accuracy: 0.6078 - val_loss: 3.9818 - val_accuracy: 0.0848\n",
+      "Epoch 3761/5000\n",
+      "919/919 - 3s - loss: 1.4930 - accuracy: 0.6107 - val_loss: 3.9967 - val_accuracy: 0.0848\n",
+      "Epoch 3762/5000\n",
+      "919/919 - 3s - loss: 1.2428 - accuracy: 0.6122 - val_loss: 3.9969 - val_accuracy: 0.0851\n",
+      "Epoch 3763/5000\n",
+      "919/919 - 3s - loss: 1.3503 - accuracy: 0.6105 - val_loss: 4.0122 - val_accuracy: 0.0848\n",
+      "Epoch 3764/5000\n",
+      "919/919 - 3s - loss: 1.2614 - accuracy: 0.6104 - val_loss: 4.0045 - val_accuracy: 0.0851\n",
+      "Epoch 3765/5000\n",
+      "919/919 - 3s - loss: 1.2652 - accuracy: 0.6088 - val_loss: 4.0095 - val_accuracy: 0.0848\n",
+      "Epoch 3766/5000\n",
+      "919/919 - 3s - loss: 1.2457 - accuracy: 0.6092 - val_loss: 3.9999 - val_accuracy: 0.0845\n",
+      "Epoch 3767/5000\n",
+      "919/919 - 3s - loss: 1.2491 - accuracy: 0.6070 - val_loss: 3.9805 - val_accuracy: 0.0851\n",
+      "Epoch 3768/5000\n",
+      "919/919 - 3s - loss: 1.2774 - accuracy: 0.6093 - val_loss: 3.9713 - val_accuracy: 0.0853\n",
+      "Epoch 3769/5000\n",
+      "919/919 - 3s - loss: 1.2319 - accuracy: 0.6116 - val_loss: 3.9730 - val_accuracy: 0.0858\n",
+      "Epoch 3770/5000\n",
+      "919/919 - 3s - loss: 1.2414 - accuracy: 0.6059 - val_loss: 3.9791 - val_accuracy: 0.0863\n",
+      "Epoch 3771/5000\n",
+      "919/919 - 3s - loss: 1.2763 - accuracy: 0.6117 - val_loss: 3.9888 - val_accuracy: 0.0857\n",
+      "Epoch 3772/5000\n",
+      "919/919 - 3s - loss: 1.2258 - accuracy: 0.6122 - val_loss: 3.9801 - val_accuracy: 0.0858\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3773/5000\n",
+      "919/919 - 3s - loss: 1.2362 - accuracy: 0.6112 - val_loss: 3.9744 - val_accuracy: 0.0858\n",
+      "Epoch 3774/5000\n",
+      "919/919 - 3s - loss: 1.2417 - accuracy: 0.6131 - val_loss: 3.9693 - val_accuracy: 0.0857\n",
+      "Epoch 3775/5000\n",
+      "919/919 - 3s - loss: 1.2455 - accuracy: 0.6084 - val_loss: 3.9763 - val_accuracy: 0.0859\n",
+      "Epoch 3776/5000\n",
+      "919/919 - 3s - loss: 1.2509 - accuracy: 0.6057 - val_loss: 3.9740 - val_accuracy: 0.0857\n",
+      "Epoch 3777/5000\n",
+      "919/919 - 3s - loss: 1.2444 - accuracy: 0.6115 - val_loss: 3.9673 - val_accuracy: 0.0853\n",
+      "Epoch 3778/5000\n",
+      "919/919 - 3s - loss: 1.2417 - accuracy: 0.6106 - val_loss: 3.9757 - val_accuracy: 0.0848\n",
+      "Epoch 3779/5000\n",
+      "919/919 - 3s - loss: 1.2397 - accuracy: 0.6094 - val_loss: 3.9913 - val_accuracy: 0.0851\n",
+      "Epoch 3780/5000\n",
+      "919/919 - 3s - loss: 1.2241 - accuracy: 0.6095 - val_loss: 3.9809 - val_accuracy: 0.0853\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3781/5000\n",
+      "919/919 - 3s - loss: 1.2549 - accuracy: 0.6080 - val_loss: 3.9753 - val_accuracy: 0.0851\n",
+      "Epoch 3782/5000\n",
+      "919/919 - 3s - loss: 1.2453 - accuracy: 0.6120 - val_loss: 3.9764 - val_accuracy: 0.0855\n",
+      "Epoch 3783/5000\n",
+      "919/919 - 3s - loss: 1.2307 - accuracy: 0.6112 - val_loss: 3.9850 - val_accuracy: 0.0857\n",
+      "Epoch 3784/5000\n",
+      "919/919 - 3s - loss: 1.2545 - accuracy: 0.6105 - val_loss: 3.9901 - val_accuracy: 0.0851\n",
+      "Epoch 3785/5000\n",
+      "919/919 - 3s - loss: 1.2217 - accuracy: 0.6144 - val_loss: 3.9976 - val_accuracy: 0.0852\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3786/5000\n",
+      "919/919 - 3s - loss: 1.2484 - accuracy: 0.6118 - val_loss: 3.9995 - val_accuracy: 0.0851\n",
+      "Epoch 3787/5000\n",
+      "919/919 - 3s - loss: 1.2719 - accuracy: 0.6108 - val_loss: 3.9933 - val_accuracy: 0.0856\n",
+      "Epoch 3788/5000\n",
+      "919/919 - 3s - loss: 1.2413 - accuracy: 0.6119 - val_loss: 3.9861 - val_accuracy: 0.0854\n",
+      "Epoch 3789/5000\n",
+      "919/919 - 3s - loss: 1.2400 - accuracy: 0.6128 - val_loss: 3.9809 - val_accuracy: 0.0851\n",
+      "Epoch 3790/5000\n",
+      "919/919 - 3s - loss: 1.2437 - accuracy: 0.6110 - val_loss: 3.9744 - val_accuracy: 0.0854\n",
+      "Epoch 3791/5000\n",
+      "919/919 - 3s - loss: 1.2410 - accuracy: 0.6103 - val_loss: 3.9894 - val_accuracy: 0.0851\n",
+      "Epoch 3792/5000\n",
+      "919/919 - 3s - loss: 1.2598 - accuracy: 0.6073 - val_loss: 4.0036 - val_accuracy: 0.0849\n",
+      "Epoch 3793/5000\n",
+      "919/919 - 3s - loss: 1.2465 - accuracy: 0.6063 - val_loss: 3.9862 - val_accuracy: 0.0854\n",
+      "Epoch 3794/5000\n",
+      "919/919 - 3s - loss: 1.2287 - accuracy: 0.6110 - val_loss: 3.9888 - val_accuracy: 0.0851\n",
+      "Epoch 3795/5000\n",
+      "919/919 - 3s - loss: 1.2588 - accuracy: 0.6097 - val_loss: 3.9902 - val_accuracy: 0.0856\n",
+      "Epoch 3796/5000\n",
+      "919/919 - 3s - loss: 1.2400 - accuracy: 0.6101 - val_loss: 3.9978 - val_accuracy: 0.0859\n",
+      "Epoch 3797/5000\n",
+      "919/919 - 3s - loss: 1.2532 - accuracy: 0.6106 - val_loss: 3.9895 - val_accuracy: 0.0857\n",
+      "Epoch 3798/5000\n",
+      "919/919 - 3s - loss: 1.2574 - accuracy: 0.6067 - val_loss: 3.9796 - val_accuracy: 0.0859\n",
+      "Epoch 3799/5000\n",
+      "919/919 - 3s - loss: 1.2494 - accuracy: 0.6088 - val_loss: 3.9878 - val_accuracy: 0.0849\n",
+      "Epoch 3800/5000\n",
+      "919/919 - 3s - loss: 1.2509 - accuracy: 0.6095 - val_loss: 3.9943 - val_accuracy: 0.0852\n",
+      "Epoch 3801/5000\n",
+      "919/919 - 3s - loss: 1.2445 - accuracy: 0.6097 - val_loss: 4.0030 - val_accuracy: 0.0848\n",
+      "Epoch 3802/5000\n",
+      "919/919 - 3s - loss: 1.2569 - accuracy: 0.6127 - val_loss: 3.9897 - val_accuracy: 0.0856\n",
+      "Epoch 3803/5000\n",
+      "919/919 - 3s - loss: 1.2449 - accuracy: 0.6124 - val_loss: 3.9893 - val_accuracy: 0.0863\n",
+      "Epoch 3804/5000\n",
+      "919/919 - 3s - loss: 1.2477 - accuracy: 0.6144 - val_loss: 3.9876 - val_accuracy: 0.0862\n",
+      "Epoch 3805/5000\n",
+      "919/919 - 3s - loss: 1.2585 - accuracy: 0.6038 - val_loss: 3.9897 - val_accuracy: 0.0859\n",
+      "Epoch 3806/5000\n",
+      "919/919 - 3s - loss: 1.2403 - accuracy: 0.6105 - val_loss: 3.9856 - val_accuracy: 0.0858\n",
+      "Epoch 3807/5000\n",
+      "919/919 - 3s - loss: 1.2383 - accuracy: 0.6093 - val_loss: 3.9903 - val_accuracy: 0.0857\n",
+      "Epoch 3808/5000\n",
+      "919/919 - 3s - loss: 1.2456 - accuracy: 0.6114 - val_loss: 4.0019 - val_accuracy: 0.0855\n",
+      "Epoch 3809/5000\n",
+      "919/919 - 3s - loss: 1.2535 - accuracy: 0.6113 - val_loss: 4.0079 - val_accuracy: 0.0854\n",
+      "Epoch 3810/5000\n",
+      "919/919 - 3s - loss: 1.2413 - accuracy: 0.6112 - val_loss: 4.0028 - val_accuracy: 0.0859\n",
+      "Epoch 3811/5000\n",
+      "919/919 - 3s - loss: 1.2320 - accuracy: 0.6119 - val_loss: 4.0142 - val_accuracy: 0.0860\n",
+      "Epoch 3812/5000\n",
+      "919/919 - 3s - loss: 1.2356 - accuracy: 0.6138 - val_loss: 4.0053 - val_accuracy: 0.0858\n",
+      "Epoch 3813/5000\n",
+      "919/919 - 3s - loss: 1.2440 - accuracy: 0.6098 - val_loss: 4.0064 - val_accuracy: 0.0862\n",
+      "Epoch 3814/5000\n",
+      "919/919 - 3s - loss: 1.2366 - accuracy: 0.6136 - val_loss: 3.9907 - val_accuracy: 0.0857\n",
+      "Epoch 3815/5000\n",
+      "919/919 - 3s - loss: 1.2315 - accuracy: 0.6116 - val_loss: 3.9999 - val_accuracy: 0.0863\n",
+      "Epoch 3816/5000\n",
+      "919/919 - 3s - loss: 1.2329 - accuracy: 0.6118 - val_loss: 3.9959 - val_accuracy: 0.0853\n",
+      "Epoch 3817/5000\n",
+      "919/919 - 3s - loss: 1.3272 - accuracy: 0.6092 - val_loss: 4.0056 - val_accuracy: 0.0853\n",
+      "Epoch 3818/5000\n",
+      "919/919 - 3s - loss: 1.2450 - accuracy: 0.6122 - val_loss: 3.9920 - val_accuracy: 0.0854\n",
+      "Epoch 3819/5000\n",
+      "919/919 - 3s - loss: 1.2544 - accuracy: 0.6067 - val_loss: 3.9881 - val_accuracy: 0.0862\n",
+      "Epoch 3820/5000\n",
+      "919/919 - 3s - loss: 1.2385 - accuracy: 0.6104 - val_loss: 3.9899 - val_accuracy: 0.0856\n",
+      "Epoch 3821/5000\n",
+      "919/919 - 3s - loss: 1.2355 - accuracy: 0.6082 - val_loss: 3.9909 - val_accuracy: 0.0856\n",
+      "Epoch 3822/5000\n",
+      "919/919 - 3s - loss: 1.2563 - accuracy: 0.6117 - val_loss: 3.9865 - val_accuracy: 0.0856\n",
+      "Epoch 3823/5000\n",
+      "919/919 - 3s - loss: 1.2976 - accuracy: 0.6123 - val_loss: 3.9843 - val_accuracy: 0.0853\n",
+      "Epoch 3824/5000\n",
+      "919/919 - 3s - loss: 1.2367 - accuracy: 0.6113 - val_loss: 3.9961 - val_accuracy: 0.0854\n",
+      "Epoch 3825/5000\n",
+      "919/919 - 3s - loss: 1.2952 - accuracy: 0.6146 - val_loss: 3.9911 - val_accuracy: 0.0857\n",
+      "Epoch 3826/5000\n",
+      "919/919 - 3s - loss: 1.2456 - accuracy: 0.6141 - val_loss: 3.9852 - val_accuracy: 0.0860\n",
+      "Epoch 3827/5000\n",
+      "919/919 - 3s - loss: 1.2303 - accuracy: 0.6134 - val_loss: 3.9910 - val_accuracy: 0.0852\n",
+      "Epoch 3828/5000\n",
+      "919/919 - 3s - loss: 1.2387 - accuracy: 0.6137 - val_loss: 3.9867 - val_accuracy: 0.0865\n",
+      "Epoch 3829/5000\n",
+      "919/919 - 3s - loss: 1.2412 - accuracy: 0.6087 - val_loss: 3.9998 - val_accuracy: 0.0863\n",
+      "Epoch 3830/5000\n",
+      "919/919 - 3s - loss: 1.2308 - accuracy: 0.6114 - val_loss: 3.9993 - val_accuracy: 0.0865\n",
+      "Epoch 3831/5000\n",
+      "919/919 - 3s - loss: 1.3257 - accuracy: 0.6104 - val_loss: 3.9869 - val_accuracy: 0.0867\n",
+      "Epoch 3832/5000\n",
+      "919/919 - 3s - loss: 1.2449 - accuracy: 0.6085 - val_loss: 3.9902 - val_accuracy: 0.0859\n",
+      "Epoch 3833/5000\n",
+      "919/919 - 3s - loss: 1.2370 - accuracy: 0.6097 - val_loss: 3.9851 - val_accuracy: 0.0865\n",
+      "Epoch 3834/5000\n",
+      "919/919 - 3s - loss: 1.3094 - accuracy: 0.6129 - val_loss: 4.0084 - val_accuracy: 0.0864\n",
+      "Epoch 3835/5000\n",
+      "919/919 - 3s - loss: 1.2353 - accuracy: 0.6130 - val_loss: 4.0039 - val_accuracy: 0.0858\n",
+      "Epoch 3836/5000\n",
+      "919/919 - 3s - loss: 1.2318 - accuracy: 0.6146 - val_loss: 4.0159 - val_accuracy: 0.0855\n",
+      "Epoch 3837/5000\n",
+      "919/919 - 3s - loss: 1.2290 - accuracy: 0.6133 - val_loss: 4.0138 - val_accuracy: 0.0858\n",
+      "Epoch 3838/5000\n",
+      "919/919 - 3s - loss: 1.2199 - accuracy: 0.6189 - val_loss: 4.0184 - val_accuracy: 0.0864\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3839/5000\n",
+      "919/919 - 3s - loss: 1.2724 - accuracy: 0.6095 - val_loss: 4.0080 - val_accuracy: 0.0859\n",
+      "Epoch 3840/5000\n",
+      "919/919 - 3s - loss: 1.2561 - accuracy: 0.6077 - val_loss: 4.0085 - val_accuracy: 0.0865\n",
+      "Epoch 3841/5000\n",
+      "919/919 - 3s - loss: 1.3261 - accuracy: 0.6158 - val_loss: 4.0216 - val_accuracy: 0.0867\n",
+      "Epoch 3842/5000\n",
+      "919/919 - 3s - loss: 1.3002 - accuracy: 0.6110 - val_loss: 4.0258 - val_accuracy: 0.0860\n",
+      "Epoch 3843/5000\n",
+      "919/919 - 3s - loss: 1.2564 - accuracy: 0.6103 - val_loss: 4.0157 - val_accuracy: 0.0856\n",
+      "Epoch 3844/5000\n",
+      "919/919 - 3s - loss: 1.2433 - accuracy: 0.6120 - val_loss: 4.0115 - val_accuracy: 0.0858\n",
+      "Epoch 3845/5000\n",
+      "919/919 - 3s - loss: 1.2382 - accuracy: 0.6103 - val_loss: 4.0079 - val_accuracy: 0.0861\n",
+      "Epoch 3846/5000\n",
+      "919/919 - 3s - loss: 1.2718 - accuracy: 0.6101 - val_loss: 4.0127 - val_accuracy: 0.0854\n",
+      "Epoch 3847/5000\n",
+      "919/919 - 3s - loss: 1.2236 - accuracy: 0.6127 - val_loss: 4.0178 - val_accuracy: 0.0859\n",
+      "Epoch 3848/5000\n",
+      "919/919 - 3s - loss: 1.2388 - accuracy: 0.6120 - val_loss: 4.0162 - val_accuracy: 0.0865\n",
+      "Epoch 3849/5000\n",
+      "919/919 - 3s - loss: 1.2448 - accuracy: 0.6086 - val_loss: 4.0172 - val_accuracy: 0.0871\n",
+      "Epoch 3850/5000\n",
+      "919/919 - 3s - loss: 1.2343 - accuracy: 0.6153 - val_loss: 4.0010 - val_accuracy: 0.0874\n",
+      "Epoch 3851/5000\n",
+      "919/919 - 3s - loss: 1.2396 - accuracy: 0.6099 - val_loss: 3.9989 - val_accuracy: 0.0874\n",
+      "Epoch 3852/5000\n",
+      "919/919 - 3s - loss: 1.2377 - accuracy: 0.6135 - val_loss: 4.0113 - val_accuracy: 0.0874\n",
+      "Epoch 3853/5000\n",
+      "919/919 - 3s - loss: 1.2349 - accuracy: 0.6103 - val_loss: 3.9964 - val_accuracy: 0.0873\n",
+      "Epoch 3854/5000\n",
+      "919/919 - 3s - loss: 1.2441 - accuracy: 0.6110 - val_loss: 4.0162 - val_accuracy: 0.0865\n",
+      "Epoch 3855/5000\n",
+      "919/919 - 3s - loss: 1.2734 - accuracy: 0.6086 - val_loss: 4.0163 - val_accuracy: 0.0858\n",
+      "Epoch 3856/5000\n",
+      "919/919 - 3s - loss: 1.3110 - accuracy: 0.6106 - val_loss: 4.0141 - val_accuracy: 0.0861\n",
+      "Epoch 3857/5000\n",
+      "919/919 - 3s - loss: 1.2508 - accuracy: 0.6095 - val_loss: 4.0061 - val_accuracy: 0.0857\n",
+      "Epoch 3858/5000\n",
+      "919/919 - 3s - loss: 1.2490 - accuracy: 0.6077 - val_loss: 4.0056 - val_accuracy: 0.0856\n",
+      "Epoch 3859/5000\n",
+      "919/919 - 3s - loss: 1.2667 - accuracy: 0.6110 - val_loss: 4.0105 - val_accuracy: 0.0857\n",
+      "Epoch 3860/5000\n",
+      "919/919 - 3s - loss: 1.2320 - accuracy: 0.6098 - val_loss: 4.0122 - val_accuracy: 0.0854\n",
+      "Epoch 3861/5000\n",
+      "919/919 - 3s - loss: 1.2283 - accuracy: 0.6150 - val_loss: 4.0172 - val_accuracy: 0.0857\n",
+      "Epoch 3862/5000\n",
+      "919/919 - 3s - loss: 1.2593 - accuracy: 0.6156 - val_loss: 4.0165 - val_accuracy: 0.0857\n",
+      "Epoch 3863/5000\n",
+      "919/919 - 3s - loss: 1.2317 - accuracy: 0.6148 - val_loss: 4.0138 - val_accuracy: 0.0861\n",
+      "Epoch 3864/5000\n",
+      "919/919 - 3s - loss: 1.2323 - accuracy: 0.6133 - val_loss: 4.0250 - val_accuracy: 0.0870\n",
+      "Epoch 3865/5000\n",
+      "919/919 - 3s - loss: 1.2389 - accuracy: 0.6149 - val_loss: 4.0295 - val_accuracy: 0.0870\n",
+      "Epoch 3866/5000\n",
+      "919/919 - 3s - loss: 1.2542 - accuracy: 0.6071 - val_loss: 4.0078 - val_accuracy: 0.0869\n",
+      "Epoch 3867/5000\n",
+      "919/919 - 3s - loss: 1.2481 - accuracy: 0.6090 - val_loss: 3.9979 - val_accuracy: 0.0872\n",
+      "Epoch 3868/5000\n",
+      "919/919 - 3s - loss: 1.2412 - accuracy: 0.6126 - val_loss: 4.0093 - val_accuracy: 0.0868\n",
+      "Epoch 3869/5000\n",
+      "919/919 - 3s - loss: 1.2484 - accuracy: 0.6100 - val_loss: 4.0251 - val_accuracy: 0.0864\n",
+      "Epoch 3870/5000\n",
+      "919/919 - 3s - loss: 1.2381 - accuracy: 0.6120 - val_loss: 4.0233 - val_accuracy: 0.0866\n",
+      "Epoch 3871/5000\n",
+      "919/919 - 3s - loss: 1.2416 - accuracy: 0.6122 - val_loss: 4.0310 - val_accuracy: 0.0863\n",
+      "Epoch 3872/5000\n",
+      "919/919 - 3s - loss: 1.2291 - accuracy: 0.6095 - val_loss: 4.0176 - val_accuracy: 0.0864\n",
+      "Epoch 3873/5000\n",
+      "919/919 - 3s - loss: 1.2587 - accuracy: 0.6106 - val_loss: 4.0233 - val_accuracy: 0.0863\n",
+      "Epoch 3874/5000\n",
+      "919/919 - 3s - loss: 1.2474 - accuracy: 0.6102 - val_loss: 4.0200 - val_accuracy: 0.0865\n",
+      "Epoch 3875/5000\n",
+      "919/919 - 3s - loss: 1.2330 - accuracy: 0.6162 - val_loss: 4.0119 - val_accuracy: 0.0869\n",
+      "Epoch 3876/5000\n",
+      "919/919 - 3s - loss: 1.3097 - accuracy: 0.6139 - val_loss: 4.0197 - val_accuracy: 0.0869\n",
+      "Epoch 3877/5000\n",
+      "919/919 - 3s - loss: 1.2346 - accuracy: 0.6121 - val_loss: 4.0232 - val_accuracy: 0.0859\n",
+      "Epoch 3878/5000\n",
+      "919/919 - 3s - loss: 1.2356 - accuracy: 0.6120 - val_loss: 4.0310 - val_accuracy: 0.0862\n",
+      "Epoch 3879/5000\n",
+      "919/919 - 3s - loss: 1.2217 - accuracy: 0.6162 - val_loss: 4.0304 - val_accuracy: 0.0865\n",
+      "Epoch 3880/5000\n",
+      "919/919 - 3s - loss: 1.2308 - accuracy: 0.6144 - val_loss: 4.0256 - val_accuracy: 0.0865\n",
+      "Epoch 3881/5000\n",
+      "919/919 - 3s - loss: 1.2938 - accuracy: 0.6078 - val_loss: 4.0155 - val_accuracy: 0.0868\n",
+      "Epoch 3882/5000\n",
+      "919/919 - 3s - loss: 1.2285 - accuracy: 0.6161 - val_loss: 4.0147 - val_accuracy: 0.0865\n",
+      "Epoch 3883/5000\n",
+      "919/919 - 3s - loss: 1.2392 - accuracy: 0.6122 - val_loss: 4.0138 - val_accuracy: 0.0865\n",
+      "Epoch 3884/5000\n",
+      "919/919 - 3s - loss: 1.2320 - accuracy: 0.6163 - val_loss: 4.0146 - val_accuracy: 0.0865\n",
+      "Epoch 3885/5000\n",
+      "919/919 - 3s - loss: 1.2339 - accuracy: 0.6150 - val_loss: 4.0161 - val_accuracy: 0.0867\n",
+      "Epoch 3886/5000\n",
+      "919/919 - 3s - loss: 1.2226 - accuracy: 0.6154 - val_loss: 4.0167 - val_accuracy: 0.0868\n",
+      "Epoch 3887/5000\n",
+      "919/919 - 3s - loss: 1.2442 - accuracy: 0.6139 - val_loss: 4.0135 - val_accuracy: 0.0865\n",
+      "Epoch 3888/5000\n",
+      "919/919 - 3s - loss: 1.2231 - accuracy: 0.6156 - val_loss: 4.0189 - val_accuracy: 0.0868\n",
+      "Epoch 3889/5000\n",
+      "919/919 - 3s - loss: 1.2183 - accuracy: 0.6184 - val_loss: 4.0424 - val_accuracy: 0.0865\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3890/5000\n",
+      "919/919 - 3s - loss: 1.2693 - accuracy: 0.6088 - val_loss: 4.0143 - val_accuracy: 0.0864\n",
+      "Epoch 3891/5000\n",
+      "919/919 - 3s - loss: 1.2686 - accuracy: 0.6124 - val_loss: 4.0283 - val_accuracy: 0.0863\n",
+      "Epoch 3892/5000\n",
+      "919/919 - 3s - loss: 1.2348 - accuracy: 0.6106 - val_loss: 4.0319 - val_accuracy: 0.0862\n",
+      "Epoch 3893/5000\n",
+      "919/919 - 3s - loss: 1.2487 - accuracy: 0.6106 - val_loss: 4.0306 - val_accuracy: 0.0865\n",
+      "Epoch 3894/5000\n",
+      "919/919 - 3s - loss: 1.3952 - accuracy: 0.6160 - val_loss: 4.0271 - val_accuracy: 0.0862\n",
+      "Epoch 3895/5000\n",
+      "919/919 - 3s - loss: 1.2360 - accuracy: 0.6123 - val_loss: 4.0332 - val_accuracy: 0.0868\n",
+      "Epoch 3896/5000\n",
+      "919/919 - 3s - loss: 1.2165 - accuracy: 0.6139 - val_loss: 4.0033 - val_accuracy: 0.0866\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3897/5000\n",
+      "919/919 - 3s - loss: 1.3063 - accuracy: 0.6148 - val_loss: 4.0126 - val_accuracy: 0.0871\n",
+      "Epoch 3898/5000\n",
+      "919/919 - 3s - loss: 1.2337 - accuracy: 0.6121 - val_loss: 4.0395 - val_accuracy: 0.0871\n",
+      "Epoch 3899/5000\n",
+      "919/919 - 3s - loss: 1.2293 - accuracy: 0.6159 - val_loss: 4.0272 - val_accuracy: 0.0872\n",
+      "Epoch 3900/5000\n",
+      "919/919 - 3s - loss: 1.2169 - accuracy: 0.6157 - val_loss: 4.0418 - val_accuracy: 0.0865\n",
+      "Epoch 3901/5000\n",
+      "919/919 - 3s - loss: 1.2454 - accuracy: 0.6133 - val_loss: 4.0376 - val_accuracy: 0.0865\n",
+      "Epoch 3902/5000\n",
+      "919/919 - 3s - loss: 1.2215 - accuracy: 0.6114 - val_loss: 4.0353 - val_accuracy: 0.0865\n",
+      "Epoch 3903/5000\n",
+      "919/919 - 3s - loss: 1.2636 - accuracy: 0.6147 - val_loss: 4.0272 - val_accuracy: 0.0868\n",
+      "Epoch 3904/5000\n",
+      "919/919 - 3s - loss: 1.2382 - accuracy: 0.6144 - val_loss: 4.0271 - val_accuracy: 0.0867\n",
+      "Epoch 3905/5000\n",
+      "919/919 - 3s - loss: 1.2282 - accuracy: 0.6167 - val_loss: 4.0229 - val_accuracy: 0.0869\n",
+      "Epoch 3906/5000\n",
+      "919/919 - 3s - loss: 1.3109 - accuracy: 0.6171 - val_loss: 4.0318 - val_accuracy: 0.0869\n",
+      "Epoch 3907/5000\n",
+      "919/919 - 3s - loss: 1.2422 - accuracy: 0.6152 - val_loss: 4.0262 - val_accuracy: 0.0874\n",
+      "Epoch 3908/5000\n",
+      "919/919 - 3s - loss: 1.2273 - accuracy: 0.6131 - val_loss: 4.0234 - val_accuracy: 0.0871\n",
+      "Epoch 3909/5000\n",
+      "919/919 - 3s - loss: 1.2253 - accuracy: 0.6170 - val_loss: 4.0361 - val_accuracy: 0.0874\n",
+      "Epoch 3910/5000\n",
+      "919/919 - 3s - loss: 1.2337 - accuracy: 0.6107 - val_loss: 4.0467 - val_accuracy: 0.0871\n",
+      "Epoch 3911/5000\n",
+      "919/919 - 3s - loss: 1.2163 - accuracy: 0.6147 - val_loss: 4.0498 - val_accuracy: 0.0868\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3912/5000\n",
+      "919/919 - 3s - loss: 1.2287 - accuracy: 0.6131 - val_loss: 4.0414 - val_accuracy: 0.0858\n",
+      "Epoch 3913/5000\n",
+      "919/919 - 3s - loss: 1.2467 - accuracy: 0.6128 - val_loss: 4.0417 - val_accuracy: 0.0862\n",
+      "Epoch 3914/5000\n",
+      "919/919 - 3s - loss: 1.2552 - accuracy: 0.6137 - val_loss: 4.0243 - val_accuracy: 0.0861\n",
+      "Epoch 3915/5000\n",
+      "919/919 - 3s - loss: 1.2537 - accuracy: 0.6124 - val_loss: 4.0138 - val_accuracy: 0.0865\n",
+      "Epoch 3916/5000\n",
+      "919/919 - 3s - loss: 1.2447 - accuracy: 0.6152 - val_loss: 4.0278 - val_accuracy: 0.0866\n",
+      "Epoch 3917/5000\n",
+      "919/919 - 3s - loss: 1.2889 - accuracy: 0.6138 - val_loss: 4.0195 - val_accuracy: 0.0869\n",
+      "Epoch 3918/5000\n",
+      "919/919 - 3s - loss: 1.2295 - accuracy: 0.6141 - val_loss: 4.0196 - val_accuracy: 0.0866\n",
+      "Epoch 3919/5000\n",
+      "919/919 - 3s - loss: 1.2240 - accuracy: 0.6150 - val_loss: 4.0162 - val_accuracy: 0.0867\n",
+      "Epoch 3920/5000\n",
+      "919/919 - 3s - loss: 1.2256 - accuracy: 0.6129 - val_loss: 4.0245 - val_accuracy: 0.0866\n",
+      "Epoch 3921/5000\n",
+      "919/919 - 3s - loss: 1.2392 - accuracy: 0.6122 - val_loss: 4.0316 - val_accuracy: 0.0866\n",
+      "Epoch 3922/5000\n",
+      "919/919 - 3s - loss: 1.2332 - accuracy: 0.6157 - val_loss: 4.0351 - val_accuracy: 0.0865\n",
+      "Epoch 3923/5000\n",
+      "919/919 - 3s - loss: 1.2505 - accuracy: 0.6137 - val_loss: 4.0357 - val_accuracy: 0.0866\n",
+      "Epoch 3924/5000\n",
+      "919/919 - 3s - loss: 1.2362 - accuracy: 0.6134 - val_loss: 4.0348 - val_accuracy: 0.0859\n",
+      "Epoch 3925/5000\n",
+      "919/919 - 3s - loss: 1.2208 - accuracy: 0.6142 - val_loss: 4.0350 - val_accuracy: 0.0864\n",
+      "Epoch 3926/5000\n",
+      "919/919 - 3s - loss: 1.2370 - accuracy: 0.6178 - val_loss: 4.0401 - val_accuracy: 0.0865\n",
+      "Epoch 3927/5000\n",
+      "919/919 - 3s - loss: 1.3261 - accuracy: 0.6147 - val_loss: 4.0394 - val_accuracy: 0.0868\n",
+      "Epoch 3928/5000\n",
+      "919/919 - 3s - loss: 1.2414 - accuracy: 0.6128 - val_loss: 4.0365 - val_accuracy: 0.0868\n",
+      "Epoch 3929/5000\n",
+      "919/919 - 3s - loss: 1.2757 - accuracy: 0.6159 - val_loss: 4.0308 - val_accuracy: 0.0865\n",
+      "Epoch 3930/5000\n",
+      "919/919 - 3s - loss: 1.2309 - accuracy: 0.6122 - val_loss: 4.0294 - val_accuracy: 0.0865\n",
+      "Epoch 3931/5000\n",
+      "919/919 - 3s - loss: 1.2375 - accuracy: 0.6176 - val_loss: 4.0315 - val_accuracy: 0.0864\n",
+      "Epoch 3932/5000\n",
+      "919/919 - 3s - loss: 1.2266 - accuracy: 0.6199 - val_loss: 4.0358 - val_accuracy: 0.0863\n",
+      "Epoch 3933/5000\n",
+      "919/919 - 3s - loss: 1.2222 - accuracy: 0.6142 - val_loss: 4.0385 - val_accuracy: 0.0865\n",
+      "Epoch 3934/5000\n",
+      "919/919 - 3s - loss: 1.2293 - accuracy: 0.6149 - val_loss: 4.0352 - val_accuracy: 0.0868\n",
+      "Epoch 3935/5000\n",
+      "919/919 - 3s - loss: 1.2229 - accuracy: 0.6164 - val_loss: 4.0366 - val_accuracy: 0.0874\n",
+      "Epoch 3936/5000\n",
+      "919/919 - 3s - loss: 1.2258 - accuracy: 0.6122 - val_loss: 4.0475 - val_accuracy: 0.0867\n",
+      "Epoch 3937/5000\n",
+      "919/919 - 3s - loss: 1.2274 - accuracy: 0.6147 - val_loss: 4.0390 - val_accuracy: 0.0868\n",
+      "Epoch 3938/5000\n",
+      "919/919 - 3s - loss: 1.2366 - accuracy: 0.6152 - val_loss: 4.0447 - val_accuracy: 0.0876\n",
+      "Epoch 3939/5000\n",
+      "919/919 - 3s - loss: 1.2255 - accuracy: 0.6197 - val_loss: 4.0340 - val_accuracy: 0.0874\n",
+      "Epoch 3940/5000\n",
+      "919/919 - 3s - loss: 1.2242 - accuracy: 0.6158 - val_loss: 4.0304 - val_accuracy: 0.0868\n",
+      "Epoch 3941/5000\n",
+      "919/919 - 3s - loss: 1.2271 - accuracy: 0.6131 - val_loss: 4.0287 - val_accuracy: 0.0866\n",
+      "Epoch 3942/5000\n",
+      "919/919 - 3s - loss: 1.2314 - accuracy: 0.6153 - val_loss: 4.0348 - val_accuracy: 0.0866\n",
+      "Epoch 3943/5000\n",
+      "919/919 - 3s - loss: 1.2178 - accuracy: 0.6150 - val_loss: 4.0260 - val_accuracy: 0.0868\n",
+      "Epoch 3944/5000\n",
+      "919/919 - 3s - loss: 1.2529 - accuracy: 0.6150 - val_loss: 4.0238 - val_accuracy: 0.0860\n",
+      "Epoch 3945/5000\n",
+      "919/919 - 3s - loss: 1.3002 - accuracy: 0.6154 - val_loss: 4.0332 - val_accuracy: 0.0862\n",
+      "Epoch 3946/5000\n",
+      "919/919 - 3s - loss: 1.2409 - accuracy: 0.6112 - val_loss: 4.0327 - val_accuracy: 0.0869\n",
+      "Epoch 3947/5000\n",
+      "919/919 - 3s - loss: 1.2362 - accuracy: 0.6166 - val_loss: 4.0295 - val_accuracy: 0.0866\n",
+      "Epoch 3948/5000\n",
+      "919/919 - 3s - loss: 1.2191 - accuracy: 0.6157 - val_loss: 4.0369 - val_accuracy: 0.0867\n",
+      "Epoch 3949/5000\n",
+      "919/919 - 3s - loss: 1.2568 - accuracy: 0.6159 - val_loss: 4.0407 - val_accuracy: 0.0863\n",
+      "Epoch 3950/5000\n",
+      "919/919 - 3s - loss: 1.2372 - accuracy: 0.6176 - val_loss: 4.0304 - val_accuracy: 0.0865\n",
+      "Epoch 3951/5000\n",
+      "919/919 - 3s - loss: 1.2928 - accuracy: 0.6078 - val_loss: 4.0322 - val_accuracy: 0.0865\n",
+      "Epoch 3952/5000\n",
+      "919/919 - 3s - loss: 1.2817 - accuracy: 0.6134 - val_loss: 4.0281 - val_accuracy: 0.0865\n",
+      "Epoch 3953/5000\n",
+      "919/919 - 3s - loss: 1.2288 - accuracy: 0.6167 - val_loss: 4.0360 - val_accuracy: 0.0867\n",
+      "Epoch 3954/5000\n",
+      "919/919 - 3s - loss: 1.2315 - accuracy: 0.6156 - val_loss: 4.0443 - val_accuracy: 0.0865\n",
+      "Epoch 3955/5000\n",
+      "919/919 - 3s - loss: 1.2491 - accuracy: 0.6129 - val_loss: 4.0375 - val_accuracy: 0.0867\n",
+      "Epoch 3956/5000\n",
+      "919/919 - 3s - loss: 1.4319 - accuracy: 0.6140 - val_loss: 4.0506 - val_accuracy: 0.0863\n",
+      "Epoch 3957/5000\n",
+      "919/919 - 3s - loss: 1.2735 - accuracy: 0.6152 - val_loss: 4.0441 - val_accuracy: 0.0863\n",
+      "Epoch 3958/5000\n",
+      "919/919 - 3s - loss: 1.2862 - accuracy: 0.6139 - val_loss: 4.0399 - val_accuracy: 0.0864\n",
+      "Epoch 3959/5000\n",
+      "919/919 - 3s - loss: 1.2319 - accuracy: 0.6105 - val_loss: 4.0418 - val_accuracy: 0.0865\n",
+      "Epoch 3960/5000\n",
+      "919/919 - 3s - loss: 1.2348 - accuracy: 0.6122 - val_loss: 4.0262 - val_accuracy: 0.0871\n",
+      "Epoch 3961/5000\n",
+      "919/919 - 3s - loss: 1.2278 - accuracy: 0.6154 - val_loss: 4.0308 - val_accuracy: 0.0880\n",
+      "Epoch 3962/5000\n",
+      "919/919 - 3s - loss: 1.2580 - accuracy: 0.6141 - val_loss: 4.0238 - val_accuracy: 0.0871\n",
+      "Epoch 3963/5000\n",
+      "919/919 - 3s - loss: 1.2401 - accuracy: 0.6152 - val_loss: 4.0294 - val_accuracy: 0.0875\n",
+      "Epoch 3964/5000\n",
+      "919/919 - 3s - loss: 1.2852 - accuracy: 0.6146 - val_loss: 4.0335 - val_accuracy: 0.0868\n",
+      "Epoch 3965/5000\n",
+      "919/919 - 3s - loss: 1.2337 - accuracy: 0.6109 - val_loss: 4.0345 - val_accuracy: 0.0869\n",
+      "Epoch 3966/5000\n",
+      "919/919 - 3s - loss: 1.2317 - accuracy: 0.6141 - val_loss: 4.0344 - val_accuracy: 0.0872\n",
+      "Epoch 3967/5000\n",
+      "919/919 - 3s - loss: 1.2226 - accuracy: 0.6161 - val_loss: 4.0460 - val_accuracy: 0.0869\n",
+      "Epoch 3968/5000\n",
+      "919/919 - 3s - loss: 1.2471 - accuracy: 0.6188 - val_loss: 4.0441 - val_accuracy: 0.0874\n",
+      "Epoch 3969/5000\n",
+      "919/919 - 3s - loss: 1.2315 - accuracy: 0.6191 - val_loss: 4.0692 - val_accuracy: 0.0867\n",
+      "Epoch 3970/5000\n",
+      "919/919 - 3s - loss: 1.2428 - accuracy: 0.6146 - val_loss: 4.0605 - val_accuracy: 0.0866\n",
+      "Epoch 3971/5000\n",
+      "919/919 - 3s - loss: 1.2541 - accuracy: 0.6133 - val_loss: 4.0371 - val_accuracy: 0.0875\n",
+      "Epoch 3972/5000\n",
+      "919/919 - 3s - loss: 1.2285 - accuracy: 0.6159 - val_loss: 4.0332 - val_accuracy: 0.0862\n",
+      "Epoch 3973/5000\n",
+      "919/919 - 3s - loss: 1.2260 - accuracy: 0.6165 - val_loss: 4.0171 - val_accuracy: 0.0869\n",
+      "Epoch 3974/5000\n",
+      "919/919 - 3s - loss: 1.2322 - accuracy: 0.6124 - val_loss: 4.0180 - val_accuracy: 0.0861\n",
+      "Epoch 3975/5000\n",
+      "919/919 - 3s - loss: 1.2321 - accuracy: 0.6165 - val_loss: 4.0197 - val_accuracy: 0.0866\n",
+      "Epoch 3976/5000\n",
+      "919/919 - 3s - loss: 1.2182 - accuracy: 0.6194 - val_loss: 4.0407 - val_accuracy: 0.0860\n",
+      "Epoch 3977/5000\n",
+      "919/919 - 3s - loss: 1.2255 - accuracy: 0.6181 - val_loss: 4.0333 - val_accuracy: 0.0869\n",
+      "Epoch 3978/5000\n",
+      "919/919 - 3s - loss: 1.2296 - accuracy: 0.6144 - val_loss: 4.0425 - val_accuracy: 0.0865\n",
+      "Epoch 3979/5000\n",
+      "919/919 - 3s - loss: 1.2347 - accuracy: 0.6148 - val_loss: 4.0356 - val_accuracy: 0.0867\n",
+      "Epoch 3980/5000\n",
+      "919/919 - 3s - loss: 1.2404 - accuracy: 0.6159 - val_loss: 4.0309 - val_accuracy: 0.0868\n",
+      "Epoch 3981/5000\n",
+      "919/919 - 3s - loss: 1.2292 - accuracy: 0.6152 - val_loss: 4.0257 - val_accuracy: 0.0865\n",
+      "Epoch 3982/5000\n",
+      "919/919 - 3s - loss: 1.2259 - accuracy: 0.6131 - val_loss: 4.0277 - val_accuracy: 0.0867\n",
+      "Epoch 3983/5000\n",
+      "919/919 - 3s - loss: 1.2286 - accuracy: 0.6165 - val_loss: 4.0291 - val_accuracy: 0.0865\n",
+      "Epoch 3984/5000\n",
+      "919/919 - 3s - loss: 1.2313 - accuracy: 0.6167 - val_loss: 4.0286 - val_accuracy: 0.0861\n",
+      "Epoch 3985/5000\n",
+      "919/919 - 3s - loss: 1.2365 - accuracy: 0.6171 - val_loss: 4.0263 - val_accuracy: 0.0869\n",
+      "Epoch 3986/5000\n",
+      "919/919 - 3s - loss: 1.2113 - accuracy: 0.6202 - val_loss: 4.0308 - val_accuracy: 0.0864\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 3987/5000\n",
+      "919/919 - 3s - loss: 1.2203 - accuracy: 0.6170 - val_loss: 4.0316 - val_accuracy: 0.0866\n",
+      "Epoch 3988/5000\n",
+      "919/919 - 3s - loss: 1.2134 - accuracy: 0.6175 - val_loss: 4.0306 - val_accuracy: 0.0863\n",
+      "Epoch 3989/5000\n",
+      "919/919 - 3s - loss: 1.2551 - accuracy: 0.6136 - val_loss: 4.0169 - val_accuracy: 0.0863\n",
+      "Epoch 3990/5000\n",
+      "919/919 - 3s - loss: 1.2944 - accuracy: 0.6137 - val_loss: 4.0229 - val_accuracy: 0.0870\n",
+      "Epoch 3991/5000\n",
+      "919/919 - 3s - loss: 1.2336 - accuracy: 0.6158 - val_loss: 4.0431 - val_accuracy: 0.0872\n",
+      "Epoch 3992/5000\n",
+      "919/919 - 3s - loss: 1.2333 - accuracy: 0.6153 - val_loss: 4.0579 - val_accuracy: 0.0869\n",
+      "Epoch 3993/5000\n",
+      "919/919 - 3s - loss: 1.2236 - accuracy: 0.6184 - val_loss: 4.0456 - val_accuracy: 0.0878\n",
+      "Epoch 3994/5000\n",
+      "919/919 - 3s - loss: 1.2185 - accuracy: 0.6152 - val_loss: 4.0410 - val_accuracy: 0.0873\n",
+      "Epoch 3995/5000\n",
+      "919/919 - 3s - loss: 1.2598 - accuracy: 0.6127 - val_loss: 4.0376 - val_accuracy: 0.0874\n",
+      "Epoch 3996/5000\n",
+      "919/919 - 3s - loss: 1.2165 - accuracy: 0.6207 - val_loss: 4.0458 - val_accuracy: 0.0872\n",
+      "Epoch 3997/5000\n",
+      "919/919 - 3s - loss: 1.2317 - accuracy: 0.6161 - val_loss: 4.0531 - val_accuracy: 0.0874\n",
+      "Epoch 3998/5000\n",
+      "919/919 - 3s - loss: 1.2370 - accuracy: 0.6171 - val_loss: 4.0430 - val_accuracy: 0.0872\n",
+      "Epoch 3999/5000\n",
+      "919/919 - 3s - loss: 1.2315 - accuracy: 0.6169 - val_loss: 4.0421 - val_accuracy: 0.0880\n",
+      "Epoch 4000/5000\n",
+      "919/919 - 3s - loss: 1.2354 - accuracy: 0.6178 - val_loss: 4.0405 - val_accuracy: 0.0878\n",
+      "Epoch 4001/5000\n",
+      "919/919 - 3s - loss: 1.2282 - accuracy: 0.6173 - val_loss: 4.0559 - val_accuracy: 0.0874\n",
+      "Epoch 4002/5000\n",
+      "919/919 - 3s - loss: 1.2342 - accuracy: 0.6140 - val_loss: 4.0547 - val_accuracy: 0.0873\n",
+      "Epoch 4003/5000\n",
+      "919/919 - 3s - loss: 1.2339 - accuracy: 0.6154 - val_loss: 4.0471 - val_accuracy: 0.0874\n",
+      "Epoch 4004/5000\n",
+      "919/919 - 3s - loss: 1.2278 - accuracy: 0.6185 - val_loss: 4.0451 - val_accuracy: 0.0865\n",
+      "Epoch 4005/5000\n",
+      "919/919 - 3s - loss: 1.2625 - accuracy: 0.6139 - val_loss: 4.0278 - val_accuracy: 0.0872\n",
+      "Epoch 4006/5000\n",
+      "919/919 - 3s - loss: 1.2381 - accuracy: 0.6155 - val_loss: 4.0400 - val_accuracy: 0.0872\n",
+      "Epoch 4007/5000\n",
+      "919/919 - 3s - loss: 1.2777 - accuracy: 0.6169 - val_loss: 4.0381 - val_accuracy: 0.0875\n",
+      "Epoch 4008/5000\n",
+      "919/919 - 3s - loss: 1.2151 - accuracy: 0.6190 - val_loss: 4.0464 - val_accuracy: 0.0873\n",
+      "Epoch 4009/5000\n",
+      "919/919 - 3s - loss: 1.2383 - accuracy: 0.6164 - val_loss: 4.0386 - val_accuracy: 0.0875\n",
+      "Epoch 4010/5000\n",
+      "919/919 - 3s - loss: 1.2153 - accuracy: 0.6172 - val_loss: 4.0421 - val_accuracy: 0.0877\n",
+      "Epoch 4011/5000\n",
+      "919/919 - 3s - loss: 1.2245 - accuracy: 0.6169 - val_loss: 4.0491 - val_accuracy: 0.0877\n",
+      "Epoch 4012/5000\n",
+      "919/919 - 3s - loss: 1.2265 - accuracy: 0.6173 - val_loss: 4.0498 - val_accuracy: 0.0875\n",
+      "Epoch 4013/5000\n",
+      "919/919 - 3s - loss: 1.2512 - accuracy: 0.6158 - val_loss: 4.0464 - val_accuracy: 0.0876\n",
+      "Epoch 4014/5000\n",
+      "919/919 - 3s - loss: 1.2200 - accuracy: 0.6156 - val_loss: 4.0617 - val_accuracy: 0.0875\n",
+      "Epoch 4015/5000\n",
+      "919/919 - 3s - loss: 1.2237 - accuracy: 0.6142 - val_loss: 4.0634 - val_accuracy: 0.0874\n",
+      "Epoch 4016/5000\n",
+      "919/919 - 3s - loss: 1.2387 - accuracy: 0.6197 - val_loss: 4.0757 - val_accuracy: 0.0872\n",
+      "Epoch 4017/5000\n",
+      "919/919 - 3s - loss: 1.2263 - accuracy: 0.6186 - val_loss: 4.0658 - val_accuracy: 0.0868\n",
+      "Epoch 4018/5000\n",
+      "919/919 - 3s - loss: 1.2208 - accuracy: 0.6203 - val_loss: 4.0688 - val_accuracy: 0.0871\n",
+      "Epoch 4019/5000\n",
+      "919/919 - 3s - loss: 1.2419 - accuracy: 0.6185 - val_loss: 4.0647 - val_accuracy: 0.0874\n",
+      "Epoch 4020/5000\n",
+      "919/919 - 3s - loss: 1.2844 - accuracy: 0.6177 - val_loss: 4.0644 - val_accuracy: 0.0871\n",
+      "Epoch 4021/5000\n",
+      "919/919 - 3s - loss: 1.2309 - accuracy: 0.6133 - val_loss: 4.0515 - val_accuracy: 0.0867\n",
+      "Epoch 4022/5000\n",
+      "919/919 - 3s - loss: 1.2105 - accuracy: 0.6205 - val_loss: 4.0632 - val_accuracy: 0.0869\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4023/5000\n",
+      "919/919 - 3s - loss: 1.2296 - accuracy: 0.6162 - val_loss: 4.0695 - val_accuracy: 0.0869\n",
+      "Epoch 4024/5000\n",
+      "919/919 - 3s - loss: 1.2232 - accuracy: 0.6166 - val_loss: 4.0588 - val_accuracy: 0.0872\n",
+      "Epoch 4025/5000\n",
+      "919/919 - 3s - loss: 1.2360 - accuracy: 0.6154 - val_loss: 4.0762 - val_accuracy: 0.0871\n",
+      "Epoch 4026/5000\n",
+      "919/919 - 3s - loss: 1.2292 - accuracy: 0.6143 - val_loss: 4.0760 - val_accuracy: 0.0870\n",
+      "Epoch 4027/5000\n",
+      "919/919 - 3s - loss: 1.2388 - accuracy: 0.6144 - val_loss: 4.0566 - val_accuracy: 0.0873\n",
+      "Epoch 4028/5000\n",
+      "919/919 - 3s - loss: 1.2476 - accuracy: 0.6137 - val_loss: 4.0710 - val_accuracy: 0.0864\n",
+      "Epoch 4029/5000\n",
+      "919/919 - 3s - loss: 1.2321 - accuracy: 0.6218 - val_loss: 4.0688 - val_accuracy: 0.0865\n",
+      "Epoch 4030/5000\n",
+      "919/919 - 3s - loss: 1.2252 - accuracy: 0.6160 - val_loss: 4.0552 - val_accuracy: 0.0862\n",
+      "Epoch 4031/5000\n",
+      "919/919 - 3s - loss: 1.2426 - accuracy: 0.6174 - val_loss: 4.0659 - val_accuracy: 0.0870\n",
+      "Epoch 4032/5000\n",
+      "919/919 - 3s - loss: 1.2168 - accuracy: 0.6156 - val_loss: 4.0744 - val_accuracy: 0.0873\n",
+      "Epoch 4033/5000\n",
+      "919/919 - 3s - loss: 1.2386 - accuracy: 0.6149 - val_loss: 4.0564 - val_accuracy: 0.0870\n",
+      "Epoch 4034/5000\n",
+      "919/919 - 3s - loss: 1.2137 - accuracy: 0.6179 - val_loss: 4.0474 - val_accuracy: 0.0869\n",
+      "Epoch 4035/5000\n",
+      "919/919 - 3s - loss: 1.3286 - accuracy: 0.6150 - val_loss: 4.0470 - val_accuracy: 0.0872\n",
+      "Epoch 4036/5000\n",
+      "919/919 - 3s - loss: 1.2212 - accuracy: 0.6173 - val_loss: 4.0557 - val_accuracy: 0.0873\n",
+      "Epoch 4037/5000\n",
+      "919/919 - 3s - loss: 1.2702 - accuracy: 0.6167 - val_loss: 4.0587 - val_accuracy: 0.0870\n",
+      "Epoch 4038/5000\n",
+      "919/919 - 3s - loss: 1.2192 - accuracy: 0.6183 - val_loss: 4.0504 - val_accuracy: 0.0873\n",
+      "Epoch 4039/5000\n",
+      "919/919 - 3s - loss: 1.2332 - accuracy: 0.6163 - val_loss: 4.0486 - val_accuracy: 0.0874\n",
+      "Epoch 4040/5000\n",
+      "919/919 - 3s - loss: 1.2149 - accuracy: 0.6152 - val_loss: 4.0472 - val_accuracy: 0.0871\n",
+      "Epoch 4041/5000\n",
+      "919/919 - 3s - loss: 1.2277 - accuracy: 0.6152 - val_loss: 4.0472 - val_accuracy: 0.0875\n",
+      "Epoch 4042/5000\n",
+      "919/919 - 3s - loss: 1.2080 - accuracy: 0.6231 - val_loss: 4.0497 - val_accuracy: 0.0873\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4043/5000\n",
+      "919/919 - 3s - loss: 1.2512 - accuracy: 0.6132 - val_loss: 4.0490 - val_accuracy: 0.0876\n",
+      "Epoch 4044/5000\n",
+      "919/919 - 3s - loss: 1.2419 - accuracy: 0.6137 - val_loss: 4.0642 - val_accuracy: 0.0878\n",
+      "Epoch 4045/5000\n",
+      "919/919 - 3s - loss: 1.2029 - accuracy: 0.6230 - val_loss: 4.0635 - val_accuracy: 0.0876\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4046/5000\n",
+      "919/919 - 3s - loss: 1.2176 - accuracy: 0.6224 - val_loss: 4.0629 - val_accuracy: 0.0874\n",
+      "Epoch 4047/5000\n",
+      "919/919 - 3s - loss: 1.2087 - accuracy: 0.6210 - val_loss: 4.0945 - val_accuracy: 0.0874\n",
+      "Epoch 4048/5000\n",
+      "919/919 - 3s - loss: 1.2317 - accuracy: 0.6136 - val_loss: 4.0799 - val_accuracy: 0.0874\n",
+      "Epoch 4049/5000\n",
+      "919/919 - 3s - loss: 1.3826 - accuracy: 0.6142 - val_loss: 4.0803 - val_accuracy: 0.0874\n",
+      "Epoch 4050/5000\n",
+      "919/919 - 3s - loss: 1.2063 - accuracy: 0.6201 - val_loss: 4.0668 - val_accuracy: 0.0874\n",
+      "Epoch 4051/5000\n",
+      "919/919 - 3s - loss: 1.2030 - accuracy: 0.6207 - val_loss: 4.0728 - val_accuracy: 0.0876\n",
+      "Epoch 4052/5000\n",
+      "919/919 - 3s - loss: 1.2089 - accuracy: 0.6170 - val_loss: 4.0882 - val_accuracy: 0.0880\n",
+      "Epoch 4053/5000\n",
+      "919/919 - 3s - loss: 1.2398 - accuracy: 0.6152 - val_loss: 4.0792 - val_accuracy: 0.0874\n",
+      "Epoch 4054/5000\n",
+      "919/919 - 3s - loss: 1.2128 - accuracy: 0.6179 - val_loss: 4.0844 - val_accuracy: 0.0878\n",
+      "Epoch 4055/5000\n",
+      "919/919 - 3s - loss: 1.2258 - accuracy: 0.6188 - val_loss: 4.0960 - val_accuracy: 0.0875\n",
+      "Epoch 4056/5000\n",
+      "919/919 - 3s - loss: 1.3312 - accuracy: 0.6199 - val_loss: 4.0836 - val_accuracy: 0.0868\n",
+      "Epoch 4057/5000\n",
+      "919/919 - 3s - loss: 1.2312 - accuracy: 0.6154 - val_loss: 4.0671 - val_accuracy: 0.0874\n",
+      "Epoch 4058/5000\n",
+      "919/919 - 3s - loss: 1.2251 - accuracy: 0.6167 - val_loss: 4.0677 - val_accuracy: 0.0868\n",
+      "Epoch 4059/5000\n",
+      "919/919 - 3s - loss: 1.2719 - accuracy: 0.6208 - val_loss: 4.0755 - val_accuracy: 0.0871\n",
+      "Epoch 4060/5000\n",
+      "919/919 - 3s - loss: 1.2296 - accuracy: 0.6156 - val_loss: 4.0666 - val_accuracy: 0.0871\n",
+      "Epoch 4061/5000\n",
+      "919/919 - 3s - loss: 1.2186 - accuracy: 0.6154 - val_loss: 4.0628 - val_accuracy: 0.0869\n",
+      "Epoch 4062/5000\n",
+      "919/919 - 3s - loss: 1.2209 - accuracy: 0.6176 - val_loss: 4.0668 - val_accuracy: 0.0873\n",
+      "Epoch 4063/5000\n",
+      "919/919 - 3s - loss: 1.2318 - accuracy: 0.6206 - val_loss: 4.0562 - val_accuracy: 0.0869\n",
+      "Epoch 4064/5000\n",
+      "919/919 - 3s - loss: 1.2428 - accuracy: 0.6135 - val_loss: 4.0555 - val_accuracy: 0.0873\n",
+      "Epoch 4065/5000\n",
+      "919/919 - 3s - loss: 1.2222 - accuracy: 0.6176 - val_loss: 4.0560 - val_accuracy: 0.0876\n",
+      "Epoch 4066/5000\n",
+      "919/919 - 3s - loss: 1.2202 - accuracy: 0.6179 - val_loss: 4.0575 - val_accuracy: 0.0874\n",
+      "Epoch 4067/5000\n",
+      "919/919 - 3s - loss: 1.2274 - accuracy: 0.6165 - val_loss: 4.0625 - val_accuracy: 0.0874\n",
+      "Epoch 4068/5000\n",
+      "919/919 - 3s - loss: 1.3463 - accuracy: 0.6195 - val_loss: 4.0490 - val_accuracy: 0.0874\n",
+      "Epoch 4069/5000\n",
+      "919/919 - 3s - loss: 1.2253 - accuracy: 0.6186 - val_loss: 4.0583 - val_accuracy: 0.0869\n",
+      "Epoch 4070/5000\n",
+      "919/919 - 3s - loss: 1.2647 - accuracy: 0.6138 - val_loss: 4.0676 - val_accuracy: 0.0870\n",
+      "Epoch 4071/5000\n",
+      "919/919 - 3s - loss: 1.2310 - accuracy: 0.6173 - val_loss: 4.0580 - val_accuracy: 0.0869\n",
+      "Epoch 4072/5000\n",
+      "919/919 - 3s - loss: 1.2141 - accuracy: 0.6182 - val_loss: 4.0493 - val_accuracy: 0.0871\n",
+      "Epoch 4073/5000\n",
+      "919/919 - 3s - loss: 1.2230 - accuracy: 0.6158 - val_loss: 4.0418 - val_accuracy: 0.0867\n",
+      "Epoch 4074/5000\n",
+      "919/919 - 3s - loss: 1.2431 - accuracy: 0.6172 - val_loss: 4.0598 - val_accuracy: 0.0870\n",
+      "Epoch 4075/5000\n",
+      "919/919 - 3s - loss: 1.2508 - accuracy: 0.6137 - val_loss: 4.0569 - val_accuracy: 0.0863\n",
+      "Epoch 4076/5000\n",
+      "919/919 - 3s - loss: 1.2117 - accuracy: 0.6197 - val_loss: 4.0663 - val_accuracy: 0.0867\n",
+      "Epoch 4077/5000\n",
+      "919/919 - 3s - loss: 1.2021 - accuracy: 0.6206 - val_loss: 4.0804 - val_accuracy: 0.0868\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4078/5000\n",
+      "919/919 - 3s - loss: 1.2116 - accuracy: 0.6213 - val_loss: 4.0958 - val_accuracy: 0.0873\n",
+      "Epoch 4079/5000\n",
+      "919/919 - 3s - loss: 1.2255 - accuracy: 0.6195 - val_loss: 4.0933 - val_accuracy: 0.0870\n",
+      "Epoch 4080/5000\n",
+      "919/919 - 3s - loss: 1.2223 - accuracy: 0.6181 - val_loss: 4.0882 - val_accuracy: 0.0873\n",
+      "Epoch 4081/5000\n",
+      "919/919 - 3s - loss: 1.2306 - accuracy: 0.6148 - val_loss: 4.0761 - val_accuracy: 0.0875\n",
+      "Epoch 4082/5000\n",
+      "919/919 - 3s - loss: 1.2242 - accuracy: 0.6176 - val_loss: 4.0708 - val_accuracy: 0.0881\n",
+      "Epoch 4083/5000\n",
+      "919/919 - 3s - loss: 1.2217 - accuracy: 0.6185 - val_loss: 4.0761 - val_accuracy: 0.0880\n",
+      "Epoch 4084/5000\n",
+      "919/919 - 3s - loss: 1.2215 - accuracy: 0.6160 - val_loss: 4.0782 - val_accuracy: 0.0882\n",
+      "Epoch 4085/5000\n",
+      "919/919 - 3s - loss: 1.3251 - accuracy: 0.6194 - val_loss: 4.0766 - val_accuracy: 0.0882\n",
+      "Epoch 4086/5000\n",
+      "919/919 - 3s - loss: 1.2183 - accuracy: 0.6150 - val_loss: 4.0710 - val_accuracy: 0.0883\n",
+      "Epoch 4087/5000\n",
+      "919/919 - 3s - loss: 1.3116 - accuracy: 0.6201 - val_loss: 4.0764 - val_accuracy: 0.0875\n",
+      "Epoch 4088/5000\n",
+      "919/919 - 3s - loss: 1.2425 - accuracy: 0.6143 - val_loss: 4.0743 - val_accuracy: 0.0881\n",
+      "Epoch 4089/5000\n",
+      "919/919 - 3s - loss: 1.2384 - accuracy: 0.6151 - val_loss: 4.0573 - val_accuracy: 0.0882\n",
+      "Epoch 4090/5000\n",
+      "919/919 - 3s - loss: 1.2143 - accuracy: 0.6207 - val_loss: 4.0686 - val_accuracy: 0.0875\n",
+      "Epoch 4091/5000\n",
+      "919/919 - 3s - loss: 1.2271 - accuracy: 0.6164 - val_loss: 4.0659 - val_accuracy: 0.0874\n",
+      "Epoch 4092/5000\n",
+      "919/919 - 3s - loss: 1.2550 - accuracy: 0.6217 - val_loss: 4.0564 - val_accuracy: 0.0880\n",
+      "Epoch 4093/5000\n",
+      "919/919 - 3s - loss: 1.2253 - accuracy: 0.6193 - val_loss: 4.0578 - val_accuracy: 0.0879\n",
+      "Epoch 4094/5000\n",
+      "919/919 - 3s - loss: 1.2244 - accuracy: 0.6184 - val_loss: 4.0664 - val_accuracy: 0.0875\n",
+      "Epoch 4095/5000\n",
+      "919/919 - 3s - loss: 1.3064 - accuracy: 0.6171 - val_loss: 4.0667 - val_accuracy: 0.0871\n",
+      "Epoch 4096/5000\n",
+      "919/919 - 3s - loss: 1.2087 - accuracy: 0.6208 - val_loss: 4.0806 - val_accuracy: 0.0873\n",
+      "Epoch 4097/5000\n",
+      "919/919 - 3s - loss: 1.2030 - accuracy: 0.6218 - val_loss: 4.0773 - val_accuracy: 0.0874\n",
+      "Epoch 4098/5000\n",
+      "919/919 - 3s - loss: 1.2237 - accuracy: 0.6169 - val_loss: 4.0735 - val_accuracy: 0.0876\n",
+      "Epoch 4099/5000\n",
+      "919/919 - 3s - loss: 1.2594 - accuracy: 0.6182 - val_loss: 4.0778 - val_accuracy: 0.0874\n",
+      "Epoch 4100/5000\n",
+      "919/919 - 3s - loss: 1.2466 - accuracy: 0.6169 - val_loss: 4.0779 - val_accuracy: 0.0874\n",
+      "Epoch 4101/5000\n",
+      "919/919 - 3s - loss: 1.2138 - accuracy: 0.6200 - val_loss: 4.0632 - val_accuracy: 0.0882\n",
+      "Epoch 4102/5000\n",
+      "919/919 - 3s - loss: 1.2355 - accuracy: 0.6168 - val_loss: 4.0722 - val_accuracy: 0.0878\n",
+      "Epoch 4103/5000\n",
+      "919/919 - 3s - loss: 1.2171 - accuracy: 0.6214 - val_loss: 4.0876 - val_accuracy: 0.0880\n",
+      "Epoch 4104/5000\n",
+      "919/919 - 3s - loss: 1.2116 - accuracy: 0.6188 - val_loss: 4.0831 - val_accuracy: 0.0876\n",
+      "Epoch 4105/5000\n",
+      "919/919 - 3s - loss: 1.2074 - accuracy: 0.6220 - val_loss: 4.0780 - val_accuracy: 0.0876\n",
+      "Epoch 4106/5000\n",
+      "919/919 - 3s - loss: 1.2369 - accuracy: 0.6152 - val_loss: 4.0787 - val_accuracy: 0.0874\n",
+      "Epoch 4107/5000\n",
+      "919/919 - 3s - loss: 1.2196 - accuracy: 0.6193 - val_loss: 4.0701 - val_accuracy: 0.0880\n",
+      "Epoch 4108/5000\n",
+      "919/919 - 3s - loss: 1.2278 - accuracy: 0.6151 - val_loss: 4.0799 - val_accuracy: 0.0875\n",
+      "Epoch 4109/5000\n",
+      "919/919 - 3s - loss: 1.2378 - accuracy: 0.6174 - val_loss: 4.0969 - val_accuracy: 0.0872\n",
+      "Epoch 4110/5000\n",
+      "919/919 - 3s - loss: 1.2974 - accuracy: 0.6180 - val_loss: 4.0829 - val_accuracy: 0.0882\n",
+      "Epoch 4111/5000\n",
+      "919/919 - 3s - loss: 1.2107 - accuracy: 0.6213 - val_loss: 4.0911 - val_accuracy: 0.0878\n",
+      "Epoch 4112/5000\n",
+      "919/919 - 3s - loss: 1.2170 - accuracy: 0.6228 - val_loss: 4.0912 - val_accuracy: 0.0871\n",
+      "Epoch 4113/5000\n",
+      "919/919 - 3s - loss: 1.2134 - accuracy: 0.6182 - val_loss: 4.0978 - val_accuracy: 0.0874\n",
+      "Epoch 4114/5000\n",
+      "919/919 - 3s - loss: 1.2234 - accuracy: 0.6180 - val_loss: 4.0871 - val_accuracy: 0.0878\n",
+      "Epoch 4115/5000\n",
+      "919/919 - 3s - loss: 1.2214 - accuracy: 0.6185 - val_loss: 4.0785 - val_accuracy: 0.0874\n",
+      "Epoch 4116/5000\n",
+      "919/919 - 3s - loss: 1.2084 - accuracy: 0.6206 - val_loss: 4.0998 - val_accuracy: 0.0878\n",
+      "Epoch 4117/5000\n",
+      "919/919 - 3s - loss: 1.1998 - accuracy: 0.6202 - val_loss: 4.1030 - val_accuracy: 0.0872\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4118/5000\n",
+      "919/919 - 3s - loss: 1.2164 - accuracy: 0.6163 - val_loss: 4.1032 - val_accuracy: 0.0874\n",
+      "Epoch 4119/5000\n",
+      "919/919 - 3s - loss: 1.2246 - accuracy: 0.6204 - val_loss: 4.1026 - val_accuracy: 0.0877\n",
+      "Epoch 4120/5000\n",
+      "919/919 - 3s - loss: 1.2167 - accuracy: 0.6220 - val_loss: 4.0799 - val_accuracy: 0.0876\n",
+      "Epoch 4121/5000\n",
+      "919/919 - 3s - loss: 1.2300 - accuracy: 0.6167 - val_loss: 4.0882 - val_accuracy: 0.0871\n",
+      "Epoch 4122/5000\n",
+      "919/919 - 3s - loss: 1.2175 - accuracy: 0.6183 - val_loss: 4.0840 - val_accuracy: 0.0877\n",
+      "Epoch 4123/5000\n",
+      "919/919 - 3s - loss: 1.2393 - accuracy: 0.6166 - val_loss: 4.0601 - val_accuracy: 0.0870\n",
+      "Epoch 4124/5000\n",
+      "919/919 - 3s - loss: 1.2860 - accuracy: 0.6209 - val_loss: 4.0685 - val_accuracy: 0.0869\n",
+      "Epoch 4125/5000\n",
+      "919/919 - 3s - loss: 1.2189 - accuracy: 0.6190 - val_loss: 4.0614 - val_accuracy: 0.0871\n",
+      "Epoch 4126/5000\n",
+      "919/919 - 3s - loss: 1.2198 - accuracy: 0.6178 - val_loss: 4.0564 - val_accuracy: 0.0872\n",
+      "Epoch 4127/5000\n",
+      "919/919 - 3s - loss: 1.2434 - accuracy: 0.6183 - val_loss: 4.0591 - val_accuracy: 0.0874\n",
+      "Epoch 4128/5000\n",
+      "919/919 - 3s - loss: 1.2295 - accuracy: 0.6163 - val_loss: 4.0778 - val_accuracy: 0.0871\n",
+      "Epoch 4129/5000\n",
+      "919/919 - 3s - loss: 1.2203 - accuracy: 0.6174 - val_loss: 4.0755 - val_accuracy: 0.0872\n",
+      "Epoch 4130/5000\n",
+      "919/919 - 3s - loss: 1.2412 - accuracy: 0.6191 - val_loss: 4.0707 - val_accuracy: 0.0874\n",
+      "Epoch 4131/5000\n",
+      "919/919 - 3s - loss: 1.2866 - accuracy: 0.6199 - val_loss: 4.0825 - val_accuracy: 0.0872\n",
+      "Epoch 4132/5000\n",
+      "919/919 - 3s - loss: 1.2330 - accuracy: 0.6185 - val_loss: 4.0815 - val_accuracy: 0.0868\n",
+      "Epoch 4133/5000\n",
+      "919/919 - 3s - loss: 1.2169 - accuracy: 0.6150 - val_loss: 4.0634 - val_accuracy: 0.0871\n",
+      "Epoch 4134/5000\n",
+      "919/919 - 3s - loss: 1.2054 - accuracy: 0.6220 - val_loss: 4.0754 - val_accuracy: 0.0867\n",
+      "Epoch 4135/5000\n",
+      "919/919 - 3s - loss: 1.2036 - accuracy: 0.6216 - val_loss: 4.0661 - val_accuracy: 0.0867\n",
+      "Epoch 4136/5000\n",
+      "919/919 - 3s - loss: 1.2217 - accuracy: 0.6226 - val_loss: 4.0667 - val_accuracy: 0.0874\n",
+      "Epoch 4137/5000\n",
+      "919/919 - 3s - loss: 1.2884 - accuracy: 0.6203 - val_loss: 4.0704 - val_accuracy: 0.0874\n",
+      "Epoch 4138/5000\n",
+      "919/919 - 3s - loss: 1.2206 - accuracy: 0.6193 - val_loss: 4.0880 - val_accuracy: 0.0875\n",
+      "Epoch 4139/5000\n",
+      "919/919 - 3s - loss: 1.2170 - accuracy: 0.6207 - val_loss: 4.0870 - val_accuracy: 0.0872\n",
+      "Epoch 4140/5000\n",
+      "919/919 - 3s - loss: 1.2150 - accuracy: 0.6216 - val_loss: 4.0908 - val_accuracy: 0.0880\n",
+      "Epoch 4141/5000\n",
+      "919/919 - 3s - loss: 1.2235 - accuracy: 0.6203 - val_loss: 4.0884 - val_accuracy: 0.0875\n",
+      "Epoch 4142/5000\n",
+      "919/919 - 3s - loss: 1.2096 - accuracy: 0.6222 - val_loss: 4.0872 - val_accuracy: 0.0874\n",
+      "Epoch 4143/5000\n",
+      "919/919 - 3s - loss: 1.2146 - accuracy: 0.6193 - val_loss: 4.0808 - val_accuracy: 0.0879\n",
+      "Epoch 4144/5000\n",
+      "919/919 - 3s - loss: 1.2182 - accuracy: 0.6194 - val_loss: 4.0688 - val_accuracy: 0.0875\n",
+      "Epoch 4145/5000\n",
+      "919/919 - 3s - loss: 1.2189 - accuracy: 0.6169 - val_loss: 4.0600 - val_accuracy: 0.0873\n",
+      "Epoch 4146/5000\n",
+      "919/919 - 3s - loss: 1.2283 - accuracy: 0.6156 - val_loss: 4.0675 - val_accuracy: 0.0877\n",
+      "Epoch 4147/5000\n",
+      "919/919 - 3s - loss: 1.2164 - accuracy: 0.6191 - val_loss: 4.0838 - val_accuracy: 0.0871\n",
+      "Epoch 4148/5000\n",
+      "919/919 - 3s - loss: 1.2172 - accuracy: 0.6196 - val_loss: 4.0702 - val_accuracy: 0.0874\n",
+      "Epoch 4149/5000\n",
+      "919/919 - 3s - loss: 1.2189 - accuracy: 0.6164 - val_loss: 4.0827 - val_accuracy: 0.0878\n",
+      "Epoch 4150/5000\n",
+      "919/919 - 3s - loss: 1.2222 - accuracy: 0.6173 - val_loss: 4.0719 - val_accuracy: 0.0876\n",
+      "Epoch 4151/5000\n",
+      "919/919 - 3s - loss: 1.1982 - accuracy: 0.6209 - val_loss: 4.0718 - val_accuracy: 0.0883\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4152/5000\n",
+      "919/919 - 3s - loss: 1.2192 - accuracy: 0.6170 - val_loss: 4.0766 - val_accuracy: 0.0876\n",
+      "Epoch 4153/5000\n",
+      "919/919 - 3s - loss: 1.2158 - accuracy: 0.6160 - val_loss: 4.0678 - val_accuracy: 0.0874\n",
+      "Epoch 4154/5000\n",
+      "919/919 - 3s - loss: 1.2007 - accuracy: 0.6226 - val_loss: 4.0988 - val_accuracy: 0.0874\n",
+      "Epoch 4155/5000\n",
+      "919/919 - 3s - loss: 1.2194 - accuracy: 0.6201 - val_loss: 4.0917 - val_accuracy: 0.0880\n",
+      "Epoch 4156/5000\n",
+      "919/919 - 3s - loss: 1.3282 - accuracy: 0.6185 - val_loss: 4.0875 - val_accuracy: 0.0879\n",
+      "Epoch 4157/5000\n",
+      "919/919 - 3s - loss: 1.2125 - accuracy: 0.6226 - val_loss: 4.0825 - val_accuracy: 0.0872\n",
+      "Epoch 4158/5000\n",
+      "919/919 - 3s - loss: 1.2026 - accuracy: 0.6190 - val_loss: 4.0869 - val_accuracy: 0.0878\n",
+      "Epoch 4159/5000\n",
+      "919/919 - 3s - loss: 1.2775 - accuracy: 0.6207 - val_loss: 4.0958 - val_accuracy: 0.0878\n",
+      "Epoch 4160/5000\n",
+      "919/919 - 3s - loss: 1.2069 - accuracy: 0.6195 - val_loss: 4.0878 - val_accuracy: 0.0874\n",
+      "Epoch 4161/5000\n",
+      "919/919 - 3s - loss: 1.2045 - accuracy: 0.6232 - val_loss: 4.0971 - val_accuracy: 0.0876\n",
+      "Epoch 4162/5000\n",
+      "919/919 - 3s - loss: 1.2086 - accuracy: 0.6231 - val_loss: 4.0916 - val_accuracy: 0.0881\n",
+      "Epoch 4163/5000\n",
+      "919/919 - 3s - loss: 1.2229 - accuracy: 0.6181 - val_loss: 4.0881 - val_accuracy: 0.0877\n",
+      "Epoch 4164/5000\n",
+      "919/919 - 3s - loss: 1.2219 - accuracy: 0.6203 - val_loss: 4.0857 - val_accuracy: 0.0875\n",
+      "Epoch 4165/5000\n",
+      "919/919 - 3s - loss: 1.2193 - accuracy: 0.6171 - val_loss: 4.0927 - val_accuracy: 0.0882\n",
+      "Epoch 4166/5000\n",
+      "919/919 - 3s - loss: 1.2321 - accuracy: 0.6161 - val_loss: 4.0880 - val_accuracy: 0.0882\n",
+      "Epoch 4167/5000\n",
+      "919/919 - 3s - loss: 1.2277 - accuracy: 0.6171 - val_loss: 4.0908 - val_accuracy: 0.0881\n",
+      "Epoch 4168/5000\n",
+      "919/919 - 3s - loss: 1.2094 - accuracy: 0.6188 - val_loss: 4.0803 - val_accuracy: 0.0876\n",
+      "Epoch 4169/5000\n",
+      "919/919 - 3s - loss: 1.2329 - accuracy: 0.6182 - val_loss: 4.0831 - val_accuracy: 0.0885\n",
+      "Epoch 4170/5000\n",
+      "919/919 - 3s - loss: 1.2239 - accuracy: 0.6220 - val_loss: 4.0919 - val_accuracy: 0.0884\n",
+      "Epoch 4171/5000\n",
+      "919/919 - 3s - loss: 1.1919 - accuracy: 0.6210 - val_loss: 4.0956 - val_accuracy: 0.0885\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4172/5000\n",
+      "919/919 - 3s - loss: 1.2083 - accuracy: 0.6231 - val_loss: 4.0979 - val_accuracy: 0.0874\n",
+      "Epoch 4173/5000\n",
+      "919/919 - 3s - loss: 1.2122 - accuracy: 0.6248 - val_loss: 4.1026 - val_accuracy: 0.0872\n",
+      "Epoch 4174/5000\n",
+      "919/919 - 3s - loss: 1.2190 - accuracy: 0.6176 - val_loss: 4.0987 - val_accuracy: 0.0874\n",
+      "Epoch 4175/5000\n",
+      "919/919 - 3s - loss: 1.2129 - accuracy: 0.6216 - val_loss: 4.0989 - val_accuracy: 0.0880\n",
+      "Epoch 4176/5000\n",
+      "919/919 - 3s - loss: 1.2237 - accuracy: 0.6207 - val_loss: 4.0949 - val_accuracy: 0.0875\n",
+      "Epoch 4177/5000\n",
+      "919/919 - 3s - loss: 1.2409 - accuracy: 0.6174 - val_loss: 4.1089 - val_accuracy: 0.0874\n",
+      "Epoch 4178/5000\n",
+      "919/919 - 3s - loss: 1.2531 - accuracy: 0.6176 - val_loss: 4.1188 - val_accuracy: 0.0877\n",
+      "Epoch 4179/5000\n",
+      "919/919 - 3s - loss: 1.1979 - accuracy: 0.6278 - val_loss: 4.1273 - val_accuracy: 0.0881\n",
+      "Epoch 4180/5000\n",
+      "919/919 - 3s - loss: 1.2202 - accuracy: 0.6207 - val_loss: 4.1154 - val_accuracy: 0.0874\n",
+      "Epoch 4181/5000\n",
+      "919/919 - 3s - loss: 1.2620 - accuracy: 0.6222 - val_loss: 4.1153 - val_accuracy: 0.0879\n",
+      "Epoch 4182/5000\n",
+      "919/919 - 3s - loss: 1.2193 - accuracy: 0.6183 - val_loss: 4.1218 - val_accuracy: 0.0874\n",
+      "Epoch 4183/5000\n",
+      "919/919 - 3s - loss: 1.2123 - accuracy: 0.6230 - val_loss: 4.1130 - val_accuracy: 0.0874\n",
+      "Epoch 4184/5000\n",
+      "919/919 - 3s - loss: 1.2266 - accuracy: 0.6181 - val_loss: 4.1141 - val_accuracy: 0.0874\n",
+      "Epoch 4185/5000\n",
+      "919/919 - 3s - loss: 1.2034 - accuracy: 0.6230 - val_loss: 4.1076 - val_accuracy: 0.0876\n",
+      "Epoch 4186/5000\n",
+      "919/919 - 3s - loss: 1.2192 - accuracy: 0.6217 - val_loss: 4.0982 - val_accuracy: 0.0876\n",
+      "Epoch 4187/5000\n",
+      "919/919 - 3s - loss: 1.3379 - accuracy: 0.6210 - val_loss: 4.0933 - val_accuracy: 0.0877\n",
+      "Epoch 4188/5000\n",
+      "919/919 - 3s - loss: 1.2486 - accuracy: 0.6216 - val_loss: 4.1047 - val_accuracy: 0.0878\n",
+      "Epoch 4189/5000\n",
+      "919/919 - 3s - loss: 1.2138 - accuracy: 0.6178 - val_loss: 4.0993 - val_accuracy: 0.0871\n",
+      "Epoch 4190/5000\n",
+      "919/919 - 3s - loss: 1.2253 - accuracy: 0.6154 - val_loss: 4.0906 - val_accuracy: 0.0877\n",
+      "Epoch 4191/5000\n",
+      "919/919 - 3s - loss: 1.2211 - accuracy: 0.6191 - val_loss: 4.0945 - val_accuracy: 0.0878\n",
+      "Epoch 4192/5000\n",
+      "919/919 - 3s - loss: 1.2345 - accuracy: 0.6168 - val_loss: 4.0995 - val_accuracy: 0.0871\n",
+      "Epoch 4193/5000\n",
+      "919/919 - 3s - loss: 1.2432 - accuracy: 0.6210 - val_loss: 4.1058 - val_accuracy: 0.0873\n",
+      "Epoch 4194/5000\n",
+      "919/919 - 3s - loss: 1.2090 - accuracy: 0.6202 - val_loss: 4.1104 - val_accuracy: 0.0876\n",
+      "Epoch 4195/5000\n",
+      "919/919 - 3s - loss: 1.2323 - accuracy: 0.6213 - val_loss: 4.1122 - val_accuracy: 0.0878\n",
+      "Epoch 4196/5000\n",
+      "919/919 - 3s - loss: 1.3923 - accuracy: 0.6197 - val_loss: 4.1100 - val_accuracy: 0.0871\n",
+      "Epoch 4197/5000\n",
+      "919/919 - 3s - loss: 1.2101 - accuracy: 0.6238 - val_loss: 4.0982 - val_accuracy: 0.0877\n",
+      "Epoch 4198/5000\n",
+      "919/919 - 3s - loss: 1.2305 - accuracy: 0.6178 - val_loss: 4.0882 - val_accuracy: 0.0874\n",
+      "Epoch 4199/5000\n",
+      "919/919 - 3s - loss: 1.1950 - accuracy: 0.6238 - val_loss: 4.1016 - val_accuracy: 0.0874\n",
+      "Epoch 4200/5000\n",
+      "919/919 - 3s - loss: 1.2156 - accuracy: 0.6218 - val_loss: 4.1058 - val_accuracy: 0.0875\n",
+      "Epoch 4201/5000\n",
+      "919/919 - 3s - loss: 1.2203 - accuracy: 0.6194 - val_loss: 4.0907 - val_accuracy: 0.0876\n",
+      "Epoch 4202/5000\n",
+      "919/919 - 3s - loss: 1.2096 - accuracy: 0.6231 - val_loss: 4.0952 - val_accuracy: 0.0880\n",
+      "Epoch 4203/5000\n",
+      "919/919 - 3s - loss: 1.2301 - accuracy: 0.6217 - val_loss: 4.0810 - val_accuracy: 0.0874\n",
+      "Epoch 4204/5000\n",
+      "919/919 - 3s - loss: 1.2089 - accuracy: 0.6212 - val_loss: 4.0986 - val_accuracy: 0.0874\n",
+      "Epoch 4205/5000\n",
+      "919/919 - 3s - loss: 1.2816 - accuracy: 0.6249 - val_loss: 4.1191 - val_accuracy: 0.0877\n",
+      "Epoch 4206/5000\n",
+      "919/919 - 3s - loss: 1.2246 - accuracy: 0.6191 - val_loss: 4.1200 - val_accuracy: 0.0876\n",
+      "Epoch 4207/5000\n",
+      "919/919 - 3s - loss: 1.1977 - accuracy: 0.6220 - val_loss: 4.1397 - val_accuracy: 0.0872\n",
+      "Epoch 4208/5000\n",
+      "919/919 - 3s - loss: 1.2254 - accuracy: 0.6223 - val_loss: 4.1347 - val_accuracy: 0.0874\n",
+      "Epoch 4209/5000\n",
+      "919/919 - 3s - loss: 1.2208 - accuracy: 0.6195 - val_loss: 4.1330 - val_accuracy: 0.0871\n",
+      "Epoch 4210/5000\n",
+      "919/919 - 3s - loss: 1.2200 - accuracy: 0.6214 - val_loss: 4.1131 - val_accuracy: 0.0879\n",
+      "Epoch 4211/5000\n",
+      "919/919 - 3s - loss: 1.1841 - accuracy: 0.6256 - val_loss: 4.1218 - val_accuracy: 0.0877\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4212/5000\n",
+      "919/919 - 3s - loss: 1.2158 - accuracy: 0.6205 - val_loss: 4.1154 - val_accuracy: 0.0879\n",
+      "Epoch 4213/5000\n",
+      "919/919 - 3s - loss: 1.2196 - accuracy: 0.6214 - val_loss: 4.1040 - val_accuracy: 0.0879\n",
+      "Epoch 4214/5000\n",
+      "919/919 - 3s - loss: 1.2229 - accuracy: 0.6222 - val_loss: 4.0985 - val_accuracy: 0.0878\n",
+      "Epoch 4215/5000\n",
+      "919/919 - 3s - loss: 1.2159 - accuracy: 0.6193 - val_loss: 4.1034 - val_accuracy: 0.0876\n",
+      "Epoch 4216/5000\n",
+      "919/919 - 3s - loss: 1.2202 - accuracy: 0.6229 - val_loss: 4.1031 - val_accuracy: 0.0877\n",
+      "Epoch 4217/5000\n",
+      "919/919 - 3s - loss: 1.2305 - accuracy: 0.6188 - val_loss: 4.1036 - val_accuracy: 0.0883\n",
+      "Epoch 4218/5000\n",
+      "919/919 - 3s - loss: 1.2159 - accuracy: 0.6214 - val_loss: 4.0989 - val_accuracy: 0.0885\n",
+      "Epoch 4219/5000\n",
+      "919/919 - 3s - loss: 1.2098 - accuracy: 0.6250 - val_loss: 4.0893 - val_accuracy: 0.0892\n",
+      "Epoch 4220/5000\n",
+      "919/919 - 3s - loss: 1.2755 - accuracy: 0.6185 - val_loss: 4.0950 - val_accuracy: 0.0882\n",
+      "Epoch 4221/5000\n",
+      "919/919 - 3s - loss: 1.2181 - accuracy: 0.6223 - val_loss: 4.0917 - val_accuracy: 0.0884\n",
+      "Epoch 4222/5000\n",
+      "919/919 - 3s - loss: 1.2031 - accuracy: 0.6219 - val_loss: 4.0978 - val_accuracy: 0.0880\n",
+      "Epoch 4223/5000\n",
+      "919/919 - 3s - loss: 1.2102 - accuracy: 0.6220 - val_loss: 4.0858 - val_accuracy: 0.0883\n",
+      "Epoch 4224/5000\n",
+      "919/919 - 3s - loss: 1.2669 - accuracy: 0.6205 - val_loss: 4.0980 - val_accuracy: 0.0883\n",
+      "Epoch 4225/5000\n",
+      "919/919 - 3s - loss: 1.2089 - accuracy: 0.6230 - val_loss: 4.0855 - val_accuracy: 0.0884\n",
+      "Epoch 4226/5000\n",
+      "919/919 - 3s - loss: 1.1986 - accuracy: 0.6230 - val_loss: 4.1088 - val_accuracy: 0.0877\n",
+      "Epoch 4227/5000\n",
+      "919/919 - 3s - loss: 1.2323 - accuracy: 0.6201 - val_loss: 4.0964 - val_accuracy: 0.0876\n",
+      "Epoch 4228/5000\n",
+      "919/919 - 3s - loss: 1.2595 - accuracy: 0.6222 - val_loss: 4.0979 - val_accuracy: 0.0875\n",
+      "Epoch 4229/5000\n",
+      "919/919 - 3s - loss: 1.3044 - accuracy: 0.6216 - val_loss: 4.0938 - val_accuracy: 0.0875\n",
+      "Epoch 4230/5000\n",
+      "919/919 - 3s - loss: 1.2055 - accuracy: 0.6203 - val_loss: 4.0966 - val_accuracy: 0.0874\n",
+      "Epoch 4231/5000\n",
+      "919/919 - 3s - loss: 1.2185 - accuracy: 0.6215 - val_loss: 4.1006 - val_accuracy: 0.0879\n",
+      "Epoch 4232/5000\n",
+      "919/919 - 3s - loss: 1.2726 - accuracy: 0.6205 - val_loss: 4.1068 - val_accuracy: 0.0879\n",
+      "Epoch 4233/5000\n",
+      "919/919 - 3s - loss: 1.2059 - accuracy: 0.6222 - val_loss: 4.1208 - val_accuracy: 0.0880\n",
+      "Epoch 4234/5000\n",
+      "919/919 - 3s - loss: 1.2035 - accuracy: 0.6218 - val_loss: 4.1146 - val_accuracy: 0.0880\n",
+      "Epoch 4235/5000\n",
+      "919/919 - 3s - loss: 1.2105 - accuracy: 0.6224 - val_loss: 4.1239 - val_accuracy: 0.0873\n",
+      "Epoch 4236/5000\n",
+      "919/919 - 3s - loss: 1.2251 - accuracy: 0.6197 - val_loss: 4.1054 - val_accuracy: 0.0873\n",
+      "Epoch 4237/5000\n",
+      "919/919 - 3s - loss: 1.2091 - accuracy: 0.6202 - val_loss: 4.1171 - val_accuracy: 0.0876\n",
+      "Epoch 4238/5000\n",
+      "919/919 - 3s - loss: 1.1876 - accuracy: 0.6281 - val_loss: 4.1156 - val_accuracy: 0.0876\n",
+      "Epoch 4239/5000\n",
+      "919/919 - 3s - loss: 1.2072 - accuracy: 0.6239 - val_loss: 4.1115 - val_accuracy: 0.0877\n",
+      "Epoch 4240/5000\n",
+      "919/919 - 3s - loss: 1.2074 - accuracy: 0.6226 - val_loss: 4.0994 - val_accuracy: 0.0876\n",
+      "Epoch 4241/5000\n",
+      "919/919 - 3s - loss: 1.2020 - accuracy: 0.6211 - val_loss: 4.1083 - val_accuracy: 0.0873\n",
+      "Epoch 4242/5000\n",
+      "919/919 - 3s - loss: 1.2164 - accuracy: 0.6205 - val_loss: 4.0867 - val_accuracy: 0.0876\n",
+      "Epoch 4243/5000\n",
+      "919/919 - 3s - loss: 1.2080 - accuracy: 0.6259 - val_loss: 4.0864 - val_accuracy: 0.0880\n",
+      "Epoch 4244/5000\n",
+      "919/919 - 3s - loss: 1.2122 - accuracy: 0.6190 - val_loss: 4.0945 - val_accuracy: 0.0875\n",
+      "Epoch 4245/5000\n",
+      "919/919 - 3s - loss: 1.2062 - accuracy: 0.6246 - val_loss: 4.0942 - val_accuracy: 0.0877\n",
+      "Epoch 4246/5000\n",
+      "919/919 - 3s - loss: 1.2294 - accuracy: 0.6176 - val_loss: 4.0957 - val_accuracy: 0.0875\n",
+      "Epoch 4247/5000\n",
+      "919/919 - 3s - loss: 1.2075 - accuracy: 0.6192 - val_loss: 4.0849 - val_accuracy: 0.0880\n",
+      "Epoch 4248/5000\n",
+      "919/919 - 3s - loss: 1.2183 - accuracy: 0.6235 - val_loss: 4.0904 - val_accuracy: 0.0877\n",
+      "Epoch 4249/5000\n",
+      "919/919 - 3s - loss: 1.2086 - accuracy: 0.6229 - val_loss: 4.0932 - val_accuracy: 0.0882\n",
+      "Epoch 4250/5000\n",
+      "919/919 - 3s - loss: 1.2098 - accuracy: 0.6213 - val_loss: 4.0934 - val_accuracy: 0.0883\n",
+      "Epoch 4251/5000\n",
+      "919/919 - 3s - loss: 1.2407 - accuracy: 0.6212 - val_loss: 4.0937 - val_accuracy: 0.0883\n",
+      "Epoch 4252/5000\n",
+      "919/919 - 3s - loss: 1.3168 - accuracy: 0.6176 - val_loss: 4.1034 - val_accuracy: 0.0885\n",
+      "Epoch 4253/5000\n",
+      "919/919 - 3s - loss: 1.2157 - accuracy: 0.6226 - val_loss: 4.0975 - val_accuracy: 0.0882\n",
+      "Epoch 4254/5000\n",
+      "919/919 - 3s - loss: 1.2161 - accuracy: 0.6207 - val_loss: 4.1021 - val_accuracy: 0.0881\n",
+      "Epoch 4255/5000\n",
+      "919/919 - 3s - loss: 1.2194 - accuracy: 0.6244 - val_loss: 4.1206 - val_accuracy: 0.0879\n",
+      "Epoch 4256/5000\n",
+      "919/919 - 3s - loss: 1.2031 - accuracy: 0.6236 - val_loss: 4.1321 - val_accuracy: 0.0879\n",
+      "Epoch 4257/5000\n",
+      "919/919 - 3s - loss: 1.2500 - accuracy: 0.6201 - val_loss: 4.1160 - val_accuracy: 0.0876\n",
+      "Epoch 4258/5000\n",
+      "919/919 - 3s - loss: 1.2192 - accuracy: 0.6171 - val_loss: 4.0996 - val_accuracy: 0.0874\n",
+      "Epoch 4259/5000\n",
+      "919/919 - 3s - loss: 1.1999 - accuracy: 0.6214 - val_loss: 4.1195 - val_accuracy: 0.0877\n",
+      "Epoch 4260/5000\n",
+      "919/919 - 3s - loss: 1.2182 - accuracy: 0.6198 - val_loss: 4.1001 - val_accuracy: 0.0874\n",
+      "Epoch 4261/5000\n",
+      "919/919 - 3s - loss: 1.2172 - accuracy: 0.6223 - val_loss: 4.1064 - val_accuracy: 0.0872\n",
+      "Epoch 4262/5000\n",
+      "919/919 - 3s - loss: 1.2087 - accuracy: 0.6224 - val_loss: 4.1131 - val_accuracy: 0.0867\n",
+      "Epoch 4263/5000\n",
+      "919/919 - 3s - loss: 1.2006 - accuracy: 0.6199 - val_loss: 4.1126 - val_accuracy: 0.0873\n",
+      "Epoch 4264/5000\n",
+      "919/919 - 3s - loss: 1.2112 - accuracy: 0.6218 - val_loss: 4.1076 - val_accuracy: 0.0881\n",
+      "Epoch 4265/5000\n",
+      "919/919 - 3s - loss: 1.2138 - accuracy: 0.6193 - val_loss: 4.0931 - val_accuracy: 0.0877\n",
+      "Epoch 4266/5000\n",
+      "919/919 - 3s - loss: 1.2183 - accuracy: 0.6237 - val_loss: 4.0822 - val_accuracy: 0.0878\n",
+      "Epoch 4267/5000\n",
+      "919/919 - 3s - loss: 1.2018 - accuracy: 0.6285 - val_loss: 4.0990 - val_accuracy: 0.0879\n",
+      "Epoch 4268/5000\n",
+      "919/919 - 3s - loss: 1.2236 - accuracy: 0.6209 - val_loss: 4.1101 - val_accuracy: 0.0881\n",
+      "Epoch 4269/5000\n",
+      "919/919 - 3s - loss: 1.2238 - accuracy: 0.6195 - val_loss: 4.0994 - val_accuracy: 0.0879\n",
+      "Epoch 4270/5000\n",
+      "919/919 - 3s - loss: 1.2235 - accuracy: 0.6224 - val_loss: 4.1037 - val_accuracy: 0.0878\n",
+      "Epoch 4271/5000\n",
+      "919/919 - 3s - loss: 1.2176 - accuracy: 0.6194 - val_loss: 4.1005 - val_accuracy: 0.0874\n",
+      "Epoch 4272/5000\n",
+      "919/919 - 3s - loss: 1.2172 - accuracy: 0.6246 - val_loss: 4.1041 - val_accuracy: 0.0877\n",
+      "Epoch 4273/5000\n",
+      "919/919 - 3s - loss: 1.2820 - accuracy: 0.6207 - val_loss: 4.1197 - val_accuracy: 0.0874\n",
+      "Epoch 4274/5000\n",
+      "919/919 - 3s - loss: 1.2032 - accuracy: 0.6260 - val_loss: 4.1117 - val_accuracy: 0.0872\n",
+      "Epoch 4275/5000\n",
+      "919/919 - 3s - loss: 1.2221 - accuracy: 0.6203 - val_loss: 4.1064 - val_accuracy: 0.0874\n",
+      "Epoch 4276/5000\n",
+      "919/919 - 3s - loss: 1.2285 - accuracy: 0.6249 - val_loss: 4.1161 - val_accuracy: 0.0877\n",
+      "Epoch 4277/5000\n",
+      "919/919 - 3s - loss: 1.2123 - accuracy: 0.6219 - val_loss: 4.1254 - val_accuracy: 0.0876\n",
+      "Epoch 4278/5000\n",
+      "919/919 - 3s - loss: 1.2167 - accuracy: 0.6202 - val_loss: 4.1289 - val_accuracy: 0.0878\n",
+      "Epoch 4279/5000\n",
+      "919/919 - 3s - loss: 1.2156 - accuracy: 0.6257 - val_loss: 4.1371 - val_accuracy: 0.0876\n",
+      "Epoch 4280/5000\n",
+      "919/919 - 3s - loss: 1.2152 - accuracy: 0.6207 - val_loss: 4.1341 - val_accuracy: 0.0879\n",
+      "Epoch 4281/5000\n",
+      "919/919 - 3s - loss: 1.2226 - accuracy: 0.6207 - val_loss: 4.1271 - val_accuracy: 0.0873\n",
+      "Epoch 4282/5000\n",
+      "919/919 - 3s - loss: 1.2291 - accuracy: 0.6201 - val_loss: 4.1141 - val_accuracy: 0.0868\n",
+      "Epoch 4283/5000\n",
+      "919/919 - 3s - loss: 1.1938 - accuracy: 0.6300 - val_loss: 4.1171 - val_accuracy: 0.0874\n",
+      "Epoch 4284/5000\n",
+      "919/919 - 3s - loss: 1.2045 - accuracy: 0.6203 - val_loss: 4.1165 - val_accuracy: 0.0878\n",
+      "Epoch 4285/5000\n",
+      "919/919 - 3s - loss: 1.2092 - accuracy: 0.6226 - val_loss: 4.1152 - val_accuracy: 0.0875\n",
+      "Epoch 4286/5000\n",
+      "919/919 - 3s - loss: 1.1957 - accuracy: 0.6250 - val_loss: 4.1239 - val_accuracy: 0.0878\n",
+      "Epoch 4287/5000\n",
+      "919/919 - 3s - loss: 1.1961 - accuracy: 0.6234 - val_loss: 4.1119 - val_accuracy: 0.0881\n",
+      "Epoch 4288/5000\n",
+      "919/919 - 3s - loss: 1.2261 - accuracy: 0.6241 - val_loss: 4.1079 - val_accuracy: 0.0888\n",
+      "Epoch 4289/5000\n",
+      "919/919 - 3s - loss: 1.2080 - accuracy: 0.6218 - val_loss: 4.1091 - val_accuracy: 0.0883\n",
+      "Epoch 4290/5000\n",
+      "919/919 - 3s - loss: 1.2176 - accuracy: 0.6216 - val_loss: 4.0996 - val_accuracy: 0.0879\n",
+      "Epoch 4291/5000\n",
+      "919/919 - 3s - loss: 1.2042 - accuracy: 0.6211 - val_loss: 4.1107 - val_accuracy: 0.0882\n",
+      "Epoch 4292/5000\n",
+      "919/919 - 3s - loss: 1.2308 - accuracy: 0.6199 - val_loss: 4.1050 - val_accuracy: 0.0885\n",
+      "Epoch 4293/5000\n",
+      "919/919 - 3s - loss: 1.2000 - accuracy: 0.6270 - val_loss: 4.1196 - val_accuracy: 0.0879\n",
+      "Epoch 4294/5000\n",
+      "919/919 - 3s - loss: 1.2496 - accuracy: 0.6214 - val_loss: 4.1138 - val_accuracy: 0.0881\n",
+      "Epoch 4295/5000\n",
+      "919/919 - 3s - loss: 1.1909 - accuracy: 0.6254 - val_loss: 4.1159 - val_accuracy: 0.0882\n",
+      "Epoch 4296/5000\n",
+      "919/919 - 3s - loss: 1.2026 - accuracy: 0.6248 - val_loss: 4.1234 - val_accuracy: 0.0882\n",
+      "Epoch 4297/5000\n",
+      "919/919 - 3s - loss: 1.2020 - accuracy: 0.6236 - val_loss: 4.1216 - val_accuracy: 0.0885\n",
+      "Epoch 4298/5000\n",
+      "919/919 - 3s - loss: 1.2174 - accuracy: 0.6241 - val_loss: 4.1162 - val_accuracy: 0.0878\n",
+      "Epoch 4299/5000\n",
+      "919/919 - 3s - loss: 1.2083 - accuracy: 0.6247 - val_loss: 4.1371 - val_accuracy: 0.0883\n",
+      "Epoch 4300/5000\n",
+      "919/919 - 3s - loss: 1.2180 - accuracy: 0.6198 - val_loss: 4.1161 - val_accuracy: 0.0876\n",
+      "Epoch 4301/5000\n",
+      "919/919 - 3s - loss: 1.1903 - accuracy: 0.6263 - val_loss: 4.1149 - val_accuracy: 0.0871\n",
+      "Epoch 4302/5000\n",
+      "919/919 - 3s - loss: 1.2084 - accuracy: 0.6212 - val_loss: 4.1017 - val_accuracy: 0.0872\n",
+      "Epoch 4303/5000\n",
+      "919/919 - 3s - loss: 1.2072 - accuracy: 0.6239 - val_loss: 4.0901 - val_accuracy: 0.0883\n",
+      "Epoch 4304/5000\n",
+      "919/919 - 3s - loss: 1.2278 - accuracy: 0.6245 - val_loss: 4.1030 - val_accuracy: 0.0884\n",
+      "Epoch 4305/5000\n",
+      "919/919 - 3s - loss: 1.2043 - accuracy: 0.6228 - val_loss: 4.0973 - val_accuracy: 0.0886\n",
+      "Epoch 4306/5000\n",
+      "919/919 - 3s - loss: 1.2305 - accuracy: 0.6212 - val_loss: 4.0980 - val_accuracy: 0.0883\n",
+      "Epoch 4307/5000\n",
+      "919/919 - 3s - loss: 1.3399 - accuracy: 0.6237 - val_loss: 4.1241 - val_accuracy: 0.0878\n",
+      "Epoch 4308/5000\n",
+      "919/919 - 3s - loss: 1.1859 - accuracy: 0.6259 - val_loss: 4.1250 - val_accuracy: 0.0874\n",
+      "Epoch 4309/5000\n",
+      "919/919 - 3s - loss: 1.2078 - accuracy: 0.6203 - val_loss: 4.1389 - val_accuracy: 0.0877\n",
+      "Epoch 4310/5000\n",
+      "919/919 - 3s - loss: 1.2046 - accuracy: 0.6234 - val_loss: 4.1459 - val_accuracy: 0.0883\n",
+      "Epoch 4311/5000\n",
+      "919/919 - 3s - loss: 1.2796 - accuracy: 0.6237 - val_loss: 4.1377 - val_accuracy: 0.0880\n",
+      "Epoch 4312/5000\n",
+      "919/919 - 3s - loss: 1.2027 - accuracy: 0.6211 - val_loss: 4.1246 - val_accuracy: 0.0883\n",
+      "Epoch 4313/5000\n",
+      "919/919 - 3s - loss: 1.2014 - accuracy: 0.6215 - val_loss: 4.1236 - val_accuracy: 0.0886\n",
+      "Epoch 4314/5000\n",
+      "919/919 - 3s - loss: 1.4562 - accuracy: 0.6232 - val_loss: 4.1176 - val_accuracy: 0.0876\n",
+      "Epoch 4315/5000\n",
+      "919/919 - 3s - loss: 1.2041 - accuracy: 0.6273 - val_loss: 4.1154 - val_accuracy: 0.0887\n",
+      "Epoch 4316/5000\n",
+      "919/919 - 3s - loss: 1.2053 - accuracy: 0.6239 - val_loss: 4.1243 - val_accuracy: 0.0879\n",
+      "Epoch 4317/5000\n",
+      "919/919 - 3s - loss: 1.1983 - accuracy: 0.6229 - val_loss: 4.1299 - val_accuracy: 0.0889\n",
+      "Epoch 4318/5000\n",
+      "919/919 - 3s - loss: 1.2019 - accuracy: 0.6221 - val_loss: 4.1067 - val_accuracy: 0.0887\n",
+      "Epoch 4319/5000\n",
+      "919/919 - 3s - loss: 1.2013 - accuracy: 0.6259 - val_loss: 4.1091 - val_accuracy: 0.0883\n",
+      "Epoch 4320/5000\n",
+      "919/919 - 3s - loss: 1.1975 - accuracy: 0.6261 - val_loss: 4.1037 - val_accuracy: 0.0889\n",
+      "Epoch 4321/5000\n",
+      "919/919 - 3s - loss: 1.2259 - accuracy: 0.6249 - val_loss: 4.1124 - val_accuracy: 0.0890\n",
+      "Epoch 4322/5000\n",
+      "919/919 - 3s - loss: 1.1873 - accuracy: 0.6248 - val_loss: 4.1218 - val_accuracy: 0.0889\n",
+      "Epoch 4323/5000\n",
+      "919/919 - 3s - loss: 1.2093 - accuracy: 0.6224 - val_loss: 4.1181 - val_accuracy: 0.0884\n",
+      "Epoch 4324/5000\n",
+      "919/919 - 3s - loss: 1.2175 - accuracy: 0.6244 - val_loss: 4.1103 - val_accuracy: 0.0884\n",
+      "Epoch 4325/5000\n",
+      "919/919 - 3s - loss: 1.2679 - accuracy: 0.6236 - val_loss: 4.1192 - val_accuracy: 0.0885\n",
+      "Epoch 4326/5000\n",
+      "919/919 - 3s - loss: 1.2073 - accuracy: 0.6250 - val_loss: 4.1322 - val_accuracy: 0.0884\n",
+      "Epoch 4327/5000\n",
+      "919/919 - 3s - loss: 1.1952 - accuracy: 0.6228 - val_loss: 4.1376 - val_accuracy: 0.0886\n",
+      "Epoch 4328/5000\n",
+      "919/919 - 3s - loss: 1.1994 - accuracy: 0.6256 - val_loss: 4.1430 - val_accuracy: 0.0883\n",
+      "Epoch 4329/5000\n",
+      "919/919 - 3s - loss: 1.2056 - accuracy: 0.6269 - val_loss: 4.1483 - val_accuracy: 0.0877\n",
+      "Epoch 4330/5000\n",
+      "919/919 - 3s - loss: 1.2106 - accuracy: 0.6203 - val_loss: 4.1239 - val_accuracy: 0.0876\n",
+      "Epoch 4331/5000\n",
+      "919/919 - 3s - loss: 1.2124 - accuracy: 0.6223 - val_loss: 4.1218 - val_accuracy: 0.0879\n",
+      "Epoch 4332/5000\n",
+      "919/919 - 3s - loss: 1.2026 - accuracy: 0.6209 - val_loss: 4.1121 - val_accuracy: 0.0879\n",
+      "Epoch 4333/5000\n",
+      "919/919 - 3s - loss: 1.2021 - accuracy: 0.6197 - val_loss: 4.1165 - val_accuracy: 0.0882\n",
+      "Epoch 4334/5000\n",
+      "919/919 - 3s - loss: 1.2045 - accuracy: 0.6221 - val_loss: 4.1099 - val_accuracy: 0.0888\n",
+      "Epoch 4335/5000\n",
+      "919/919 - 3s - loss: 1.2135 - accuracy: 0.6247 - val_loss: 4.1274 - val_accuracy: 0.0889\n",
+      "Epoch 4336/5000\n",
+      "919/919 - 3s - loss: 1.2019 - accuracy: 0.6250 - val_loss: 4.1377 - val_accuracy: 0.0881\n",
+      "Epoch 4337/5000\n",
+      "919/919 - 3s - loss: 1.1935 - accuracy: 0.6231 - val_loss: 4.1436 - val_accuracy: 0.0888\n",
+      "Epoch 4338/5000\n",
+      "919/919 - 3s - loss: 1.2089 - accuracy: 0.6241 - val_loss: 4.1240 - val_accuracy: 0.0888\n",
+      "Epoch 4339/5000\n",
+      "919/919 - 3s - loss: 1.2149 - accuracy: 0.6232 - val_loss: 4.1288 - val_accuracy: 0.0883\n",
+      "Epoch 4340/5000\n",
+      "919/919 - 3s - loss: 1.2152 - accuracy: 0.6197 - val_loss: 4.1390 - val_accuracy: 0.0883\n",
+      "Epoch 4341/5000\n",
+      "919/919 - 3s - loss: 1.1968 - accuracy: 0.6241 - val_loss: 4.1354 - val_accuracy: 0.0881\n",
+      "Epoch 4342/5000\n",
+      "919/919 - 3s - loss: 1.2070 - accuracy: 0.6251 - val_loss: 4.1515 - val_accuracy: 0.0876\n",
+      "Epoch 4343/5000\n",
+      "919/919 - 3s - loss: 1.2031 - accuracy: 0.6240 - val_loss: 4.1386 - val_accuracy: 0.0884\n",
+      "Epoch 4344/5000\n",
+      "919/919 - 3s - loss: 1.2119 - accuracy: 0.6238 - val_loss: 4.1480 - val_accuracy: 0.0885\n",
+      "Epoch 4345/5000\n",
+      "919/919 - 3s - loss: 1.2012 - accuracy: 0.6234 - val_loss: 4.1376 - val_accuracy: 0.0882\n",
+      "Epoch 4346/5000\n",
+      "919/919 - 3s - loss: 1.1975 - accuracy: 0.6297 - val_loss: 4.1393 - val_accuracy: 0.0882\n",
+      "Epoch 4347/5000\n",
+      "919/919 - 3s - loss: 1.1976 - accuracy: 0.6237 - val_loss: 4.1426 - val_accuracy: 0.0883\n",
+      "Epoch 4348/5000\n",
+      "919/919 - 3s - loss: 1.2012 - accuracy: 0.6233 - val_loss: 4.1293 - val_accuracy: 0.0883\n",
+      "Epoch 4349/5000\n",
+      "919/919 - 3s - loss: 1.2029 - accuracy: 0.6229 - val_loss: 4.1412 - val_accuracy: 0.0890\n",
+      "Epoch 4350/5000\n",
+      "919/919 - 3s - loss: 1.1892 - accuracy: 0.6278 - val_loss: 4.1512 - val_accuracy: 0.0881\n",
+      "Epoch 4351/5000\n",
+      "919/919 - 3s - loss: 1.2303 - accuracy: 0.6293 - val_loss: 4.1561 - val_accuracy: 0.0885\n",
+      "Epoch 4352/5000\n",
+      "919/919 - 3s - loss: 1.1980 - accuracy: 0.6278 - val_loss: 4.1418 - val_accuracy: 0.0881\n",
+      "Epoch 4353/5000\n",
+      "919/919 - 3s - loss: 1.2185 - accuracy: 0.6214 - val_loss: 4.1331 - val_accuracy: 0.0884\n",
+      "Epoch 4354/5000\n",
+      "919/919 - 3s - loss: 1.2403 - accuracy: 0.6234 - val_loss: 4.1454 - val_accuracy: 0.0883\n",
+      "Epoch 4355/5000\n",
+      "919/919 - 3s - loss: 1.2071 - accuracy: 0.6237 - val_loss: 4.1160 - val_accuracy: 0.0878\n",
+      "Epoch 4356/5000\n",
+      "919/919 - 3s - loss: 1.2219 - accuracy: 0.6231 - val_loss: 4.1084 - val_accuracy: 0.0871\n",
+      "Epoch 4357/5000\n",
+      "919/919 - 3s - loss: 1.1953 - accuracy: 0.6242 - val_loss: 4.1134 - val_accuracy: 0.0874\n",
+      "Epoch 4358/5000\n",
+      "919/919 - 3s - loss: 1.2026 - accuracy: 0.6261 - val_loss: 4.1220 - val_accuracy: 0.0876\n",
+      "Epoch 4359/5000\n",
+      "919/919 - 3s - loss: 1.2100 - accuracy: 0.6230 - val_loss: 4.1278 - val_accuracy: 0.0880\n",
+      "Epoch 4360/5000\n",
+      "919/919 - 3s - loss: 1.1917 - accuracy: 0.6269 - val_loss: 4.1184 - val_accuracy: 0.0881\n",
+      "Epoch 4361/5000\n",
+      "919/919 - 3s - loss: 1.1805 - accuracy: 0.6259 - val_loss: 4.1373 - val_accuracy: 0.0889\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4362/5000\n",
+      "919/919 - 3s - loss: 1.1967 - accuracy: 0.6277 - val_loss: 4.1409 - val_accuracy: 0.0889\n",
+      "Epoch 4363/5000\n",
+      "919/919 - 3s - loss: 1.2078 - accuracy: 0.6241 - val_loss: 4.1381 - val_accuracy: 0.0884\n",
+      "Epoch 4364/5000\n",
+      "919/919 - 3s - loss: 1.2120 - accuracy: 0.6218 - val_loss: 4.1402 - val_accuracy: 0.0881\n",
+      "Epoch 4365/5000\n",
+      "919/919 - 3s - loss: 1.2251 - accuracy: 0.6232 - val_loss: 4.1338 - val_accuracy: 0.0887\n",
+      "Epoch 4366/5000\n",
+      "919/919 - 3s - loss: 1.1904 - accuracy: 0.6244 - val_loss: 4.1530 - val_accuracy: 0.0894\n",
+      "Epoch 4367/5000\n",
+      "919/919 - 3s - loss: 1.1956 - accuracy: 0.6268 - val_loss: 4.1411 - val_accuracy: 0.0885\n",
+      "Epoch 4368/5000\n",
+      "919/919 - 3s - loss: 1.3050 - accuracy: 0.6246 - val_loss: 4.1244 - val_accuracy: 0.0880\n",
+      "Epoch 4369/5000\n",
+      "919/919 - 3s - loss: 1.2028 - accuracy: 0.6251 - val_loss: 4.1355 - val_accuracy: 0.0882\n",
+      "Epoch 4370/5000\n",
+      "919/919 - 3s - loss: 1.1865 - accuracy: 0.6263 - val_loss: 4.1346 - val_accuracy: 0.0881\n",
+      "Epoch 4371/5000\n",
+      "919/919 - 3s - loss: 1.1936 - accuracy: 0.6266 - val_loss: 4.1338 - val_accuracy: 0.0887\n",
+      "Epoch 4372/5000\n",
+      "919/919 - 3s - loss: 1.2033 - accuracy: 0.6215 - val_loss: 4.1393 - val_accuracy: 0.0885\n",
+      "Epoch 4373/5000\n",
+      "919/919 - 3s - loss: 1.2110 - accuracy: 0.6257 - val_loss: 4.1510 - val_accuracy: 0.0887\n",
+      "Epoch 4374/5000\n",
+      "919/919 - 3s - loss: 1.2115 - accuracy: 0.6206 - val_loss: 4.1309 - val_accuracy: 0.0887\n",
+      "Epoch 4375/5000\n",
+      "919/919 - 3s - loss: 1.1865 - accuracy: 0.6304 - val_loss: 4.1463 - val_accuracy: 0.0889\n",
+      "Epoch 4376/5000\n",
+      "919/919 - 3s - loss: 1.2032 - accuracy: 0.6247 - val_loss: 4.1403 - val_accuracy: 0.0883\n",
+      "Epoch 4377/5000\n",
+      "919/919 - 3s - loss: 1.2377 - accuracy: 0.6266 - val_loss: 4.1488 - val_accuracy: 0.0882\n",
+      "Epoch 4378/5000\n",
+      "919/919 - 3s - loss: 1.1775 - accuracy: 0.6254 - val_loss: 4.1363 - val_accuracy: 0.0882\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4379/5000\n",
+      "919/919 - 3s - loss: 1.2382 - accuracy: 0.6268 - val_loss: 4.1333 - val_accuracy: 0.0886\n",
+      "Epoch 4380/5000\n",
+      "919/919 - 3s - loss: 1.1863 - accuracy: 0.6280 - val_loss: 4.1281 - val_accuracy: 0.0886\n",
+      "Epoch 4381/5000\n",
+      "919/919 - 3s - loss: 1.1887 - accuracy: 0.6267 - val_loss: 4.1295 - val_accuracy: 0.0880\n",
+      "Epoch 4382/5000\n",
+      "919/919 - 3s - loss: 1.1956 - accuracy: 0.6253 - val_loss: 4.1433 - val_accuracy: 0.0878\n",
+      "Epoch 4383/5000\n",
+      "919/919 - 3s - loss: 1.3545 - accuracy: 0.6281 - val_loss: 4.1385 - val_accuracy: 0.0881\n",
+      "Epoch 4384/5000\n",
+      "919/919 - 3s - loss: 1.1940 - accuracy: 0.6272 - val_loss: 4.1334 - val_accuracy: 0.0882\n",
+      "Epoch 4385/5000\n",
+      "919/919 - 3s - loss: 1.1967 - accuracy: 0.6257 - val_loss: 4.1319 - val_accuracy: 0.0880\n",
+      "Epoch 4386/5000\n",
+      "919/919 - 3s - loss: 1.3621 - accuracy: 0.6227 - val_loss: 4.1460 - val_accuracy: 0.0881\n",
+      "Epoch 4387/5000\n",
+      "919/919 - 3s - loss: 1.1923 - accuracy: 0.6259 - val_loss: 4.1361 - val_accuracy: 0.0876\n",
+      "Epoch 4388/5000\n",
+      "919/919 - 3s - loss: 1.1998 - accuracy: 0.6228 - val_loss: 4.1333 - val_accuracy: 0.0875\n",
+      "Epoch 4389/5000\n",
+      "919/919 - 3s - loss: 1.2152 - accuracy: 0.6238 - val_loss: 4.1302 - val_accuracy: 0.0880\n",
+      "Epoch 4390/5000\n",
+      "919/919 - 3s - loss: 1.1918 - accuracy: 0.6293 - val_loss: 4.1442 - val_accuracy: 0.0880\n",
+      "Epoch 4391/5000\n",
+      "919/919 - 3s - loss: 1.1903 - accuracy: 0.6245 - val_loss: 4.1435 - val_accuracy: 0.0882\n",
+      "Epoch 4392/5000\n",
+      "919/919 - 3s - loss: 1.2163 - accuracy: 0.6242 - val_loss: 4.1464 - val_accuracy: 0.0878\n",
+      "Epoch 4393/5000\n",
+      "919/919 - 3s - loss: 1.2039 - accuracy: 0.6224 - val_loss: 4.1530 - val_accuracy: 0.0874\n",
+      "Epoch 4394/5000\n",
+      "919/919 - 3s - loss: 1.2170 - accuracy: 0.6233 - val_loss: 4.1446 - val_accuracy: 0.0880\n",
+      "Epoch 4395/5000\n",
+      "919/919 - 3s - loss: 1.1949 - accuracy: 0.6244 - val_loss: 4.1478 - val_accuracy: 0.0878\n",
+      "Epoch 4396/5000\n",
+      "919/919 - 3s - loss: 1.2023 - accuracy: 0.6216 - val_loss: 4.1429 - val_accuracy: 0.0879\n",
+      "Epoch 4397/5000\n",
+      "919/919 - 3s - loss: 1.2032 - accuracy: 0.6231 - val_loss: 4.1426 - val_accuracy: 0.0873\n",
+      "Epoch 4398/5000\n",
+      "919/919 - 3s - loss: 1.2087 - accuracy: 0.6208 - val_loss: 4.1325 - val_accuracy: 0.0875\n",
+      "Epoch 4399/5000\n",
+      "919/919 - 3s - loss: 1.1972 - accuracy: 0.6275 - val_loss: 4.1408 - val_accuracy: 0.0874\n",
+      "Epoch 4400/5000\n",
+      "919/919 - 3s - loss: 1.2754 - accuracy: 0.6252 - val_loss: 4.1379 - val_accuracy: 0.0880\n",
+      "Epoch 4401/5000\n",
+      "919/919 - 3s - loss: 1.1953 - accuracy: 0.6294 - val_loss: 4.1325 - val_accuracy: 0.0883\n",
+      "Epoch 4402/5000\n",
+      "919/919 - 3s - loss: 1.1958 - accuracy: 0.6256 - val_loss: 4.1361 - val_accuracy: 0.0890\n",
+      "Epoch 4403/5000\n",
+      "919/919 - 3s - loss: 1.1975 - accuracy: 0.6291 - val_loss: 4.1415 - val_accuracy: 0.0886\n",
+      "Epoch 4404/5000\n",
+      "919/919 - 3s - loss: 1.1802 - accuracy: 0.6294 - val_loss: 4.1610 - val_accuracy: 0.0887\n",
+      "Epoch 4405/5000\n",
+      "919/919 - 3s - loss: 1.1818 - accuracy: 0.6310 - val_loss: 4.1507 - val_accuracy: 0.0884\n",
+      "Epoch 4406/5000\n",
+      "919/919 - 3s - loss: 1.1956 - accuracy: 0.6276 - val_loss: 4.1517 - val_accuracy: 0.0880\n",
+      "Epoch 4407/5000\n",
+      "919/919 - 3s - loss: 1.2020 - accuracy: 0.6273 - val_loss: 4.1456 - val_accuracy: 0.0876\n",
+      "Epoch 4408/5000\n",
+      "919/919 - 3s - loss: 1.1988 - accuracy: 0.6245 - val_loss: 4.1588 - val_accuracy: 0.0882\n",
+      "Epoch 4409/5000\n",
+      "919/919 - 3s - loss: 1.2330 - accuracy: 0.6231 - val_loss: 4.1522 - val_accuracy: 0.0877\n",
+      "Epoch 4410/5000\n",
+      "919/919 - 3s - loss: 1.1880 - accuracy: 0.6288 - val_loss: 4.1550 - val_accuracy: 0.0881\n",
+      "Epoch 4411/5000\n",
+      "919/919 - 3s - loss: 1.1946 - accuracy: 0.6241 - val_loss: 4.1439 - val_accuracy: 0.0886\n",
+      "Epoch 4412/5000\n",
+      "919/919 - 3s - loss: 1.2641 - accuracy: 0.6229 - val_loss: 4.1441 - val_accuracy: 0.0885\n",
+      "Epoch 4413/5000\n",
+      "919/919 - 3s - loss: 1.2145 - accuracy: 0.6266 - val_loss: 4.1383 - val_accuracy: 0.0890\n",
+      "Epoch 4414/5000\n",
+      "919/919 - 3s - loss: 1.2063 - accuracy: 0.6269 - val_loss: 4.1327 - val_accuracy: 0.0889\n",
+      "Epoch 4415/5000\n",
+      "919/919 - 3s - loss: 1.1988 - accuracy: 0.6269 - val_loss: 4.1273 - val_accuracy: 0.0883\n",
+      "Epoch 4416/5000\n",
+      "919/919 - 3s - loss: 1.2088 - accuracy: 0.6229 - val_loss: 4.1310 - val_accuracy: 0.0888\n",
+      "Epoch 4417/5000\n",
+      "919/919 - 3s - loss: 1.2364 - accuracy: 0.6269 - val_loss: 4.1170 - val_accuracy: 0.0883\n",
+      "Epoch 4418/5000\n",
+      "919/919 - 3s - loss: 1.1925 - accuracy: 0.6243 - val_loss: 4.1282 - val_accuracy: 0.0883\n",
+      "Epoch 4419/5000\n",
+      "919/919 - 3s - loss: 1.2206 - accuracy: 0.6245 - val_loss: 4.1260 - val_accuracy: 0.0879\n",
+      "Epoch 4420/5000\n",
+      "919/919 - 3s - loss: 1.2018 - accuracy: 0.6288 - val_loss: 4.1269 - val_accuracy: 0.0875\n",
+      "Epoch 4421/5000\n",
+      "919/919 - 3s - loss: 1.2154 - accuracy: 0.6240 - val_loss: 4.1373 - val_accuracy: 0.0885\n",
+      "Epoch 4422/5000\n",
+      "919/919 - 3s - loss: 1.2120 - accuracy: 0.6274 - val_loss: 4.1364 - val_accuracy: 0.0883\n",
+      "Epoch 4423/5000\n",
+      "919/919 - 3s - loss: 1.1995 - accuracy: 0.6263 - val_loss: 4.1422 - val_accuracy: 0.0883\n",
+      "Epoch 4424/5000\n",
+      "919/919 - 3s - loss: 1.1946 - accuracy: 0.6305 - val_loss: 4.1437 - val_accuracy: 0.0883\n",
+      "Epoch 4425/5000\n",
+      "919/919 - 3s - loss: 1.1960 - accuracy: 0.6263 - val_loss: 4.1478 - val_accuracy: 0.0886\n",
+      "Epoch 4426/5000\n",
+      "919/919 - 3s - loss: 1.1931 - accuracy: 0.6251 - val_loss: 4.1405 - val_accuracy: 0.0885\n",
+      "Epoch 4427/5000\n",
+      "919/919 - 3s - loss: 1.1888 - accuracy: 0.6278 - val_loss: 4.1534 - val_accuracy: 0.0891\n",
+      "Epoch 4428/5000\n",
+      "919/919 - 3s - loss: 1.1979 - accuracy: 0.6217 - val_loss: 4.1443 - val_accuracy: 0.0891\n",
+      "Epoch 4429/5000\n",
+      "919/919 - 3s - loss: 1.1936 - accuracy: 0.6253 - val_loss: 4.1581 - val_accuracy: 0.0886\n",
+      "Epoch 4430/5000\n",
+      "919/919 - 3s - loss: 1.1974 - accuracy: 0.6280 - val_loss: 4.1584 - val_accuracy: 0.0875\n",
+      "Epoch 4431/5000\n",
+      "919/919 - 3s - loss: 1.2216 - accuracy: 0.6265 - val_loss: 4.1607 - val_accuracy: 0.0874\n",
+      "Epoch 4432/5000\n",
+      "919/919 - 3s - loss: 1.5076 - accuracy: 0.6250 - val_loss: 4.1550 - val_accuracy: 0.0876\n",
+      "Epoch 4433/5000\n",
+      "919/919 - 3s - loss: 1.1924 - accuracy: 0.6261 - val_loss: 4.1800 - val_accuracy: 0.0876\n",
+      "Epoch 4434/5000\n",
+      "919/919 - 3s - loss: 1.1983 - accuracy: 0.6250 - val_loss: 4.1600 - val_accuracy: 0.0870\n",
+      "Epoch 4435/5000\n",
+      "919/919 - 3s - loss: 1.2161 - accuracy: 0.6254 - val_loss: 4.1405 - val_accuracy: 0.0877\n",
+      "Epoch 4436/5000\n",
+      "919/919 - 3s - loss: 1.2083 - accuracy: 0.6231 - val_loss: 4.1437 - val_accuracy: 0.0876\n",
+      "Epoch 4437/5000\n",
+      "919/919 - 3s - loss: 1.1985 - accuracy: 0.6253 - val_loss: 4.1364 - val_accuracy: 0.0883\n",
+      "Epoch 4438/5000\n",
+      "919/919 - 3s - loss: 1.2006 - accuracy: 0.6293 - val_loss: 4.1421 - val_accuracy: 0.0877\n",
+      "Epoch 4439/5000\n",
+      "919/919 - 3s - loss: 1.1880 - accuracy: 0.6287 - val_loss: 4.1461 - val_accuracy: 0.0880\n",
+      "Epoch 4440/5000\n",
+      "919/919 - 3s - loss: 1.2097 - accuracy: 0.6287 - val_loss: 4.1375 - val_accuracy: 0.0884\n",
+      "Epoch 4441/5000\n",
+      "919/919 - 3s - loss: 1.2118 - accuracy: 0.6261 - val_loss: 4.1234 - val_accuracy: 0.0885\n",
+      "Epoch 4442/5000\n",
+      "919/919 - 3s - loss: 1.1867 - accuracy: 0.6267 - val_loss: 4.1330 - val_accuracy: 0.0889\n",
+      "Epoch 4443/5000\n",
+      "919/919 - 3s - loss: 1.2011 - accuracy: 0.6286 - val_loss: 4.1312 - val_accuracy: 0.0889\n",
+      "Epoch 4444/5000\n",
+      "919/919 - 3s - loss: 1.2369 - accuracy: 0.6265 - val_loss: 4.1263 - val_accuracy: 0.0889\n",
+      "Epoch 4445/5000\n",
+      "919/919 - 3s - loss: 1.2135 - accuracy: 0.6235 - val_loss: 4.1357 - val_accuracy: 0.0887\n",
+      "Epoch 4446/5000\n",
+      "919/919 - 3s - loss: 1.2671 - accuracy: 0.6238 - val_loss: 4.1404 - val_accuracy: 0.0887\n",
+      "Epoch 4447/5000\n",
+      "919/919 - 3s - loss: 1.1949 - accuracy: 0.6279 - val_loss: 4.1348 - val_accuracy: 0.0884\n",
+      "Epoch 4448/5000\n",
+      "919/919 - 3s - loss: 1.1991 - accuracy: 0.6299 - val_loss: 4.1277 - val_accuracy: 0.0876\n",
+      "Epoch 4449/5000\n",
+      "919/919 - 3s - loss: 1.2029 - accuracy: 0.6267 - val_loss: 4.1447 - val_accuracy: 0.0882\n",
+      "Epoch 4450/5000\n",
+      "919/919 - 3s - loss: 1.2126 - accuracy: 0.6273 - val_loss: 4.1415 - val_accuracy: 0.0885\n",
+      "Epoch 4451/5000\n",
+      "919/919 - 3s - loss: 1.1913 - accuracy: 0.6308 - val_loss: 4.1425 - val_accuracy: 0.0886\n",
+      "Epoch 4452/5000\n",
+      "919/919 - 3s - loss: 1.2653 - accuracy: 0.6252 - val_loss: 4.1501 - val_accuracy: 0.0883\n",
+      "Epoch 4453/5000\n",
+      "919/919 - 3s - loss: 1.2049 - accuracy: 0.6233 - val_loss: 4.1556 - val_accuracy: 0.0880\n",
+      "Epoch 4454/5000\n",
+      "919/919 - 3s - loss: 1.1943 - accuracy: 0.6269 - val_loss: 4.1351 - val_accuracy: 0.0880\n",
+      "Epoch 4455/5000\n",
+      "919/919 - 3s - loss: 1.2267 - accuracy: 0.6243 - val_loss: 4.1572 - val_accuracy: 0.0889\n",
+      "Epoch 4456/5000\n",
+      "919/919 - 3s - loss: 1.1993 - accuracy: 0.6258 - val_loss: 4.1546 - val_accuracy: 0.0883\n",
+      "Epoch 4457/5000\n",
+      "919/919 - 3s - loss: 1.2083 - accuracy: 0.6269 - val_loss: 4.1557 - val_accuracy: 0.0883\n",
+      "Epoch 4458/5000\n",
+      "919/919 - 3s - loss: 1.1855 - accuracy: 0.6257 - val_loss: 4.1530 - val_accuracy: 0.0881\n",
+      "Epoch 4459/5000\n",
+      "919/919 - 3s - loss: 1.2900 - accuracy: 0.6272 - val_loss: 4.1486 - val_accuracy: 0.0877\n",
+      "Epoch 4460/5000\n",
+      "919/919 - 3s - loss: 1.1924 - accuracy: 0.6263 - val_loss: 4.1367 - val_accuracy: 0.0874\n",
+      "Epoch 4461/5000\n",
+      "919/919 - 3s - loss: 1.2447 - accuracy: 0.6269 - val_loss: 4.1276 - val_accuracy: 0.0887\n",
+      "Epoch 4462/5000\n",
+      "919/919 - 3s - loss: 1.1996 - accuracy: 0.6296 - val_loss: 4.1303 - val_accuracy: 0.0892\n",
+      "Epoch 4463/5000\n",
+      "919/919 - 3s - loss: 1.1964 - accuracy: 0.6241 - val_loss: 4.1339 - val_accuracy: 0.0889\n",
+      "Epoch 4464/5000\n",
+      "919/919 - 3s - loss: 1.1788 - accuracy: 0.6250 - val_loss: 4.1525 - val_accuracy: 0.0886\n",
+      "Epoch 4465/5000\n",
+      "919/919 - 3s - loss: 1.1897 - accuracy: 0.6284 - val_loss: 4.1560 - val_accuracy: 0.0891\n",
+      "Epoch 4466/5000\n",
+      "919/919 - 3s - loss: 1.2531 - accuracy: 0.6283 - val_loss: 4.1387 - val_accuracy: 0.0883\n",
+      "Epoch 4467/5000\n",
+      "919/919 - 3s - loss: 1.1904 - accuracy: 0.6293 - val_loss: 4.1381 - val_accuracy: 0.0884\n",
+      "Epoch 4468/5000\n",
+      "919/919 - 3s - loss: 1.1907 - accuracy: 0.6279 - val_loss: 4.1435 - val_accuracy: 0.0882\n",
+      "Epoch 4469/5000\n",
+      "919/919 - 3s - loss: 1.1883 - accuracy: 0.6303 - val_loss: 4.1427 - val_accuracy: 0.0884\n",
+      "Epoch 4470/5000\n",
+      "919/919 - 3s - loss: 1.2064 - accuracy: 0.6245 - val_loss: 4.1322 - val_accuracy: 0.0883\n",
+      "Epoch 4471/5000\n",
+      "919/919 - 3s - loss: 1.1830 - accuracy: 0.6302 - val_loss: 4.1482 - val_accuracy: 0.0877\n",
+      "Epoch 4472/5000\n",
+      "919/919 - 3s - loss: 1.1821 - accuracy: 0.6297 - val_loss: 4.1452 - val_accuracy: 0.0876\n",
+      "Epoch 4473/5000\n",
+      "919/919 - 3s - loss: 1.1875 - accuracy: 0.6293 - val_loss: 4.1581 - val_accuracy: 0.0876\n",
+      "Epoch 4474/5000\n",
+      "919/919 - 3s - loss: 1.1954 - accuracy: 0.6299 - val_loss: 4.1545 - val_accuracy: 0.0875\n",
+      "Epoch 4475/5000\n",
+      "919/919 - 3s - loss: 1.1967 - accuracy: 0.6276 - val_loss: 4.1536 - val_accuracy: 0.0875\n",
+      "Epoch 4476/5000\n",
+      "919/919 - 3s - loss: 1.1934 - accuracy: 0.6284 - val_loss: 4.1591 - val_accuracy: 0.0881\n",
+      "Epoch 4477/5000\n",
+      "919/919 - 3s - loss: 1.2206 - accuracy: 0.6303 - val_loss: 4.1514 - val_accuracy: 0.0884\n",
+      "Epoch 4478/5000\n",
+      "919/919 - 3s - loss: 1.2101 - accuracy: 0.6272 - val_loss: 4.1618 - val_accuracy: 0.0886\n",
+      "Epoch 4479/5000\n",
+      "919/919 - 3s - loss: 1.1845 - accuracy: 0.6288 - val_loss: 4.1461 - val_accuracy: 0.0890\n",
+      "Epoch 4480/5000\n",
+      "919/919 - 3s - loss: 1.1905 - accuracy: 0.6271 - val_loss: 4.1520 - val_accuracy: 0.0883\n",
+      "Epoch 4481/5000\n",
+      "919/919 - 3s - loss: 1.1779 - accuracy: 0.6316 - val_loss: 4.1510 - val_accuracy: 0.0888\n",
+      "Epoch 4482/5000\n",
+      "919/919 - 3s - loss: 1.1877 - accuracy: 0.6250 - val_loss: 4.1463 - val_accuracy: 0.0886\n",
+      "Epoch 4483/5000\n",
+      "919/919 - 3s - loss: 1.2003 - accuracy: 0.6271 - val_loss: 4.1450 - val_accuracy: 0.0890\n",
+      "Epoch 4484/5000\n",
+      "919/919 - 3s - loss: 1.1901 - accuracy: 0.6284 - val_loss: 4.1446 - val_accuracy: 0.0882\n",
+      "Epoch 4485/5000\n",
+      "919/919 - 3s - loss: 1.3274 - accuracy: 0.6252 - val_loss: 4.1621 - val_accuracy: 0.0893\n",
+      "Epoch 4486/5000\n",
+      "919/919 - 3s - loss: 1.1890 - accuracy: 0.6293 - val_loss: 4.1577 - val_accuracy: 0.0885\n",
+      "Epoch 4487/5000\n",
+      "919/919 - 3s - loss: 1.2021 - accuracy: 0.6286 - val_loss: 4.1493 - val_accuracy: 0.0888\n",
+      "Epoch 4488/5000\n",
+      "919/919 - 3s - loss: 1.2040 - accuracy: 0.6251 - val_loss: 4.1469 - val_accuracy: 0.0879\n",
+      "Epoch 4489/5000\n",
+      "919/919 - 3s - loss: 1.1888 - accuracy: 0.6265 - val_loss: 4.1476 - val_accuracy: 0.0879\n",
+      "Epoch 4490/5000\n",
+      "919/919 - 3s - loss: 1.1727 - accuracy: 0.6300 - val_loss: 4.1431 - val_accuracy: 0.0881\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4491/5000\n",
+      "919/919 - 3s - loss: 1.1861 - accuracy: 0.6286 - val_loss: 4.1651 - val_accuracy: 0.0886\n",
+      "Epoch 4492/5000\n",
+      "919/919 - 3s - loss: 1.2069 - accuracy: 0.6249 - val_loss: 4.1563 - val_accuracy: 0.0882\n",
+      "Epoch 4493/5000\n",
+      "919/919 - 3s - loss: 1.1913 - accuracy: 0.6278 - val_loss: 4.1513 - val_accuracy: 0.0884\n",
+      "Epoch 4494/5000\n",
+      "919/919 - 3s - loss: 1.1846 - accuracy: 0.6285 - val_loss: 4.1562 - val_accuracy: 0.0879\n",
+      "Epoch 4495/5000\n",
+      "919/919 - 3s - loss: 1.1792 - accuracy: 0.6316 - val_loss: 4.1638 - val_accuracy: 0.0880\n",
+      "Epoch 4496/5000\n",
+      "919/919 - 3s - loss: 1.1823 - accuracy: 0.6277 - val_loss: 4.1755 - val_accuracy: 0.0876\n",
+      "Epoch 4497/5000\n",
+      "919/919 - 3s - loss: 1.2913 - accuracy: 0.6237 - val_loss: 4.1741 - val_accuracy: 0.0880\n",
+      "Epoch 4498/5000\n",
+      "919/919 - 3s - loss: 1.1976 - accuracy: 0.6307 - val_loss: 4.1715 - val_accuracy: 0.0872\n",
+      "Epoch 4499/5000\n",
+      "919/919 - 3s - loss: 1.2266 - accuracy: 0.6278 - val_loss: 4.1644 - val_accuracy: 0.0884\n",
+      "Epoch 4500/5000\n",
+      "919/919 - 3s - loss: 1.2604 - accuracy: 0.6281 - val_loss: 4.1713 - val_accuracy: 0.0884\n",
+      "Epoch 4501/5000\n",
+      "919/919 - 3s - loss: 1.1964 - accuracy: 0.6273 - val_loss: 4.1815 - val_accuracy: 0.0883\n",
+      "Epoch 4502/5000\n",
+      "919/919 - 3s - loss: 1.1990 - accuracy: 0.6261 - val_loss: 4.1625 - val_accuracy: 0.0884\n",
+      "Epoch 4503/5000\n",
+      "919/919 - 3s - loss: 1.2880 - accuracy: 0.6318 - val_loss: 4.1777 - val_accuracy: 0.0885\n",
+      "Epoch 4504/5000\n",
+      "919/919 - 3s - loss: 1.2576 - accuracy: 0.6233 - val_loss: 4.1721 - val_accuracy: 0.0892\n",
+      "Epoch 4505/5000\n",
+      "919/919 - 3s - loss: 1.2059 - accuracy: 0.6233 - val_loss: 4.1667 - val_accuracy: 0.0889\n",
+      "Epoch 4506/5000\n",
+      "919/919 - 3s - loss: 1.1984 - accuracy: 0.6271 - val_loss: 4.1620 - val_accuracy: 0.0883\n",
+      "Epoch 4507/5000\n",
+      "919/919 - 3s - loss: 1.1865 - accuracy: 0.6324 - val_loss: 4.1593 - val_accuracy: 0.0883\n",
+      "Epoch 4508/5000\n",
+      "919/919 - 3s - loss: 1.1936 - accuracy: 0.6305 - val_loss: 4.1518 - val_accuracy: 0.0883\n",
+      "Epoch 4509/5000\n",
+      "919/919 - 3s - loss: 1.1938 - accuracy: 0.6297 - val_loss: 4.1543 - val_accuracy: 0.0882\n",
+      "Epoch 4510/5000\n",
+      "919/919 - 3s - loss: 1.1913 - accuracy: 0.6276 - val_loss: 4.1512 - val_accuracy: 0.0884\n",
+      "Epoch 4511/5000\n",
+      "919/919 - 3s - loss: 1.1856 - accuracy: 0.6252 - val_loss: 4.1498 - val_accuracy: 0.0875\n",
+      "Epoch 4512/5000\n",
+      "919/919 - 3s - loss: 1.1924 - accuracy: 0.6296 - val_loss: 4.1533 - val_accuracy: 0.0882\n",
+      "Epoch 4513/5000\n",
+      "919/919 - 3s - loss: 1.1812 - accuracy: 0.6286 - val_loss: 4.1602 - val_accuracy: 0.0890\n",
+      "Epoch 4514/5000\n",
+      "919/919 - 3s - loss: 1.1906 - accuracy: 0.6281 - val_loss: 4.1586 - val_accuracy: 0.0883\n",
+      "Epoch 4515/5000\n",
+      "919/919 - 3s - loss: 1.1819 - accuracy: 0.6284 - val_loss: 4.1517 - val_accuracy: 0.0884\n",
+      "Epoch 4516/5000\n",
+      "919/919 - 3s - loss: 1.2032 - accuracy: 0.6237 - val_loss: 4.1399 - val_accuracy: 0.0889\n",
+      "Epoch 4517/5000\n",
+      "919/919 - 3s - loss: 1.3222 - accuracy: 0.6254 - val_loss: 4.1513 - val_accuracy: 0.0892\n",
+      "Epoch 4518/5000\n",
+      "919/919 - 3s - loss: 1.1949 - accuracy: 0.6273 - val_loss: 4.1547 - val_accuracy: 0.0891\n",
+      "Epoch 4519/5000\n",
+      "919/919 - 3s - loss: 1.1808 - accuracy: 0.6318 - val_loss: 4.1549 - val_accuracy: 0.0886\n",
+      "Epoch 4520/5000\n",
+      "919/919 - 3s - loss: 1.2185 - accuracy: 0.6275 - val_loss: 4.1585 - val_accuracy: 0.0882\n",
+      "Epoch 4521/5000\n",
+      "919/919 - 3s - loss: 1.1891 - accuracy: 0.6304 - val_loss: 4.1748 - val_accuracy: 0.0882\n",
+      "Epoch 4522/5000\n",
+      "919/919 - 3s - loss: 1.1965 - accuracy: 0.6301 - val_loss: 4.1724 - val_accuracy: 0.0876\n",
+      "Epoch 4523/5000\n",
+      "919/919 - 3s - loss: 1.1777 - accuracy: 0.6307 - val_loss: 4.1870 - val_accuracy: 0.0876\n",
+      "Epoch 4524/5000\n",
+      "919/919 - 3s - loss: 1.1738 - accuracy: 0.6323 - val_loss: 4.1695 - val_accuracy: 0.0874\n",
+      "Epoch 4525/5000\n",
+      "919/919 - 3s - loss: 1.1957 - accuracy: 0.6278 - val_loss: 4.1659 - val_accuracy: 0.0870\n",
+      "Epoch 4526/5000\n",
+      "919/919 - 3s - loss: 1.2289 - accuracy: 0.6297 - val_loss: 4.1573 - val_accuracy: 0.0870\n",
+      "Epoch 4527/5000\n",
+      "919/919 - 3s - loss: 1.1970 - accuracy: 0.6287 - val_loss: 4.1584 - val_accuracy: 0.0871\n",
+      "Epoch 4528/5000\n",
+      "919/919 - 3s - loss: 1.1913 - accuracy: 0.6278 - val_loss: 4.1483 - val_accuracy: 0.0877\n",
+      "Epoch 4529/5000\n",
+      "919/919 - 3s - loss: 1.1961 - accuracy: 0.6292 - val_loss: 4.1621 - val_accuracy: 0.0880\n",
+      "Epoch 4530/5000\n",
+      "919/919 - 3s - loss: 1.1891 - accuracy: 0.6267 - val_loss: 4.1701 - val_accuracy: 0.0880\n",
+      "Epoch 4531/5000\n",
+      "919/919 - 3s - loss: 1.1930 - accuracy: 0.6292 - val_loss: 4.1696 - val_accuracy: 0.0883\n",
+      "Epoch 4532/5000\n",
+      "919/919 - 3s - loss: 1.1934 - accuracy: 0.6294 - val_loss: 4.1676 - val_accuracy: 0.0882\n",
+      "Epoch 4533/5000\n",
+      "919/919 - 3s - loss: 1.1893 - accuracy: 0.6290 - val_loss: 4.1544 - val_accuracy: 0.0887\n",
+      "Epoch 4534/5000\n",
+      "919/919 - 3s - loss: 1.1999 - accuracy: 0.6337 - val_loss: 4.1518 - val_accuracy: 0.0887\n",
+      "Epoch 4535/5000\n",
+      "919/919 - 3s - loss: 1.2052 - accuracy: 0.6254 - val_loss: 4.1478 - val_accuracy: 0.0884\n",
+      "Epoch 4536/5000\n",
+      "919/919 - 3s - loss: 1.1858 - accuracy: 0.6307 - val_loss: 4.1803 - val_accuracy: 0.0882\n",
+      "Epoch 4537/5000\n",
+      "919/919 - 3s - loss: 1.1759 - accuracy: 0.6333 - val_loss: 4.1950 - val_accuracy: 0.0881\n",
+      "Epoch 4538/5000\n",
+      "919/919 - 3s - loss: 1.1998 - accuracy: 0.6301 - val_loss: 4.1885 - val_accuracy: 0.0892\n",
+      "Epoch 4539/5000\n",
+      "919/919 - 3s - loss: 1.1860 - accuracy: 0.6267 - val_loss: 4.1981 - val_accuracy: 0.0889\n",
+      "Epoch 4540/5000\n",
+      "919/919 - 3s - loss: 1.1914 - accuracy: 0.6282 - val_loss: 4.1764 - val_accuracy: 0.0886\n",
+      "Epoch 4541/5000\n",
+      "919/919 - 3s - loss: 1.2285 - accuracy: 0.6263 - val_loss: 4.1810 - val_accuracy: 0.0888\n",
+      "Epoch 4542/5000\n",
+      "919/919 - 3s - loss: 1.1924 - accuracy: 0.6268 - val_loss: 4.1704 - val_accuracy: 0.0880\n",
+      "Epoch 4543/5000\n",
+      "919/919 - 3s - loss: 1.1856 - accuracy: 0.6283 - val_loss: 4.1665 - val_accuracy: 0.0889\n",
+      "Epoch 4544/5000\n",
+      "919/919 - 3s - loss: 1.1974 - accuracy: 0.6288 - val_loss: 4.1602 - val_accuracy: 0.0883\n",
+      "Epoch 4545/5000\n",
+      "919/919 - 3s - loss: 1.1948 - accuracy: 0.6252 - val_loss: 4.1589 - val_accuracy: 0.0882\n",
+      "Epoch 4546/5000\n",
+      "919/919 - 3s - loss: 1.2017 - accuracy: 0.6259 - val_loss: 4.1539 - val_accuracy: 0.0879\n",
+      "Epoch 4547/5000\n",
+      "919/919 - 3s - loss: 1.2032 - accuracy: 0.6287 - val_loss: 4.1588 - val_accuracy: 0.0883\n",
+      "Epoch 4548/5000\n",
+      "919/919 - 3s - loss: 1.1902 - accuracy: 0.6269 - val_loss: 4.1832 - val_accuracy: 0.0880\n",
+      "Epoch 4549/5000\n",
+      "919/919 - 3s - loss: 1.1807 - accuracy: 0.6314 - val_loss: 4.1903 - val_accuracy: 0.0883\n",
+      "Epoch 4550/5000\n",
+      "919/919 - 3s - loss: 1.2241 - accuracy: 0.6289 - val_loss: 4.1659 - val_accuracy: 0.0888\n",
+      "Epoch 4551/5000\n",
+      "919/919 - 3s - loss: 1.1941 - accuracy: 0.6284 - val_loss: 4.1634 - val_accuracy: 0.0887\n",
+      "Epoch 4552/5000\n",
+      "919/919 - 3s - loss: 1.1804 - accuracy: 0.6303 - val_loss: 4.1780 - val_accuracy: 0.0893\n",
+      "Epoch 4553/5000\n",
+      "919/919 - 3s - loss: 1.3501 - accuracy: 0.6339 - val_loss: 4.1895 - val_accuracy: 0.0892\n",
+      "Epoch 4554/5000\n",
+      "919/919 - 3s - loss: 1.1790 - accuracy: 0.6301 - val_loss: 4.1806 - val_accuracy: 0.0893\n",
+      "Epoch 4555/5000\n",
+      "919/919 - 3s - loss: 1.1937 - accuracy: 0.6282 - val_loss: 4.1630 - val_accuracy: 0.0899\n",
+      "Epoch 4556/5000\n",
+      "919/919 - 3s - loss: 1.2326 - accuracy: 0.6303 - val_loss: 4.1516 - val_accuracy: 0.0892\n",
+      "Epoch 4557/5000\n",
+      "919/919 - 3s - loss: 1.1800 - accuracy: 0.6282 - val_loss: 4.1583 - val_accuracy: 0.0901\n",
+      "Epoch 4558/5000\n",
+      "919/919 - 3s - loss: 1.1798 - accuracy: 0.6282 - val_loss: 4.1640 - val_accuracy: 0.0896\n",
+      "Epoch 4559/5000\n",
+      "919/919 - 3s - loss: 1.2145 - accuracy: 0.6319 - val_loss: 4.1746 - val_accuracy: 0.0892\n",
+      "Epoch 4560/5000\n",
+      "919/919 - 3s - loss: 1.2111 - accuracy: 0.6266 - val_loss: 4.1718 - val_accuracy: 0.0884\n",
+      "Epoch 4561/5000\n",
+      "919/919 - 3s - loss: 1.1933 - accuracy: 0.6316 - val_loss: 4.1872 - val_accuracy: 0.0884\n",
+      "Epoch 4562/5000\n",
+      "919/919 - 3s - loss: 1.1908 - accuracy: 0.6301 - val_loss: 4.1784 - val_accuracy: 0.0887\n",
+      "Epoch 4563/5000\n",
+      "919/919 - 3s - loss: 1.1907 - accuracy: 0.6284 - val_loss: 4.1679 - val_accuracy: 0.0901\n",
+      "Epoch 4564/5000\n",
+      "919/919 - 3s - loss: 1.2035 - accuracy: 0.6240 - val_loss: 4.1639 - val_accuracy: 0.0897\n",
+      "Epoch 4565/5000\n",
+      "919/919 - 3s - loss: 1.2155 - accuracy: 0.6309 - val_loss: 4.1776 - val_accuracy: 0.0892\n",
+      "Epoch 4566/5000\n",
+      "919/919 - 3s - loss: 1.1799 - accuracy: 0.6317 - val_loss: 4.1643 - val_accuracy: 0.0895\n",
+      "Epoch 4567/5000\n",
+      "919/919 - 3s - loss: 1.1825 - accuracy: 0.6290 - val_loss: 4.1697 - val_accuracy: 0.0897\n",
+      "Epoch 4568/5000\n",
+      "919/919 - 3s - loss: 1.1896 - accuracy: 0.6297 - val_loss: 4.1836 - val_accuracy: 0.0890\n",
+      "Epoch 4569/5000\n",
+      "919/919 - 3s - loss: 1.1946 - accuracy: 0.6262 - val_loss: 4.1932 - val_accuracy: 0.0883\n",
+      "Epoch 4570/5000\n",
+      "919/919 - 3s - loss: 1.1888 - accuracy: 0.6260 - val_loss: 4.1999 - val_accuracy: 0.0888\n",
+      "Epoch 4571/5000\n",
+      "919/919 - 3s - loss: 1.2207 - accuracy: 0.6319 - val_loss: 4.1998 - val_accuracy: 0.0889\n",
+      "Epoch 4572/5000\n",
+      "919/919 - 3s - loss: 1.1929 - accuracy: 0.6292 - val_loss: 4.1911 - val_accuracy: 0.0885\n",
+      "Epoch 4573/5000\n",
+      "919/919 - 3s - loss: 1.1850 - accuracy: 0.6317 - val_loss: 4.1825 - val_accuracy: 0.0886\n",
+      "Epoch 4574/5000\n",
+      "919/919 - 3s - loss: 1.1696 - accuracy: 0.6308 - val_loss: 4.2067 - val_accuracy: 0.0881\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4575/5000\n",
+      "919/919 - 3s - loss: 1.1995 - accuracy: 0.6283 - val_loss: 4.1995 - val_accuracy: 0.0883\n",
+      "Epoch 4576/5000\n",
+      "919/919 - 3s - loss: 1.2008 - accuracy: 0.6335 - val_loss: 4.1771 - val_accuracy: 0.0887\n",
+      "Epoch 4577/5000\n",
+      "919/919 - 3s - loss: 1.1843 - accuracy: 0.6283 - val_loss: 4.1847 - val_accuracy: 0.0883\n",
+      "Epoch 4578/5000\n",
+      "919/919 - 3s - loss: 1.2061 - accuracy: 0.6295 - val_loss: 4.1716 - val_accuracy: 0.0885\n",
+      "Epoch 4579/5000\n",
+      "919/919 - 3s - loss: 1.1779 - accuracy: 0.6300 - val_loss: 4.1854 - val_accuracy: 0.0883\n",
+      "Epoch 4580/5000\n",
+      "919/919 - 3s - loss: 1.1816 - accuracy: 0.6286 - val_loss: 4.1816 - val_accuracy: 0.0884\n",
+      "Epoch 4581/5000\n",
+      "919/919 - 3s - loss: 1.1882 - accuracy: 0.6278 - val_loss: 4.1778 - val_accuracy: 0.0881\n",
+      "Epoch 4582/5000\n",
+      "919/919 - 3s - loss: 1.1783 - accuracy: 0.6300 - val_loss: 4.1944 - val_accuracy: 0.0880\n",
+      "Epoch 4583/5000\n",
+      "919/919 - 3s - loss: 1.1787 - accuracy: 0.6317 - val_loss: 4.1831 - val_accuracy: 0.0879\n",
+      "Epoch 4584/5000\n",
+      "919/919 - 3s - loss: 1.1700 - accuracy: 0.6348 - val_loss: 4.1866 - val_accuracy: 0.0890\n",
+      "Epoch 4585/5000\n",
+      "919/919 - 3s - loss: 1.1984 - accuracy: 0.6301 - val_loss: 4.1937 - val_accuracy: 0.0888\n",
+      "Epoch 4586/5000\n",
+      "919/919 - 3s - loss: 1.2810 - accuracy: 0.6324 - val_loss: 4.1999 - val_accuracy: 0.0896\n",
+      "Epoch 4587/5000\n",
+      "919/919 - 3s - loss: 1.2052 - accuracy: 0.6269 - val_loss: 4.1867 - val_accuracy: 0.0892\n",
+      "Epoch 4588/5000\n",
+      "919/919 - 3s - loss: 1.1878 - accuracy: 0.6278 - val_loss: 4.1704 - val_accuracy: 0.0891\n",
+      "Epoch 4589/5000\n",
+      "919/919 - 3s - loss: 1.1932 - accuracy: 0.6274 - val_loss: 4.1684 - val_accuracy: 0.0888\n",
+      "Epoch 4590/5000\n",
+      "919/919 - 3s - loss: 1.1800 - accuracy: 0.6299 - val_loss: 4.1838 - val_accuracy: 0.0893\n",
+      "Epoch 4591/5000\n",
+      "919/919 - 3s - loss: 1.1808 - accuracy: 0.6292 - val_loss: 4.1890 - val_accuracy: 0.0889\n",
+      "Epoch 4592/5000\n",
+      "919/919 - 3s - loss: 1.3220 - accuracy: 0.6339 - val_loss: 4.1896 - val_accuracy: 0.0885\n",
+      "Epoch 4593/5000\n",
+      "919/919 - 3s - loss: 1.1885 - accuracy: 0.6316 - val_loss: 4.1808 - val_accuracy: 0.0888\n",
+      "Epoch 4594/5000\n",
+      "919/919 - 3s - loss: 1.1918 - accuracy: 0.6315 - val_loss: 4.1786 - val_accuracy: 0.0891\n",
+      "Epoch 4595/5000\n",
+      "919/919 - 3s - loss: 1.1890 - accuracy: 0.6260 - val_loss: 4.1845 - val_accuracy: 0.0887\n",
+      "Epoch 4596/5000\n",
+      "919/919 - 3s - loss: 1.1907 - accuracy: 0.6316 - val_loss: 4.1814 - val_accuracy: 0.0895\n",
+      "Epoch 4597/5000\n",
+      "919/919 - 3s - loss: 1.1888 - accuracy: 0.6314 - val_loss: 4.1763 - val_accuracy: 0.0886\n",
+      "Epoch 4598/5000\n",
+      "919/919 - 3s - loss: 1.1733 - accuracy: 0.6358 - val_loss: 4.1889 - val_accuracy: 0.0891\n",
+      "Epoch 4599/5000\n",
+      "919/919 - 3s - loss: 1.1893 - accuracy: 0.6293 - val_loss: 4.1995 - val_accuracy: 0.0894\n",
+      "Epoch 4600/5000\n",
+      "919/919 - 3s - loss: 1.1875 - accuracy: 0.6295 - val_loss: 4.2155 - val_accuracy: 0.0882\n",
+      "Epoch 4601/5000\n",
+      "919/919 - 3s - loss: 1.1835 - accuracy: 0.6301 - val_loss: 4.2097 - val_accuracy: 0.0880\n",
+      "Epoch 4602/5000\n",
+      "919/919 - 3s - loss: 1.1825 - accuracy: 0.6315 - val_loss: 4.2205 - val_accuracy: 0.0875\n",
+      "Epoch 4603/5000\n",
+      "919/919 - 3s - loss: 1.1793 - accuracy: 0.6310 - val_loss: 4.2242 - val_accuracy: 0.0887\n",
+      "Epoch 4604/5000\n",
+      "919/919 - 3s - loss: 1.2361 - accuracy: 0.6322 - val_loss: 4.2202 - val_accuracy: 0.0885\n",
+      "Epoch 4605/5000\n",
+      "919/919 - 3s - loss: 1.1876 - accuracy: 0.6297 - val_loss: 4.2019 - val_accuracy: 0.0891\n",
+      "Epoch 4606/5000\n",
+      "919/919 - 3s - loss: 1.2020 - accuracy: 0.6286 - val_loss: 4.1990 - val_accuracy: 0.0887\n",
+      "Epoch 4607/5000\n",
+      "919/919 - 3s - loss: 1.1999 - accuracy: 0.6293 - val_loss: 4.1903 - val_accuracy: 0.0881\n",
+      "Epoch 4608/5000\n",
+      "919/919 - 3s - loss: 1.2103 - accuracy: 0.6285 - val_loss: 4.1803 - val_accuracy: 0.0887\n",
+      "Epoch 4609/5000\n",
+      "919/919 - 3s - loss: 1.2637 - accuracy: 0.6306 - val_loss: 4.1908 - val_accuracy: 0.0883\n",
+      "Epoch 4610/5000\n",
+      "919/919 - 3s - loss: 1.1933 - accuracy: 0.6277 - val_loss: 4.2049 - val_accuracy: 0.0880\n",
+      "Epoch 4611/5000\n",
+      "919/919 - 3s - loss: 1.1728 - accuracy: 0.6301 - val_loss: 4.1968 - val_accuracy: 0.0885\n",
+      "Epoch 4612/5000\n",
+      "919/919 - 3s - loss: 1.1746 - accuracy: 0.6293 - val_loss: 4.1960 - val_accuracy: 0.0888\n",
+      "Epoch 4613/5000\n",
+      "919/919 - 3s - loss: 1.2346 - accuracy: 0.6299 - val_loss: 4.2016 - val_accuracy: 0.0881\n",
+      "Epoch 4614/5000\n",
+      "919/919 - 3s - loss: 1.2008 - accuracy: 0.6322 - val_loss: 4.2003 - val_accuracy: 0.0886\n",
+      "Epoch 4615/5000\n",
+      "919/919 - 3s - loss: 1.2042 - accuracy: 0.6266 - val_loss: 4.2077 - val_accuracy: 0.0878\n",
+      "Epoch 4616/5000\n",
+      "919/919 - 3s - loss: 1.1881 - accuracy: 0.6295 - val_loss: 4.2025 - val_accuracy: 0.0886\n",
+      "Epoch 4617/5000\n",
+      "919/919 - 3s - loss: 1.1961 - accuracy: 0.6328 - val_loss: 4.1939 - val_accuracy: 0.0888\n",
+      "Epoch 4618/5000\n",
+      "919/919 - 3s - loss: 1.1977 - accuracy: 0.6282 - val_loss: 4.2126 - val_accuracy: 0.0883\n",
+      "Epoch 4619/5000\n",
+      "919/919 - 3s - loss: 1.1836 - accuracy: 0.6308 - val_loss: 4.2080 - val_accuracy: 0.0882\n",
+      "Epoch 4620/5000\n",
+      "919/919 - 3s - loss: 1.1803 - accuracy: 0.6282 - val_loss: 4.1963 - val_accuracy: 0.0882\n",
+      "Epoch 4621/5000\n",
+      "919/919 - 3s - loss: 1.1767 - accuracy: 0.6337 - val_loss: 4.1927 - val_accuracy: 0.0886\n",
+      "Epoch 4622/5000\n",
+      "919/919 - 3s - loss: 1.1665 - accuracy: 0.6382 - val_loss: 4.1889 - val_accuracy: 0.0892\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4623/5000\n",
+      "919/919 - 3s - loss: 1.2416 - accuracy: 0.6339 - val_loss: 4.2040 - val_accuracy: 0.0897\n",
+      "Epoch 4624/5000\n",
+      "919/919 - 3s - loss: 1.2089 - accuracy: 0.6265 - val_loss: 4.2157 - val_accuracy: 0.0894\n",
+      "Epoch 4625/5000\n",
+      "919/919 - 3s - loss: 1.1902 - accuracy: 0.6269 - val_loss: 4.2106 - val_accuracy: 0.0898\n",
+      "Epoch 4626/5000\n",
+      "919/919 - 3s - loss: 1.1769 - accuracy: 0.6324 - val_loss: 4.1961 - val_accuracy: 0.0893\n",
+      "Epoch 4627/5000\n",
+      "919/919 - 3s - loss: 1.2052 - accuracy: 0.6299 - val_loss: 4.1856 - val_accuracy: 0.0890\n",
+      "Epoch 4628/5000\n",
+      "919/919 - 3s - loss: 1.1993 - accuracy: 0.6356 - val_loss: 4.1842 - val_accuracy: 0.0892\n",
+      "Epoch 4629/5000\n",
+      "919/919 - 3s - loss: 1.1823 - accuracy: 0.6312 - val_loss: 4.1853 - val_accuracy: 0.0898\n",
+      "Epoch 4630/5000\n",
+      "919/919 - 3s - loss: 1.1843 - accuracy: 0.6325 - val_loss: 4.1746 - val_accuracy: 0.0895\n",
+      "Epoch 4631/5000\n",
+      "919/919 - 3s - loss: 1.1942 - accuracy: 0.6271 - val_loss: 4.1829 - val_accuracy: 0.0894\n",
+      "Epoch 4632/5000\n",
+      "919/919 - 3s - loss: 1.1763 - accuracy: 0.6297 - val_loss: 4.1782 - val_accuracy: 0.0897\n",
+      "Epoch 4633/5000\n",
+      "919/919 - 3s - loss: 1.2948 - accuracy: 0.6327 - val_loss: 4.1879 - val_accuracy: 0.0897\n",
+      "Epoch 4634/5000\n",
+      "919/919 - 3s - loss: 1.1699 - accuracy: 0.6344 - val_loss: 4.1871 - val_accuracy: 0.0892\n",
+      "Epoch 4635/5000\n",
+      "919/919 - 3s - loss: 1.1851 - accuracy: 0.6282 - val_loss: 4.1906 - val_accuracy: 0.0891\n",
+      "Epoch 4636/5000\n",
+      "919/919 - 3s - loss: 1.2147 - accuracy: 0.6310 - val_loss: 4.1991 - val_accuracy: 0.0883\n",
+      "Epoch 4637/5000\n",
+      "919/919 - 3s - loss: 1.1855 - accuracy: 0.6297 - val_loss: 4.1903 - val_accuracy: 0.0883\n",
+      "Epoch 4638/5000\n",
+      "919/919 - 3s - loss: 1.1882 - accuracy: 0.6305 - val_loss: 4.1935 - val_accuracy: 0.0899\n",
+      "Epoch 4639/5000\n",
+      "919/919 - 3s - loss: 1.2242 - accuracy: 0.6276 - val_loss: 4.1947 - val_accuracy: 0.0901\n",
+      "Epoch 4640/5000\n",
+      "919/919 - 3s - loss: 1.1676 - accuracy: 0.6356 - val_loss: 4.2007 - val_accuracy: 0.0893\n",
+      "Epoch 4641/5000\n",
+      "919/919 - 3s - loss: 1.1773 - accuracy: 0.6320 - val_loss: 4.2018 - val_accuracy: 0.0897\n",
+      "Epoch 4642/5000\n",
+      "919/919 - 3s - loss: 1.1966 - accuracy: 0.6335 - val_loss: 4.1822 - val_accuracy: 0.0897\n",
+      "Epoch 4643/5000\n",
+      "919/919 - 3s - loss: 1.1743 - accuracy: 0.6316 - val_loss: 4.1911 - val_accuracy: 0.0897\n",
+      "Epoch 4644/5000\n",
+      "919/919 - 3s - loss: 1.1823 - accuracy: 0.6301 - val_loss: 4.1973 - val_accuracy: 0.0891\n",
+      "Epoch 4645/5000\n",
+      "919/919 - 3s - loss: 1.1938 - accuracy: 0.6286 - val_loss: 4.2149 - val_accuracy: 0.0887\n",
+      "Epoch 4646/5000\n",
+      "919/919 - 3s - loss: 1.2096 - accuracy: 0.6327 - val_loss: 4.2149 - val_accuracy: 0.0892\n",
+      "Epoch 4647/5000\n",
+      "919/919 - 3s - loss: 1.2084 - accuracy: 0.6297 - val_loss: 4.2072 - val_accuracy: 0.0892\n",
+      "Epoch 4648/5000\n",
+      "919/919 - 3s - loss: 1.1942 - accuracy: 0.6308 - val_loss: 4.2082 - val_accuracy: 0.0883\n",
+      "Epoch 4649/5000\n",
+      "919/919 - 3s - loss: 1.1900 - accuracy: 0.6356 - val_loss: 4.2200 - val_accuracy: 0.0889\n",
+      "Epoch 4650/5000\n",
+      "919/919 - 3s - loss: 1.1828 - accuracy: 0.6299 - val_loss: 4.1908 - val_accuracy: 0.0896\n",
+      "Epoch 4651/5000\n",
+      "919/919 - 3s - loss: 1.1816 - accuracy: 0.6297 - val_loss: 4.1950 - val_accuracy: 0.0900\n",
+      "Epoch 4652/5000\n",
+      "919/919 - 3s - loss: 1.2177 - accuracy: 0.6282 - val_loss: 4.1922 - val_accuracy: 0.0894\n",
+      "Epoch 4653/5000\n",
+      "919/919 - 3s - loss: 1.2549 - accuracy: 0.6346 - val_loss: 4.1956 - val_accuracy: 0.0893\n",
+      "Epoch 4654/5000\n",
+      "919/919 - 3s - loss: 1.1807 - accuracy: 0.6316 - val_loss: 4.2069 - val_accuracy: 0.0901\n",
+      "Epoch 4655/5000\n",
+      "919/919 - 3s - loss: 1.1640 - accuracy: 0.6354 - val_loss: 4.2172 - val_accuracy: 0.0895\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4656/5000\n",
+      "919/919 - 3s - loss: 1.2082 - accuracy: 0.6335 - val_loss: 4.2074 - val_accuracy: 0.0905\n",
+      "Epoch 4657/5000\n",
+      "919/919 - 3s - loss: 1.1781 - accuracy: 0.6320 - val_loss: 4.1993 - val_accuracy: 0.0890\n",
+      "Epoch 4658/5000\n",
+      "919/919 - 3s - loss: 1.1716 - accuracy: 0.6347 - val_loss: 4.1960 - val_accuracy: 0.0888\n",
+      "Epoch 4659/5000\n",
+      "919/919 - 3s - loss: 1.1948 - accuracy: 0.6273 - val_loss: 4.1811 - val_accuracy: 0.0892\n",
+      "Epoch 4660/5000\n",
+      "919/919 - 3s - loss: 1.1671 - accuracy: 0.6366 - val_loss: 4.1956 - val_accuracy: 0.0886\n",
+      "Epoch 4661/5000\n",
+      "919/919 - 3s - loss: 1.1823 - accuracy: 0.6295 - val_loss: 4.2066 - val_accuracy: 0.0889\n",
+      "Epoch 4662/5000\n",
+      "919/919 - 3s - loss: 1.1855 - accuracy: 0.6330 - val_loss: 4.2027 - val_accuracy: 0.0892\n",
+      "Epoch 4663/5000\n",
+      "919/919 - 3s - loss: 1.1955 - accuracy: 0.6286 - val_loss: 4.2025 - val_accuracy: 0.0883\n",
+      "Epoch 4664/5000\n",
+      "919/919 - 3s - loss: 1.1834 - accuracy: 0.6329 - val_loss: 4.1928 - val_accuracy: 0.0891\n",
+      "Epoch 4665/5000\n",
+      "919/919 - 3s - loss: 1.1859 - accuracy: 0.6307 - val_loss: 4.2076 - val_accuracy: 0.0883\n",
+      "Epoch 4666/5000\n",
+      "919/919 - 3s - loss: 1.1921 - accuracy: 0.6347 - val_loss: 4.1948 - val_accuracy: 0.0892\n",
+      "Epoch 4667/5000\n",
+      "919/919 - 3s - loss: 1.1836 - accuracy: 0.6337 - val_loss: 4.2083 - val_accuracy: 0.0889\n",
+      "Epoch 4668/5000\n",
+      "919/919 - 3s - loss: 1.1717 - accuracy: 0.6348 - val_loss: 4.2048 - val_accuracy: 0.0890\n",
+      "Epoch 4669/5000\n",
+      "919/919 - 3s - loss: 1.2307 - accuracy: 0.6330 - val_loss: 4.2066 - val_accuracy: 0.0885\n",
+      "Epoch 4670/5000\n",
+      "919/919 - 3s - loss: 1.1812 - accuracy: 0.6307 - val_loss: 4.1974 - val_accuracy: 0.0885\n",
+      "Epoch 4671/5000\n",
+      "919/919 - 3s - loss: 1.1843 - accuracy: 0.6351 - val_loss: 4.2057 - val_accuracy: 0.0891\n",
+      "Epoch 4672/5000\n",
+      "919/919 - 3s - loss: 1.1939 - accuracy: 0.6344 - val_loss: 4.2071 - val_accuracy: 0.0886\n",
+      "Epoch 4673/5000\n",
+      "919/919 - 3s - loss: 1.2129 - accuracy: 0.6329 - val_loss: 4.1974 - val_accuracy: 0.0890\n",
+      "Epoch 4674/5000\n",
+      "919/919 - 3s - loss: 1.2546 - accuracy: 0.6310 - val_loss: 4.1864 - val_accuracy: 0.0885\n",
+      "Epoch 4675/5000\n",
+      "919/919 - 3s - loss: 1.2194 - accuracy: 0.6318 - val_loss: 4.1999 - val_accuracy: 0.0891\n",
+      "Epoch 4676/5000\n",
+      "919/919 - 3s - loss: 1.2352 - accuracy: 0.6298 - val_loss: 4.2123 - val_accuracy: 0.0882\n",
+      "Epoch 4677/5000\n",
+      "919/919 - 3s - loss: 1.1888 - accuracy: 0.6297 - val_loss: 4.2264 - val_accuracy: 0.0874\n",
+      "Epoch 4678/5000\n",
+      "919/919 - 3s - loss: 1.1872 - accuracy: 0.6322 - val_loss: 4.2262 - val_accuracy: 0.0888\n",
+      "Epoch 4679/5000\n",
+      "919/919 - 3s - loss: 1.1912 - accuracy: 0.6326 - val_loss: 4.2023 - val_accuracy: 0.0886\n",
+      "Epoch 4680/5000\n",
+      "919/919 - 3s - loss: 1.1907 - accuracy: 0.6332 - val_loss: 4.1886 - val_accuracy: 0.0883\n",
+      "Epoch 4681/5000\n",
+      "919/919 - 3s - loss: 1.1798 - accuracy: 0.6320 - val_loss: 4.1955 - val_accuracy: 0.0891\n",
+      "Epoch 4682/5000\n",
+      "919/919 - 3s - loss: 1.1955 - accuracy: 0.6325 - val_loss: 4.1982 - val_accuracy: 0.0882\n",
+      "Epoch 4683/5000\n",
+      "919/919 - 3s - loss: 1.1859 - accuracy: 0.6312 - val_loss: 4.2001 - val_accuracy: 0.0881\n",
+      "Epoch 4684/5000\n",
+      "919/919 - 3s - loss: 1.2105 - accuracy: 0.6329 - val_loss: 4.2050 - val_accuracy: 0.0877\n",
+      "Epoch 4685/5000\n",
+      "919/919 - 3s - loss: 1.1904 - accuracy: 0.6302 - val_loss: 4.2056 - val_accuracy: 0.0883\n",
+      "Epoch 4686/5000\n",
+      "919/919 - 3s - loss: 1.1763 - accuracy: 0.6345 - val_loss: 4.2122 - val_accuracy: 0.0883\n",
+      "Epoch 4687/5000\n",
+      "919/919 - 3s - loss: 1.1743 - accuracy: 0.6320 - val_loss: 4.2083 - val_accuracy: 0.0882\n",
+      "Epoch 4688/5000\n",
+      "919/919 - 3s - loss: 1.1729 - accuracy: 0.6337 - val_loss: 4.2151 - val_accuracy: 0.0883\n",
+      "Epoch 4689/5000\n",
+      "919/919 - 3s - loss: 1.2002 - accuracy: 0.6291 - val_loss: 4.2174 - val_accuracy: 0.0873\n",
+      "Epoch 4690/5000\n",
+      "919/919 - 3s - loss: 1.1823 - accuracy: 0.6295 - val_loss: 4.2118 - val_accuracy: 0.0876\n",
+      "Epoch 4691/5000\n",
+      "919/919 - 3s - loss: 1.1849 - accuracy: 0.6313 - val_loss: 4.1984 - val_accuracy: 0.0882\n",
+      "Epoch 4692/5000\n",
+      "919/919 - 3s - loss: 1.2060 - accuracy: 0.6299 - val_loss: 4.1995 - val_accuracy: 0.0883\n",
+      "Epoch 4693/5000\n",
+      "919/919 - 3s - loss: 1.2829 - accuracy: 0.6338 - val_loss: 4.1958 - val_accuracy: 0.0885\n",
+      "Epoch 4694/5000\n",
+      "919/919 - 3s - loss: 1.1889 - accuracy: 0.6305 - val_loss: 4.2118 - val_accuracy: 0.0888\n",
+      "Epoch 4695/5000\n",
+      "919/919 - 3s - loss: 1.1928 - accuracy: 0.6314 - val_loss: 4.2096 - val_accuracy: 0.0889\n",
+      "Epoch 4696/5000\n",
+      "919/919 - 3s - loss: 1.1885 - accuracy: 0.6316 - val_loss: 4.2112 - val_accuracy: 0.0893\n",
+      "Epoch 4697/5000\n",
+      "919/919 - 3s - loss: 1.2066 - accuracy: 0.6341 - val_loss: 4.2103 - val_accuracy: 0.0889\n",
+      "Epoch 4698/5000\n",
+      "919/919 - 3s - loss: 1.1926 - accuracy: 0.6310 - val_loss: 4.1893 - val_accuracy: 0.0890\n",
+      "Epoch 4699/5000\n",
+      "919/919 - 3s - loss: 1.1602 - accuracy: 0.6357 - val_loss: 4.1737 - val_accuracy: 0.0894\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4700/5000\n",
+      "919/919 - 3s - loss: 1.1760 - accuracy: 0.6335 - val_loss: 4.1927 - val_accuracy: 0.0894\n",
+      "Epoch 4701/5000\n",
+      "919/919 - 3s - loss: 1.1941 - accuracy: 0.6317 - val_loss: 4.1995 - val_accuracy: 0.0896\n",
+      "Epoch 4702/5000\n",
+      "919/919 - 3s - loss: 1.1882 - accuracy: 0.6325 - val_loss: 4.2014 - val_accuracy: 0.0885\n",
+      "Epoch 4703/5000\n",
+      "919/919 - 3s - loss: 1.2352 - accuracy: 0.6318 - val_loss: 4.2128 - val_accuracy: 0.0890\n",
+      "Epoch 4704/5000\n",
+      "919/919 - 3s - loss: 1.2129 - accuracy: 0.6362 - val_loss: 4.2285 - val_accuracy: 0.0890\n",
+      "Epoch 4705/5000\n",
+      "919/919 - 3s - loss: 1.2169 - accuracy: 0.6339 - val_loss: 4.2335 - val_accuracy: 0.0895\n",
+      "Epoch 4706/5000\n",
+      "919/919 - 3s - loss: 1.1859 - accuracy: 0.6303 - val_loss: 4.2124 - val_accuracy: 0.0901\n",
+      "Epoch 4707/5000\n",
+      "919/919 - 3s - loss: 1.1783 - accuracy: 0.6301 - val_loss: 4.2124 - val_accuracy: 0.0896\n",
+      "Epoch 4708/5000\n",
+      "919/919 - 3s - loss: 1.1611 - accuracy: 0.6350 - val_loss: 4.1998 - val_accuracy: 0.0897\n",
+      "Epoch 4709/5000\n",
+      "919/919 - 3s - loss: 1.1579 - accuracy: 0.6365 - val_loss: 4.1971 - val_accuracy: 0.0895\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4710/5000\n",
+      "919/919 - 3s - loss: 1.1750 - accuracy: 0.6317 - val_loss: 4.1852 - val_accuracy: 0.0892\n",
+      "Epoch 4711/5000\n",
+      "919/919 - 3s - loss: 1.1931 - accuracy: 0.6288 - val_loss: 4.2052 - val_accuracy: 0.0892\n",
+      "Epoch 4712/5000\n",
+      "919/919 - 3s - loss: 1.1816 - accuracy: 0.6312 - val_loss: 4.2123 - val_accuracy: 0.0892\n",
+      "Epoch 4713/5000\n",
+      "919/919 - 3s - loss: 1.1817 - accuracy: 0.6310 - val_loss: 4.2078 - val_accuracy: 0.0893\n",
+      "Epoch 4714/5000\n",
+      "919/919 - 3s - loss: 1.1705 - accuracy: 0.6363 - val_loss: 4.2172 - val_accuracy: 0.0903\n",
+      "Epoch 4715/5000\n",
+      "919/919 - 3s - loss: 1.1902 - accuracy: 0.6309 - val_loss: 4.2104 - val_accuracy: 0.0906\n",
+      "Epoch 4716/5000\n",
+      "919/919 - 3s - loss: 1.1781 - accuracy: 0.6347 - val_loss: 4.2182 - val_accuracy: 0.0903\n",
+      "Epoch 4717/5000\n",
+      "919/919 - 3s - loss: 1.1829 - accuracy: 0.6295 - val_loss: 4.2329 - val_accuracy: 0.0898\n",
+      "Epoch 4718/5000\n",
+      "919/919 - 3s - loss: 1.1614 - accuracy: 0.6374 - val_loss: 4.2262 - val_accuracy: 0.0899\n",
+      "Epoch 4719/5000\n",
+      "919/919 - 3s - loss: 1.2538 - accuracy: 0.6312 - val_loss: 4.2243 - val_accuracy: 0.0896\n",
+      "Epoch 4720/5000\n",
+      "919/919 - 3s - loss: 1.1715 - accuracy: 0.6344 - val_loss: 4.2158 - val_accuracy: 0.0892\n",
+      "Epoch 4721/5000\n",
+      "919/919 - 3s - loss: 1.2072 - accuracy: 0.6329 - val_loss: 4.2156 - val_accuracy: 0.0900\n",
+      "Epoch 4722/5000\n",
+      "919/919 - 3s - loss: 1.2021 - accuracy: 0.6337 - val_loss: 4.2143 - val_accuracy: 0.0891\n",
+      "Epoch 4723/5000\n",
+      "919/919 - 3s - loss: 1.2974 - accuracy: 0.6314 - val_loss: 4.2014 - val_accuracy: 0.0895\n",
+      "Epoch 4724/5000\n",
+      "919/919 - 3s - loss: 1.2056 - accuracy: 0.6322 - val_loss: 4.2227 - val_accuracy: 0.0898\n",
+      "Epoch 4725/5000\n",
+      "919/919 - 3s - loss: 1.1679 - accuracy: 0.6372 - val_loss: 4.2245 - val_accuracy: 0.0894\n",
+      "Epoch 4726/5000\n",
+      "919/919 - 3s - loss: 1.2888 - accuracy: 0.6320 - val_loss: 4.2138 - val_accuracy: 0.0899\n",
+      "Epoch 4727/5000\n",
+      "919/919 - 3s - loss: 1.1629 - accuracy: 0.6345 - val_loss: 4.2232 - val_accuracy: 0.0895\n",
+      "Epoch 4728/5000\n",
+      "919/919 - 3s - loss: 1.1770 - accuracy: 0.6310 - val_loss: 4.2187 - val_accuracy: 0.0889\n",
+      "Epoch 4729/5000\n",
+      "919/919 - 3s - loss: 1.2099 - accuracy: 0.6347 - val_loss: 4.2093 - val_accuracy: 0.0900\n",
+      "Epoch 4730/5000\n",
+      "919/919 - 3s - loss: 1.1867 - accuracy: 0.6293 - val_loss: 4.2210 - val_accuracy: 0.0891\n",
+      "Epoch 4731/5000\n",
+      "919/919 - 3s - loss: 1.2030 - accuracy: 0.6341 - val_loss: 4.2104 - val_accuracy: 0.0896\n",
+      "Epoch 4732/5000\n",
+      "919/919 - 3s - loss: 1.1868 - accuracy: 0.6303 - val_loss: 4.2029 - val_accuracy: 0.0899\n",
+      "Epoch 4733/5000\n",
+      "919/919 - 3s - loss: 1.1974 - accuracy: 0.6307 - val_loss: 4.2088 - val_accuracy: 0.0900\n",
+      "Epoch 4734/5000\n",
+      "919/919 - 3s - loss: 1.1625 - accuracy: 0.6317 - val_loss: 4.2261 - val_accuracy: 0.0895\n",
+      "Epoch 4735/5000\n",
+      "919/919 - 3s - loss: 1.1792 - accuracy: 0.6347 - val_loss: 4.2313 - val_accuracy: 0.0888\n",
+      "Epoch 4736/5000\n",
+      "919/919 - 3s - loss: 1.1656 - accuracy: 0.6316 - val_loss: 4.2455 - val_accuracy: 0.0892\n",
+      "Epoch 4737/5000\n",
+      "919/919 - 3s - loss: 1.1721 - accuracy: 0.6338 - val_loss: 4.2636 - val_accuracy: 0.0893\n",
+      "Epoch 4738/5000\n",
+      "919/919 - 3s - loss: 1.3150 - accuracy: 0.6355 - val_loss: 4.2626 - val_accuracy: 0.0888\n",
+      "Epoch 4739/5000\n",
+      "919/919 - 3s - loss: 1.1739 - accuracy: 0.6349 - val_loss: 4.2426 - val_accuracy: 0.0893\n",
+      "Epoch 4740/5000\n",
+      "919/919 - 3s - loss: 1.2646 - accuracy: 0.6316 - val_loss: 4.2437 - val_accuracy: 0.0898\n",
+      "Epoch 4741/5000\n",
+      "919/919 - 3s - loss: 1.1575 - accuracy: 0.6386 - val_loss: 4.2388 - val_accuracy: 0.0893\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4742/5000\n",
+      "919/919 - 3s - loss: 1.1739 - accuracy: 0.6341 - val_loss: 4.2495 - val_accuracy: 0.0895\n",
+      "Epoch 4743/5000\n",
+      "919/919 - 3s - loss: 1.1674 - accuracy: 0.6348 - val_loss: 4.2467 - val_accuracy: 0.0896\n",
+      "Epoch 4744/5000\n",
+      "919/919 - 3s - loss: 1.1824 - accuracy: 0.6331 - val_loss: 4.2348 - val_accuracy: 0.0898\n",
+      "Epoch 4745/5000\n",
+      "919/919 - 3s - loss: 1.1912 - accuracy: 0.6309 - val_loss: 4.2280 - val_accuracy: 0.0894\n",
+      "Epoch 4746/5000\n",
+      "919/919 - 3s - loss: 1.1746 - accuracy: 0.6339 - val_loss: 4.2334 - val_accuracy: 0.0901\n",
+      "Epoch 4747/5000\n",
+      "919/919 - 3s - loss: 1.1870 - accuracy: 0.6308 - val_loss: 4.2398 - val_accuracy: 0.0898\n",
+      "Epoch 4748/5000\n",
+      "919/919 - 3s - loss: 1.2740 - accuracy: 0.6340 - val_loss: 4.2439 - val_accuracy: 0.0895\n",
+      "Epoch 4749/5000\n",
+      "919/919 - 3s - loss: 1.1902 - accuracy: 0.6333 - val_loss: 4.2315 - val_accuracy: 0.0895\n",
+      "Epoch 4750/5000\n",
+      "919/919 - 3s - loss: 1.2119 - accuracy: 0.6306 - val_loss: 4.2264 - val_accuracy: 0.0892\n",
+      "Epoch 4751/5000\n",
+      "919/919 - 3s - loss: 1.1740 - accuracy: 0.6356 - val_loss: 4.2271 - val_accuracy: 0.0896\n",
+      "Epoch 4752/5000\n",
+      "919/919 - 3s - loss: 1.1855 - accuracy: 0.6335 - val_loss: 4.2106 - val_accuracy: 0.0896\n",
+      "Epoch 4753/5000\n",
+      "919/919 - 3s - loss: 1.1871 - accuracy: 0.6352 - val_loss: 4.2100 - val_accuracy: 0.0897\n",
+      "Epoch 4754/5000\n",
+      "919/919 - 3s - loss: 1.1880 - accuracy: 0.6294 - val_loss: 4.2067 - val_accuracy: 0.0898\n",
+      "Epoch 4755/5000\n",
+      "919/919 - 3s - loss: 1.1940 - accuracy: 0.6344 - val_loss: 4.2087 - val_accuracy: 0.0896\n",
+      "Epoch 4756/5000\n",
+      "919/919 - 3s - loss: 1.1878 - accuracy: 0.6369 - val_loss: 4.2061 - val_accuracy: 0.0896\n",
+      "Epoch 4757/5000\n",
+      "919/919 - 3s - loss: 1.2985 - accuracy: 0.6345 - val_loss: 4.1990 - val_accuracy: 0.0897\n",
+      "Epoch 4758/5000\n",
+      "919/919 - 3s - loss: 1.1562 - accuracy: 0.6388 - val_loss: 4.2002 - val_accuracy: 0.0898\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4759/5000\n",
+      "919/919 - 3s - loss: 1.1592 - accuracy: 0.6351 - val_loss: 4.2091 - val_accuracy: 0.0900\n",
+      "Epoch 4760/5000\n",
+      "919/919 - 3s - loss: 1.1866 - accuracy: 0.6304 - val_loss: 4.2111 - val_accuracy: 0.0900\n",
+      "Epoch 4761/5000\n",
+      "919/919 - 3s - loss: 1.1780 - accuracy: 0.6321 - val_loss: 4.2033 - val_accuracy: 0.0900\n",
+      "Epoch 4762/5000\n",
+      "919/919 - 3s - loss: 1.1796 - accuracy: 0.6337 - val_loss: 4.2146 - val_accuracy: 0.0902\n",
+      "Epoch 4763/5000\n",
+      "919/919 - 3s - loss: 1.1779 - accuracy: 0.6335 - val_loss: 4.2178 - val_accuracy: 0.0905\n",
+      "Epoch 4764/5000\n",
+      "919/919 - 3s - loss: 1.2170 - accuracy: 0.6337 - val_loss: 4.2274 - val_accuracy: 0.0900\n",
+      "Epoch 4765/5000\n",
+      "919/919 - 3s - loss: 1.2301 - accuracy: 0.6323 - val_loss: 4.2394 - val_accuracy: 0.0903\n",
+      "Epoch 4766/5000\n",
+      "919/919 - 3s - loss: 1.1658 - accuracy: 0.6369 - val_loss: 4.2270 - val_accuracy: 0.0911\n",
+      "Epoch 4767/5000\n",
+      "919/919 - 3s - loss: 1.1717 - accuracy: 0.6350 - val_loss: 4.2270 - val_accuracy: 0.0903\n",
+      "Epoch 4768/5000\n",
+      "919/919 - 3s - loss: 1.1847 - accuracy: 0.6303 - val_loss: 4.2251 - val_accuracy: 0.0890\n",
+      "Epoch 4769/5000\n",
+      "919/919 - 3s - loss: 1.1820 - accuracy: 0.6331 - val_loss: 4.2324 - val_accuracy: 0.0892\n",
+      "Epoch 4770/5000\n",
+      "919/919 - 3s - loss: 1.1758 - accuracy: 0.6352 - val_loss: 4.2276 - val_accuracy: 0.0896\n",
+      "Epoch 4771/5000\n",
+      "919/919 - 3s - loss: 1.1828 - accuracy: 0.6300 - val_loss: 4.2304 - val_accuracy: 0.0889\n",
+      "Epoch 4772/5000\n",
+      "919/919 - 3s - loss: 1.1749 - accuracy: 0.6347 - val_loss: 4.2216 - val_accuracy: 0.0890\n",
+      "Epoch 4773/5000\n",
+      "919/919 - 3s - loss: 1.1785 - accuracy: 0.6330 - val_loss: 4.2063 - val_accuracy: 0.0892\n",
+      "Epoch 4774/5000\n",
+      "919/919 - 3s - loss: 1.1805 - accuracy: 0.6348 - val_loss: 4.2233 - val_accuracy: 0.0892\n",
+      "Epoch 4775/5000\n",
+      "919/919 - 3s - loss: 1.1551 - accuracy: 0.6365 - val_loss: 4.2327 - val_accuracy: 0.0895\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4776/5000\n",
+      "919/919 - 3s - loss: 1.1824 - accuracy: 0.6339 - val_loss: 4.2386 - val_accuracy: 0.0892\n",
+      "Epoch 4777/5000\n",
+      "919/919 - 3s - loss: 1.2046 - accuracy: 0.6348 - val_loss: 4.2341 - val_accuracy: 0.0884\n",
+      "Epoch 4778/5000\n",
+      "919/919 - 3s - loss: 1.1860 - accuracy: 0.6295 - val_loss: 4.2288 - val_accuracy: 0.0891\n",
+      "Epoch 4779/5000\n",
+      "919/919 - 3s - loss: 1.1778 - accuracy: 0.6318 - val_loss: 4.2214 - val_accuracy: 0.0895\n",
+      "Epoch 4780/5000\n",
+      "919/919 - 3s - loss: 1.1884 - accuracy: 0.6344 - val_loss: 4.2161 - val_accuracy: 0.0895\n",
+      "Epoch 4781/5000\n",
+      "919/919 - 3s - loss: 1.1833 - accuracy: 0.6335 - val_loss: 4.2175 - val_accuracy: 0.0897\n",
+      "Epoch 4782/5000\n",
+      "919/919 - 3s - loss: 1.1796 - accuracy: 0.6332 - val_loss: 4.2303 - val_accuracy: 0.0894\n",
+      "Epoch 4783/5000\n",
+      "919/919 - 3s - loss: 1.1858 - accuracy: 0.6357 - val_loss: 4.2362 - val_accuracy: 0.0893\n",
+      "Epoch 4784/5000\n",
+      "919/919 - 3s - loss: 1.1840 - accuracy: 0.6350 - val_loss: 4.2291 - val_accuracy: 0.0894\n",
+      "Epoch 4785/5000\n",
+      "919/919 - 3s - loss: 1.1688 - accuracy: 0.6348 - val_loss: 4.2182 - val_accuracy: 0.0892\n",
+      "Epoch 4786/5000\n",
+      "919/919 - 3s - loss: 1.1694 - accuracy: 0.6348 - val_loss: 4.2203 - val_accuracy: 0.0899\n",
+      "Epoch 4787/5000\n",
+      "919/919 - 3s - loss: 1.1789 - accuracy: 0.6346 - val_loss: 4.2164 - val_accuracy: 0.0898\n",
+      "Epoch 4788/5000\n",
+      "919/919 - 3s - loss: 1.1752 - accuracy: 0.6338 - val_loss: 4.2199 - val_accuracy: 0.0901\n",
+      "Epoch 4789/5000\n",
+      "919/919 - 3s - loss: 1.1838 - accuracy: 0.6330 - val_loss: 4.2017 - val_accuracy: 0.0889\n",
+      "Epoch 4790/5000\n",
+      "919/919 - 3s - loss: 1.2107 - accuracy: 0.6298 - val_loss: 4.1935 - val_accuracy: 0.0895\n",
+      "Epoch 4791/5000\n",
+      "919/919 - 3s - loss: 1.1660 - accuracy: 0.6337 - val_loss: 4.2136 - val_accuracy: 0.0901\n",
+      "Epoch 4792/5000\n",
+      "919/919 - 3s - loss: 1.1772 - accuracy: 0.6337 - val_loss: 4.2119 - val_accuracy: 0.0899\n",
+      "Epoch 4793/5000\n",
+      "919/919 - 3s - loss: 1.1775 - accuracy: 0.6307 - val_loss: 4.2098 - val_accuracy: 0.0899\n",
+      "Epoch 4794/5000\n",
+      "919/919 - 3s - loss: 1.2242 - accuracy: 0.6297 - val_loss: 4.2069 - val_accuracy: 0.0895\n",
+      "Epoch 4795/5000\n",
+      "919/919 - 3s - loss: 1.1595 - accuracy: 0.6349 - val_loss: 4.2034 - val_accuracy: 0.0898\n",
+      "Epoch 4796/5000\n",
+      "919/919 - 3s - loss: 1.1727 - accuracy: 0.6331 - val_loss: 4.2134 - val_accuracy: 0.0901\n",
+      "Epoch 4797/5000\n",
+      "919/919 - 3s - loss: 1.1850 - accuracy: 0.6318 - val_loss: 4.2195 - val_accuracy: 0.0905\n",
+      "Epoch 4798/5000\n",
+      "919/919 - 3s - loss: 1.2420 - accuracy: 0.6341 - val_loss: 4.2347 - val_accuracy: 0.0899\n",
+      "Epoch 4799/5000\n",
+      "919/919 - 3s - loss: 1.1900 - accuracy: 0.6297 - val_loss: 4.2360 - val_accuracy: 0.0903\n",
+      "Epoch 4800/5000\n",
+      "919/919 - 3s - loss: 1.1675 - accuracy: 0.6361 - val_loss: 4.2282 - val_accuracy: 0.0908\n",
+      "Epoch 4801/5000\n",
+      "919/919 - 3s - loss: 1.1669 - accuracy: 0.6325 - val_loss: 4.2425 - val_accuracy: 0.0902\n",
+      "Epoch 4802/5000\n",
+      "919/919 - 3s - loss: 1.1723 - accuracy: 0.6329 - val_loss: 4.2277 - val_accuracy: 0.0892\n",
+      "Epoch 4803/5000\n",
+      "919/919 - 3s - loss: 1.1698 - accuracy: 0.6341 - val_loss: 4.2443 - val_accuracy: 0.0891\n",
+      "Epoch 4804/5000\n",
+      "919/919 - 3s - loss: 1.1708 - accuracy: 0.6372 - val_loss: 4.2481 - val_accuracy: 0.0899\n",
+      "Epoch 4805/5000\n",
+      "919/919 - 3s - loss: 1.1589 - accuracy: 0.6373 - val_loss: 4.2391 - val_accuracy: 0.0886\n",
+      "Epoch 4806/5000\n",
+      "919/919 - 3s - loss: 1.1642 - accuracy: 0.6375 - val_loss: 4.2300 - val_accuracy: 0.0900\n",
+      "Epoch 4807/5000\n",
+      "919/919 - 3s - loss: 1.2101 - accuracy: 0.6354 - val_loss: 4.2317 - val_accuracy: 0.0897\n",
+      "Epoch 4808/5000\n",
+      "919/919 - 3s - loss: 1.1736 - accuracy: 0.6320 - val_loss: 4.2276 - val_accuracy: 0.0893\n",
+      "Epoch 4809/5000\n",
+      "919/919 - 3s - loss: 1.2076 - accuracy: 0.6344 - val_loss: 4.2064 - val_accuracy: 0.0898\n",
+      "Epoch 4810/5000\n",
+      "919/919 - 3s - loss: 1.1646 - accuracy: 0.6370 - val_loss: 4.2264 - val_accuracy: 0.0904\n",
+      "Epoch 4811/5000\n",
+      "919/919 - 3s - loss: 1.1758 - accuracy: 0.6340 - val_loss: 4.2258 - val_accuracy: 0.0903\n",
+      "Epoch 4812/5000\n",
+      "919/919 - 3s - loss: 1.1760 - accuracy: 0.6333 - val_loss: 4.2244 - val_accuracy: 0.0899\n",
+      "Epoch 4813/5000\n",
+      "919/919 - 3s - loss: 1.1818 - accuracy: 0.6343 - val_loss: 4.2195 - val_accuracy: 0.0905\n",
+      "Epoch 4814/5000\n",
+      "919/919 - 3s - loss: 1.1714 - accuracy: 0.6375 - val_loss: 4.2273 - val_accuracy: 0.0899\n",
+      "Epoch 4815/5000\n",
+      "919/919 - 3s - loss: 1.1740 - accuracy: 0.6362 - val_loss: 4.2426 - val_accuracy: 0.0906\n",
+      "Epoch 4816/5000\n",
+      "919/919 - 3s - loss: 1.2316 - accuracy: 0.6386 - val_loss: 4.2328 - val_accuracy: 0.0910\n",
+      "Epoch 4817/5000\n",
+      "919/919 - 3s - loss: 1.1636 - accuracy: 0.6355 - val_loss: 4.2417 - val_accuracy: 0.0913\n",
+      "Epoch 4818/5000\n",
+      "919/919 - 3s - loss: 1.1657 - accuracy: 0.6339 - val_loss: 4.2593 - val_accuracy: 0.0905\n",
+      "Epoch 4819/5000\n",
+      "919/919 - 3s - loss: 1.1843 - accuracy: 0.6314 - val_loss: 4.2408 - val_accuracy: 0.0906\n",
+      "Epoch 4820/5000\n",
+      "919/919 - 3s - loss: 1.1884 - accuracy: 0.6339 - val_loss: 4.2399 - val_accuracy: 0.0904\n",
+      "Epoch 4821/5000\n",
+      "919/919 - 3s - loss: 1.1847 - accuracy: 0.6371 - val_loss: 4.2451 - val_accuracy: 0.0903\n",
+      "Epoch 4822/5000\n",
+      "919/919 - 3s - loss: 1.2601 - accuracy: 0.6310 - val_loss: 4.2324 - val_accuracy: 0.0901\n",
+      "Epoch 4823/5000\n",
+      "919/919 - 3s - loss: 1.2947 - accuracy: 0.6346 - val_loss: 4.2347 - val_accuracy: 0.0899\n",
+      "Epoch 4824/5000\n",
+      "919/919 - 3s - loss: 1.1839 - accuracy: 0.6342 - val_loss: 4.2227 - val_accuracy: 0.0901\n",
+      "Epoch 4825/5000\n",
+      "919/919 - 3s - loss: 1.1967 - accuracy: 0.6371 - val_loss: 4.2388 - val_accuracy: 0.0906\n",
+      "Epoch 4826/5000\n",
+      "919/919 - 3s - loss: 1.2166 - accuracy: 0.6354 - val_loss: 4.2289 - val_accuracy: 0.0908\n",
+      "Epoch 4827/5000\n",
+      "919/919 - 3s - loss: 1.1617 - accuracy: 0.6338 - val_loss: 4.2184 - val_accuracy: 0.0907\n",
+      "Epoch 4828/5000\n",
+      "919/919 - 3s - loss: 1.2059 - accuracy: 0.6348 - val_loss: 4.2255 - val_accuracy: 0.0906\n",
+      "Epoch 4829/5000\n",
+      "919/919 - 3s - loss: 1.1802 - accuracy: 0.6322 - val_loss: 4.2274 - val_accuracy: 0.0901\n",
+      "Epoch 4830/5000\n",
+      "919/919 - 3s - loss: 1.2252 - accuracy: 0.6368 - val_loss: 4.2286 - val_accuracy: 0.0902\n",
+      "Epoch 4831/5000\n",
+      "919/919 - 3s - loss: 1.1805 - accuracy: 0.6335 - val_loss: 4.2246 - val_accuracy: 0.0900\n",
+      "Epoch 4832/5000\n",
+      "919/919 - 3s - loss: 1.1733 - accuracy: 0.6364 - val_loss: 4.2251 - val_accuracy: 0.0895\n",
+      "Epoch 4833/5000\n",
+      "919/919 - 3s - loss: 1.1622 - accuracy: 0.6339 - val_loss: 4.2263 - val_accuracy: 0.0892\n",
+      "Epoch 4834/5000\n",
+      "919/919 - 3s - loss: 1.1657 - accuracy: 0.6339 - val_loss: 4.2349 - val_accuracy: 0.0894\n",
+      "Epoch 4835/5000\n",
+      "919/919 - 3s - loss: 1.1638 - accuracy: 0.6387 - val_loss: 4.2444 - val_accuracy: 0.0901\n",
+      "Epoch 4836/5000\n",
+      "919/919 - 3s - loss: 1.1755 - accuracy: 0.6363 - val_loss: 4.2342 - val_accuracy: 0.0899\n",
+      "Epoch 4837/5000\n",
+      "919/919 - 3s - loss: 1.1590 - accuracy: 0.6377 - val_loss: 4.2357 - val_accuracy: 0.0895\n",
+      "Epoch 4838/5000\n",
+      "919/919 - 3s - loss: 1.1908 - accuracy: 0.6312 - val_loss: 4.2417 - val_accuracy: 0.0893\n",
+      "Epoch 4839/5000\n",
+      "919/919 - 3s - loss: 1.1677 - accuracy: 0.6324 - val_loss: 4.2413 - val_accuracy: 0.0893\n",
+      "Epoch 4840/5000\n",
+      "919/919 - 3s - loss: 1.1687 - accuracy: 0.6371 - val_loss: 4.2510 - val_accuracy: 0.0889\n",
+      "Epoch 4841/5000\n",
+      "919/919 - 3s - loss: 1.1760 - accuracy: 0.6352 - val_loss: 4.2412 - val_accuracy: 0.0893\n",
+      "Epoch 4842/5000\n",
+      "919/919 - 3s - loss: 1.1793 - accuracy: 0.6287 - val_loss: 4.2308 - val_accuracy: 0.0891\n",
+      "Epoch 4843/5000\n",
+      "919/919 - 3s - loss: 1.2273 - accuracy: 0.6305 - val_loss: 4.2230 - val_accuracy: 0.0896\n",
+      "Epoch 4844/5000\n",
+      "919/919 - 3s - loss: 1.1693 - accuracy: 0.6369 - val_loss: 4.2356 - val_accuracy: 0.0898\n",
+      "Epoch 4845/5000\n",
+      "919/919 - 3s - loss: 1.1594 - accuracy: 0.6371 - val_loss: 4.2337 - val_accuracy: 0.0897\n",
+      "Epoch 4846/5000\n",
+      "919/919 - 3s - loss: 1.1898 - accuracy: 0.6310 - val_loss: 4.2372 - val_accuracy: 0.0901\n",
+      "Epoch 4847/5000\n",
+      "919/919 - 3s - loss: 1.1839 - accuracy: 0.6346 - val_loss: 4.2347 - val_accuracy: 0.0898\n",
+      "Epoch 4848/5000\n",
+      "919/919 - 3s - loss: 1.1682 - accuracy: 0.6390 - val_loss: 4.2330 - val_accuracy: 0.0899\n",
+      "Epoch 4849/5000\n",
+      "919/919 - 3s - loss: 1.1671 - accuracy: 0.6376 - val_loss: 4.2321 - val_accuracy: 0.0911\n",
+      "Epoch 4850/5000\n",
+      "919/919 - 3s - loss: 1.2565 - accuracy: 0.6386 - val_loss: 4.2352 - val_accuracy: 0.0913\n",
+      "Epoch 4851/5000\n",
+      "919/919 - 3s - loss: 1.3319 - accuracy: 0.6339 - val_loss: 4.2404 - val_accuracy: 0.0904\n",
+      "Epoch 4852/5000\n",
+      "919/919 - 3s - loss: 1.1663 - accuracy: 0.6368 - val_loss: 4.2540 - val_accuracy: 0.0897\n",
+      "Epoch 4853/5000\n",
+      "919/919 - 3s - loss: 1.1545 - accuracy: 0.6390 - val_loss: 4.2466 - val_accuracy: 0.0902\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4854/5000\n",
+      "919/919 - 3s - loss: 1.1814 - accuracy: 0.6370 - val_loss: 4.2588 - val_accuracy: 0.0906\n",
+      "Epoch 4855/5000\n",
+      "919/919 - 3s - loss: 1.1656 - accuracy: 0.6377 - val_loss: 4.2536 - val_accuracy: 0.0892\n",
+      "Epoch 4856/5000\n",
+      "919/919 - 3s - loss: 1.2657 - accuracy: 0.6369 - val_loss: 4.2708 - val_accuracy: 0.0903\n",
+      "Epoch 4857/5000\n",
+      "919/919 - 3s - loss: 1.1679 - accuracy: 0.6352 - val_loss: 4.2569 - val_accuracy: 0.0901\n",
+      "Epoch 4858/5000\n",
+      "919/919 - 3s - loss: 1.1703 - accuracy: 0.6348 - val_loss: 4.2534 - val_accuracy: 0.0901\n",
+      "Epoch 4859/5000\n",
+      "919/919 - 3s - loss: 1.1756 - accuracy: 0.6345 - val_loss: 4.2424 - val_accuracy: 0.0903\n",
+      "Epoch 4860/5000\n",
+      "919/919 - 3s - loss: 1.1568 - accuracy: 0.6353 - val_loss: 4.2299 - val_accuracy: 0.0907\n",
+      "Epoch 4861/5000\n",
+      "919/919 - 3s - loss: 1.1719 - accuracy: 0.6320 - val_loss: 4.2411 - val_accuracy: 0.0909\n",
+      "Epoch 4862/5000\n",
+      "919/919 - 3s - loss: 1.1793 - accuracy: 0.6346 - val_loss: 4.2365 - val_accuracy: 0.0909\n",
+      "Epoch 4863/5000\n",
+      "919/919 - 3s - loss: 1.1823 - accuracy: 0.6356 - val_loss: 4.2445 - val_accuracy: 0.0907\n",
+      "Epoch 4864/5000\n",
+      "919/919 - 3s - loss: 1.1793 - accuracy: 0.6390 - val_loss: 4.2548 - val_accuracy: 0.0907\n",
+      "Epoch 4865/5000\n",
+      "919/919 - 3s - loss: 1.1983 - accuracy: 0.6377 - val_loss: 4.2645 - val_accuracy: 0.0900\n",
+      "Epoch 4866/5000\n",
+      "919/919 - 3s - loss: 1.2588 - accuracy: 0.6388 - val_loss: 4.2560 - val_accuracy: 0.0902\n",
+      "Epoch 4867/5000\n",
+      "919/919 - 3s - loss: 1.1765 - accuracy: 0.6344 - val_loss: 4.2675 - val_accuracy: 0.0895\n",
+      "Epoch 4868/5000\n",
+      "919/919 - 3s - loss: 1.1588 - accuracy: 0.6392 - val_loss: 4.2630 - val_accuracy: 0.0896\n",
+      "Epoch 4869/5000\n",
+      "919/919 - 3s - loss: 1.1712 - accuracy: 0.6369 - val_loss: 4.2617 - val_accuracy: 0.0908\n",
+      "Epoch 4870/5000\n",
+      "919/919 - 3s - loss: 1.1753 - accuracy: 0.6357 - val_loss: 4.2494 - val_accuracy: 0.0904\n",
+      "Epoch 4871/5000\n",
+      "919/919 - 3s - loss: 1.1641 - accuracy: 0.6369 - val_loss: 4.2553 - val_accuracy: 0.0904\n",
+      "Epoch 4872/5000\n",
+      "919/919 - 3s - loss: 1.1466 - accuracy: 0.6369 - val_loss: 4.2433 - val_accuracy: 0.0904\n",
+      "INFO:tensorflow:Assets written to: ./goat.weights/assets\n",
+      "Epoch 4873/5000\n",
+      "919/919 - 3s - loss: 1.1863 - accuracy: 0.6375 - val_loss: 4.2350 - val_accuracy: 0.0902\n",
+      "Epoch 4874/5000\n",
+      "919/919 - 3s - loss: 1.2023 - accuracy: 0.6366 - val_loss: 4.2303 - val_accuracy: 0.0902\n",
+      "Epoch 4875/5000\n",
+      "919/919 - 3s - loss: 1.1609 - accuracy: 0.6393 - val_loss: 4.2322 - val_accuracy: 0.0907\n",
+      "Epoch 4876/5000\n",
+      "919/919 - 3s - loss: 1.1597 - accuracy: 0.6414 - val_loss: 4.2525 - val_accuracy: 0.0900\n",
+      "Epoch 4877/5000\n",
+      "919/919 - 3s - loss: 1.1683 - accuracy: 0.6361 - val_loss: 4.2615 - val_accuracy: 0.0901\n",
+      "Epoch 4878/5000\n",
+      "919/919 - 3s - loss: 1.1629 - accuracy: 0.6394 - val_loss: 4.2625 - val_accuracy: 0.0904\n",
+      "Epoch 4879/5000\n",
+      "919/919 - 3s - loss: 1.1592 - accuracy: 0.6345 - val_loss: 4.2586 - val_accuracy: 0.0910\n",
+      "Epoch 4880/5000\n",
+      "919/919 - 3s - loss: 1.4077 - accuracy: 0.6370 - val_loss: 4.2454 - val_accuracy: 0.0906\n",
+      "Epoch 4881/5000\n",
+      "919/919 - 3s - loss: 1.1708 - accuracy: 0.6361 - val_loss: 4.2294 - val_accuracy: 0.0907\n",
+      "Epoch 4882/5000\n",
+      "919/919 - 3s - loss: 1.1642 - accuracy: 0.6359 - val_loss: 4.2308 - val_accuracy: 0.0894\n",
+      "Epoch 4883/5000\n",
+      "919/919 - 3s - loss: 1.1632 - accuracy: 0.6384 - val_loss: 4.2369 - val_accuracy: 0.0899\n",
+      "Epoch 4884/5000\n",
+      "919/919 - 3s - loss: 1.1728 - accuracy: 0.6355 - val_loss: 4.2468 - val_accuracy: 0.0904\n",
+      "Epoch 4885/5000\n",
+      "919/919 - 3s - loss: 1.1581 - accuracy: 0.6376 - val_loss: 4.2423 - val_accuracy: 0.0902\n",
+      "Epoch 4886/5000\n",
+      "919/919 - 3s - loss: 1.1774 - accuracy: 0.6321 - val_loss: 4.2429 - val_accuracy: 0.0899\n",
+      "Epoch 4887/5000\n",
+      "919/919 - 3s - loss: 1.1600 - accuracy: 0.6401 - val_loss: 4.2393 - val_accuracy: 0.0898\n",
+      "Epoch 4888/5000\n",
+      "919/919 - 3s - loss: 1.1725 - accuracy: 0.6365 - val_loss: 4.2617 - val_accuracy: 0.0900\n",
+      "Epoch 4889/5000\n",
+      "919/919 - 3s - loss: 1.1589 - accuracy: 0.6384 - val_loss: 4.2629 - val_accuracy: 0.0900\n",
+      "Epoch 4890/5000\n",
+      "919/919 - 3s - loss: 1.1697 - accuracy: 0.6346 - val_loss: 4.2576 - val_accuracy: 0.0904\n",
+      "Epoch 4891/5000\n",
+      "919/919 - 3s - loss: 1.1873 - accuracy: 0.6358 - val_loss: 4.2527 - val_accuracy: 0.0893\n",
+      "Epoch 4892/5000\n",
+      "919/919 - 3s - loss: 1.1714 - accuracy: 0.6349 - val_loss: 4.2559 - val_accuracy: 0.0889\n",
+      "Epoch 4893/5000\n",
+      "919/919 - 3s - loss: 1.1489 - accuracy: 0.6373 - val_loss: 4.2466 - val_accuracy: 0.0888\n",
+      "Epoch 4894/5000\n",
+      "919/919 - 3s - loss: 1.1631 - accuracy: 0.6386 - val_loss: 4.2492 - val_accuracy: 0.0889\n",
+      "Epoch 4895/5000\n",
+      "919/919 - 3s - loss: 1.1688 - accuracy: 0.6397 - val_loss: 4.2396 - val_accuracy: 0.0901\n",
+      "Epoch 4896/5000\n",
+      "919/919 - 3s - loss: 1.1676 - accuracy: 0.6380 - val_loss: 4.2501 - val_accuracy: 0.0899\n",
+      "Epoch 4897/5000\n",
+      "919/919 - 3s - loss: 1.1592 - accuracy: 0.6403 - val_loss: 4.2361 - val_accuracy: 0.0902\n",
+      "Epoch 4898/5000\n",
+      "919/919 - 3s - loss: 1.1704 - accuracy: 0.6335 - val_loss: 4.2320 - val_accuracy: 0.0899\n",
+      "Epoch 4899/5000\n",
+      "919/919 - 3s - loss: 1.2328 - accuracy: 0.6334 - val_loss: 4.2417 - val_accuracy: 0.0899\n",
+      "Epoch 4900/5000\n",
+      "919/919 - 3s - loss: 1.1687 - accuracy: 0.6350 - val_loss: 4.2363 - val_accuracy: 0.0897\n",
+      "Epoch 4901/5000\n",
+      "919/919 - 3s - loss: 1.1856 - accuracy: 0.6375 - val_loss: 4.2295 - val_accuracy: 0.0906\n",
+      "Epoch 4902/5000\n",
+      "919/919 - 3s - loss: 1.1860 - accuracy: 0.6361 - val_loss: 4.2322 - val_accuracy: 0.0912\n",
+      "Epoch 4903/5000\n",
+      "919/919 - 3s - loss: 1.1745 - accuracy: 0.6357 - val_loss: 4.2233 - val_accuracy: 0.0910\n",
+      "Epoch 4904/5000\n",
+      "919/919 - 3s - loss: 1.1699 - accuracy: 0.6371 - val_loss: 4.2344 - val_accuracy: 0.0912\n",
+      "Epoch 4905/5000\n",
+      "919/919 - 3s - loss: 1.1693 - accuracy: 0.6363 - val_loss: 4.2500 - val_accuracy: 0.0911\n",
+      "Epoch 4906/5000\n",
+      "919/919 - 3s - loss: 1.2002 - accuracy: 0.6346 - val_loss: 4.2602 - val_accuracy: 0.0901\n",
+      "Epoch 4907/5000\n",
+      "919/919 - 3s - loss: 1.1722 - accuracy: 0.6365 - val_loss: 4.2613 - val_accuracy: 0.0910\n",
+      "Epoch 4908/5000\n",
+      "919/919 - 3s - loss: 1.1677 - accuracy: 0.6399 - val_loss: 4.2500 - val_accuracy: 0.0909\n",
+      "Epoch 4909/5000\n",
+      "919/919 - 3s - loss: 1.1665 - accuracy: 0.6374 - val_loss: 4.2405 - val_accuracy: 0.0909\n",
+      "Epoch 4910/5000\n",
+      "919/919 - 3s - loss: 1.1639 - accuracy: 0.6357 - val_loss: 4.2452 - val_accuracy: 0.0902\n",
+      "Epoch 4911/5000\n",
+      "919/919 - 3s - loss: 1.1813 - accuracy: 0.6322 - val_loss: 4.2472 - val_accuracy: 0.0901\n",
+      "Epoch 4912/5000\n",
+      "919/919 - 3s - loss: 1.1627 - accuracy: 0.6397 - val_loss: 4.2600 - val_accuracy: 0.0899\n",
+      "Epoch 4913/5000\n",
+      "919/919 - 3s - loss: 1.1698 - accuracy: 0.6371 - val_loss: 4.2417 - val_accuracy: 0.0899\n",
+      "Epoch 4914/5000\n",
+      "919/919 - 3s - loss: 1.1791 - accuracy: 0.6388 - val_loss: 4.2324 - val_accuracy: 0.0906\n",
+      "Epoch 4915/5000\n",
+      "919/919 - 3s - loss: 1.1684 - accuracy: 0.6390 - val_loss: 4.2246 - val_accuracy: 0.0905\n",
+      "Epoch 4916/5000\n",
+      "919/919 - 3s - loss: 1.1693 - accuracy: 0.6346 - val_loss: 4.2495 - val_accuracy: 0.0893\n",
+      "Epoch 4917/5000\n",
+      "919/919 - 3s - loss: 1.1666 - accuracy: 0.6351 - val_loss: 4.2491 - val_accuracy: 0.0897\n",
+      "Epoch 4918/5000\n",
+      "919/919 - 3s - loss: 1.1592 - accuracy: 0.6361 - val_loss: 4.2440 - val_accuracy: 0.0905\n",
+      "Epoch 4919/5000\n",
+      "919/919 - 3s - loss: 1.1724 - accuracy: 0.6375 - val_loss: 4.2388 - val_accuracy: 0.0905\n",
+      "Epoch 4920/5000\n",
+      "919/919 - 3s - loss: 1.1734 - accuracy: 0.6369 - val_loss: 4.2558 - val_accuracy: 0.0906\n",
+      "Epoch 4921/5000\n",
+      "919/919 - 3s - loss: 1.4577 - accuracy: 0.6378 - val_loss: 4.2657 - val_accuracy: 0.0904\n",
+      "Epoch 4922/5000\n",
+      "919/919 - 3s - loss: 1.1575 - accuracy: 0.6396 - val_loss: 4.2764 - val_accuracy: 0.0892\n",
+      "Epoch 4923/5000\n",
+      "919/919 - 3s - loss: 1.1656 - accuracy: 0.6348 - val_loss: 4.2836 - val_accuracy: 0.0897\n",
+      "Epoch 4924/5000\n",
+      "919/919 - 3s - loss: 1.1645 - accuracy: 0.6397 - val_loss: 4.2865 - val_accuracy: 0.0892\n",
+      "Epoch 4925/5000\n",
+      "919/919 - 3s - loss: 1.1706 - accuracy: 0.6379 - val_loss: 4.2684 - val_accuracy: 0.0904\n",
+      "Epoch 4926/5000\n",
+      "919/919 - 3s - loss: 1.1855 - accuracy: 0.6359 - val_loss: 4.2712 - val_accuracy: 0.0900\n",
+      "Epoch 4927/5000\n",
+      "919/919 - 3s - loss: 1.1631 - accuracy: 0.6362 - val_loss: 4.2785 - val_accuracy: 0.0893\n",
+      "Epoch 4928/5000\n",
+      "919/919 - 3s - loss: 1.1728 - accuracy: 0.6329 - val_loss: 4.2653 - val_accuracy: 0.0888\n",
+      "Epoch 4929/5000\n",
+      "919/919 - 3s - loss: 1.3305 - accuracy: 0.6390 - val_loss: 4.2540 - val_accuracy: 0.0896\n",
+      "Epoch 4930/5000\n",
+      "919/919 - 3s - loss: 1.1661 - accuracy: 0.6382 - val_loss: 4.2680 - val_accuracy: 0.0892\n",
+      "Epoch 4931/5000\n",
+      "919/919 - 3s - loss: 1.1642 - accuracy: 0.6362 - val_loss: 4.2530 - val_accuracy: 0.0893\n",
+      "Epoch 4932/5000\n",
+      "919/919 - 3s - loss: 1.1639 - accuracy: 0.6380 - val_loss: 4.2461 - val_accuracy: 0.0897\n",
+      "Epoch 4933/5000\n",
+      "919/919 - 3s - loss: 1.1645 - accuracy: 0.6360 - val_loss: 4.2377 - val_accuracy: 0.0906\n",
+      "Epoch 4934/5000\n",
+      "919/919 - 3s - loss: 1.2204 - accuracy: 0.6373 - val_loss: 4.2478 - val_accuracy: 0.0898\n",
+      "Epoch 4935/5000\n",
+      "919/919 - 3s - loss: 1.1613 - accuracy: 0.6359 - val_loss: 4.2616 - val_accuracy: 0.0904\n",
+      "Epoch 4936/5000\n",
+      "919/919 - 3s - loss: 1.1548 - accuracy: 0.6402 - val_loss: 4.2579 - val_accuracy: 0.0897\n",
+      "Epoch 4937/5000\n",
+      "919/919 - 3s - loss: 1.1585 - accuracy: 0.6376 - val_loss: 4.2493 - val_accuracy: 0.0898\n",
+      "Epoch 4938/5000\n",
+      "919/919 - 3s - loss: 1.2136 - accuracy: 0.6350 - val_loss: 4.2494 - val_accuracy: 0.0894\n",
+      "Epoch 4939/5000\n",
+      "919/919 - 3s - loss: 1.1740 - accuracy: 0.6345 - val_loss: 4.2606 - val_accuracy: 0.0882\n",
+      "Epoch 4940/5000\n",
+      "919/919 - 3s - loss: 1.1601 - accuracy: 0.6399 - val_loss: 4.2622 - val_accuracy: 0.0892\n",
+      "Epoch 4941/5000\n",
+      "919/919 - 3s - loss: 1.1772 - accuracy: 0.6339 - val_loss: 4.2590 - val_accuracy: 0.0897\n",
+      "Epoch 4942/5000\n",
+      "919/919 - 3s - loss: 1.1846 - accuracy: 0.6403 - val_loss: 4.2710 - val_accuracy: 0.0890\n",
+      "Epoch 4943/5000\n",
+      "919/919 - 3s - loss: 1.1766 - accuracy: 0.6369 - val_loss: 4.2443 - val_accuracy: 0.0898\n",
+      "Epoch 4944/5000\n",
+      "919/919 - 3s - loss: 1.1612 - accuracy: 0.6395 - val_loss: 4.2260 - val_accuracy: 0.0898\n",
+      "Epoch 4945/5000\n",
+      "919/919 - 3s - loss: 1.1579 - accuracy: 0.6371 - val_loss: 4.2536 - val_accuracy: 0.0894\n",
+      "Epoch 4946/5000\n",
+      "919/919 - 3s - loss: 1.2716 - accuracy: 0.6364 - val_loss: 4.2691 - val_accuracy: 0.0894\n",
+      "Epoch 4947/5000\n",
+      "919/919 - 3s - loss: 1.1634 - accuracy: 0.6376 - val_loss: 4.2599 - val_accuracy: 0.0894\n",
+      "Epoch 4948/5000\n",
+      "919/919 - 3s - loss: 1.1699 - accuracy: 0.6360 - val_loss: 4.2640 - val_accuracy: 0.0897\n",
+      "Epoch 4949/5000\n",
+      "919/919 - 3s - loss: 1.4199 - accuracy: 0.6424 - val_loss: 4.2694 - val_accuracy: 0.0900\n",
+      "Epoch 4950/5000\n",
+      "919/919 - 3s - loss: 1.1697 - accuracy: 0.6390 - val_loss: 4.2684 - val_accuracy: 0.0891\n",
+      "Epoch 4951/5000\n",
+      "919/919 - 3s - loss: 1.1656 - accuracy: 0.6351 - val_loss: 4.2566 - val_accuracy: 0.0905\n",
+      "Epoch 4952/5000\n",
+      "919/919 - 3s - loss: 1.1538 - accuracy: 0.6403 - val_loss: 4.2657 - val_accuracy: 0.0904\n",
+      "Epoch 4953/5000\n",
+      "919/919 - 3s - loss: 1.1937 - accuracy: 0.6378 - val_loss: 4.2660 - val_accuracy: 0.0905\n",
+      "Epoch 4954/5000\n",
+      "919/919 - 3s - loss: 1.1638 - accuracy: 0.6390 - val_loss: 4.2698 - val_accuracy: 0.0899\n",
+      "Epoch 4955/5000\n",
+      "919/919 - 3s - loss: 1.1617 - accuracy: 0.6384 - val_loss: 4.2832 - val_accuracy: 0.0898\n",
+      "Epoch 4956/5000\n",
+      "919/919 - 3s - loss: 1.1727 - accuracy: 0.6414 - val_loss: 4.2905 - val_accuracy: 0.0904\n",
+      "Epoch 4957/5000\n",
+      "919/919 - 3s - loss: 1.2211 - accuracy: 0.6382 - val_loss: 4.2759 - val_accuracy: 0.0894\n",
+      "Epoch 4958/5000\n",
+      "919/919 - 3s - loss: 1.1960 - accuracy: 0.6380 - val_loss: 4.2531 - val_accuracy: 0.0902\n",
+      "Epoch 4959/5000\n",
+      "919/919 - 3s - loss: 1.1663 - accuracy: 0.6376 - val_loss: 4.2737 - val_accuracy: 0.0898\n",
+      "Epoch 4960/5000\n",
+      "919/919 - 3s - loss: 1.1626 - accuracy: 0.6373 - val_loss: 4.2671 - val_accuracy: 0.0898\n",
+      "Epoch 4961/5000\n",
+      "919/919 - 3s - loss: 1.1551 - accuracy: 0.6367 - val_loss: 4.2851 - val_accuracy: 0.0895\n",
+      "Epoch 4962/5000\n",
+      "919/919 - 3s - loss: 1.1533 - accuracy: 0.6407 - val_loss: 4.2829 - val_accuracy: 0.0896\n",
+      "Epoch 4963/5000\n",
+      "919/919 - 3s - loss: 1.1559 - accuracy: 0.6371 - val_loss: 4.2616 - val_accuracy: 0.0892\n",
+      "Epoch 4964/5000\n",
+      "919/919 - 3s - loss: 1.1892 - accuracy: 0.6352 - val_loss: 4.2655 - val_accuracy: 0.0898\n",
+      "Epoch 4965/5000\n",
+      "919/919 - 3s - loss: 1.1486 - accuracy: 0.6397 - val_loss: 4.2775 - val_accuracy: 0.0897\n",
+      "Epoch 4966/5000\n",
+      "919/919 - 3s - loss: 1.1730 - accuracy: 0.6358 - val_loss: 4.2747 - val_accuracy: 0.0901\n",
+      "Epoch 4967/5000\n",
+      "919/919 - 3s - loss: 1.1676 - accuracy: 0.6341 - val_loss: 4.2820 - val_accuracy: 0.0903\n",
+      "Epoch 4968/5000\n",
+      "919/919 - 3s - loss: 1.1503 - accuracy: 0.6414 - val_loss: 4.2752 - val_accuracy: 0.0908\n",
+      "Epoch 4969/5000\n",
+      "919/919 - 3s - loss: 1.1659 - accuracy: 0.6327 - val_loss: 4.2788 - val_accuracy: 0.0897\n",
+      "Epoch 4970/5000\n",
+      "919/919 - 3s - loss: 1.1776 - accuracy: 0.6389 - val_loss: 4.2703 - val_accuracy: 0.0897\n",
+      "Epoch 4971/5000\n",
+      "919/919 - 3s - loss: 1.1698 - accuracy: 0.6369 - val_loss: 4.2721 - val_accuracy: 0.0900\n",
+      "Epoch 4972/5000\n",
+      "919/919 - 3s - loss: 1.2219 - accuracy: 0.6428 - val_loss: 4.2830 - val_accuracy: 0.0894\n",
+      "Epoch 4973/5000\n",
+      "919/919 - 3s - loss: 1.1732 - accuracy: 0.6343 - val_loss: 4.2712 - val_accuracy: 0.0897\n",
+      "Epoch 4974/5000\n",
+      "919/919 - 3s - loss: 1.2688 - accuracy: 0.6378 - val_loss: 4.2908 - val_accuracy: 0.0896\n",
+      "Epoch 4975/5000\n",
+      "919/919 - 3s - loss: 1.1991 - accuracy: 0.6397 - val_loss: 4.2966 - val_accuracy: 0.0899\n",
+      "Epoch 4976/5000\n",
+      "919/919 - 3s - loss: 1.2552 - accuracy: 0.6389 - val_loss: 4.2833 - val_accuracy: 0.0894\n",
+      "Epoch 4977/5000\n",
+      "919/919 - 3s - loss: 1.1577 - accuracy: 0.6382 - val_loss: 4.2751 - val_accuracy: 0.0903\n",
+      "Epoch 4978/5000\n",
+      "919/919 - 3s - loss: 1.1616 - accuracy: 0.6382 - val_loss: 4.2929 - val_accuracy: 0.0888\n",
+      "Epoch 4979/5000\n",
+      "919/919 - 3s - loss: 1.1551 - accuracy: 0.6369 - val_loss: 4.2796 - val_accuracy: 0.0894\n",
+      "Epoch 4980/5000\n",
+      "919/919 - 3s - loss: 1.1895 - accuracy: 0.6332 - val_loss: 4.2895 - val_accuracy: 0.0892\n",
+      "Epoch 4981/5000\n",
+      "919/919 - 3s - loss: 1.1809 - accuracy: 0.6401 - val_loss: 4.2830 - val_accuracy: 0.0893\n",
+      "Epoch 4982/5000\n",
+      "919/919 - 3s - loss: 1.1820 - accuracy: 0.6290 - val_loss: 4.2841 - val_accuracy: 0.0897\n",
+      "Epoch 4983/5000\n",
+      "919/919 - 3s - loss: 1.1682 - accuracy: 0.6372 - val_loss: 4.2706 - val_accuracy: 0.0904\n",
+      "Epoch 4984/5000\n",
+      "919/919 - 3s - loss: 1.1769 - accuracy: 0.6387 - val_loss: 4.2721 - val_accuracy: 0.0910\n",
+      "Epoch 4985/5000\n",
+      "919/919 - 3s - loss: 1.1849 - accuracy: 0.6393 - val_loss: 4.2612 - val_accuracy: 0.0906\n",
+      "Epoch 4986/5000\n",
+      "919/919 - 3s - loss: 1.1709 - accuracy: 0.6392 - val_loss: 4.2398 - val_accuracy: 0.0903\n",
+      "Epoch 4987/5000\n",
+      "919/919 - 3s - loss: 1.1664 - accuracy: 0.6403 - val_loss: 4.2486 - val_accuracy: 0.0898\n",
+      "Epoch 4988/5000\n",
+      "919/919 - 3s - loss: 1.1846 - accuracy: 0.6398 - val_loss: 4.2491 - val_accuracy: 0.0901\n",
+      "Epoch 4989/5000\n",
+      "919/919 - 3s - loss: 1.1594 - accuracy: 0.6357 - val_loss: 4.2588 - val_accuracy: 0.0898\n",
+      "Epoch 4990/5000\n",
+      "919/919 - 3s - loss: 1.1726 - accuracy: 0.6370 - val_loss: 4.2574 - val_accuracy: 0.0904\n",
+      "Epoch 4991/5000\n",
+      "919/919 - 3s - loss: 1.1643 - accuracy: 0.6384 - val_loss: 4.2548 - val_accuracy: 0.0902\n",
+      "Epoch 4992/5000\n",
+      "919/919 - 3s - loss: 1.1513 - accuracy: 0.6428 - val_loss: 4.2553 - val_accuracy: 0.0901\n",
+      "Epoch 4993/5000\n",
+      "919/919 - 3s - loss: 1.1659 - accuracy: 0.6360 - val_loss: 4.2438 - val_accuracy: 0.0900\n",
+      "Epoch 4994/5000\n",
+      "919/919 - 3s - loss: 1.1625 - accuracy: 0.6369 - val_loss: 4.2601 - val_accuracy: 0.0905\n",
+      "Epoch 4995/5000\n",
+      "919/919 - 3s - loss: 1.1782 - accuracy: 0.6349 - val_loss: 4.2542 - val_accuracy: 0.0901\n",
+      "Epoch 4996/5000\n",
+      "919/919 - 3s - loss: 1.1734 - accuracy: 0.6395 - val_loss: 4.2734 - val_accuracy: 0.0897\n",
+      "Epoch 4997/5000\n",
+      "919/919 - 3s - loss: 1.2702 - accuracy: 0.6372 - val_loss: 4.2749 - val_accuracy: 0.0890\n",
+      "Epoch 4998/5000\n",
+      "919/919 - 3s - loss: 1.1727 - accuracy: 0.6368 - val_loss: 4.2697 - val_accuracy: 0.0886\n",
+      "Epoch 4999/5000\n",
+      "919/919 - 3s - loss: 1.1728 - accuracy: 0.6340 - val_loss: 4.2695 - val_accuracy: 0.0887\n",
+      "Epoch 5000/5000\n",
+      "919/919 - 3s - loss: 1.1608 - accuracy: 0.6394 - val_loss: 4.2721 - val_accuracy: 0.0887\n",
+      "CPU times: user 7h 2min 4s, sys: 35min 59s, total: 7h 38min 4s\n",
+      "Wall time: 4h 22min 41s\n"
+     ]
+    }
+   ],
+   "source": [
+    "%%time\n",
+    "\n",
+    "model, history = train_mlp(np.array(X_train), np.array(yy_train), np.array(X_test), np.array(yy_test))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "4b4a6ad0",
+   "metadata": {},
+   "source": [
+    "# Eval"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 25,
+   "id": "b7382ef6",
    "metadata": {},
    "outputs": [],
    "source": [
     "def predict(model, entry):\n",
     "    p_dict = dict()\n",
-    "    predictions = model.predict_classes(entry['data'])\n",
-    "    \n",
+    "    predictions = np.argmax(model.predict(entry['data']), axis=-1)\n",
     "    for p in predictions:\n",
     "        if p in p_dict:\n",
     "            p_dict[p] += 1\n",
     "        else:\n",
     "            p_dict[p] = 1\n",
     "    prediction = max(p_dict, key=p_dict.get)\n",
-    "    return prediction\n"
+    "    return prediction+1"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 24,
-   "id": "aae03bc6",
+   "execution_count": 26,
+   "id": "8c75712e",
    "metadata": {},
    "outputs": [
     {
-     "name": "stderr",
+     "name": "stdout",
      "output_type": "stream",
      "text": [
-      "/opt/jupyterhub/lib/python3.8/site-packages/tensorflow/python/keras/engine/sequential.py:455: UserWarning: `model.predict_classes()` is deprecated and will be removed after 2021-01-01. Please use instead:* `np.argmax(model.predict(x), axis=-1)`,   if your model does multi-class classification   (e.g. if it uses a `softmax` last-layer activation).* `(model.predict(x) > 0.5).astype(\"int32\")`,   if your model does binary classification   (e.g. if it uses a `sigmoid` last-layer activation).\n",
-      "  warnings.warn('`model.predict_classes()` is deprecated and '\n"
+      "CPU times: user 2.07 s, sys: 163 ms, total: 2.23 s\n",
+      "Wall time: 1.85 s\n"
      ]
     }
    ],
    "source": [
+    "%%time\n",
+    "\n",
     "ltest = [lb.inverse_transform(e['label'])[0] for e in test]\n",
     "ptest = [predict(model, e) for e in test]\n",
     "\n",
@@ -742,13 +11195,25 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 25,
-   "id": "888494f1",
+   "execution_count": 27,
+   "id": "95f82f2c",
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "CPU times: user 2.21 s, sys: 190 ms, total: 2.4 s\n",
+      "Wall time: 1.93 s\n"
+     ]
+    }
+   ],
    "source": [
+    "%%time\n",
+    "\n",
     "ltrain = [lb.inverse_transform(e['label'])[0] for e in train]\n",
     "ptrain = [predict(model, e) for e in train]\n",
+    "\n",
     "# for e in train:\n",
     "#     print(f\"Label:      {lb.inverse_transform(e['label'])[0]:2d}\")\n",
     "#     print(f\"Prediction: {predict(model, e):2d}\\n_______________\")"
@@ -756,13 +11221,13 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 26,
-   "id": "03dfed1a",
+   "execution_count": 28,
+   "id": "afdfe050",
    "metadata": {},
    "outputs": [
     {
      "data": {
-      "image/png": "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\n",
+      "image/png": "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\n",
       "text/plain": [
        "<Figure size 720x504 with 2 Axes>"
       ]
@@ -771,13 +11236,48 @@
       "needs_background": "light"
      },
      "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "              precision    recall  f1-score   support\n",
+      "\n",
+      "           1       0.00      0.00      0.00         3\n",
+      "           2       0.00      0.00      0.00         3\n",
+      "           3       0.00      0.00      0.00         3\n",
+      "           4       0.00      0.00      0.00         3\n",
+      "           5       0.00      0.00      0.00         3\n",
+      "           6       0.00      0.00      0.00         3\n",
+      "           7       0.50      0.33      0.40         3\n",
+      "           8       0.00      0.00      0.00         3\n",
+      "           9       0.00      0.00      0.00         3\n",
+      "          10       0.00      0.00      0.00         3\n",
+      "          11       0.07      1.00      0.12         3\n",
+      "          12       1.00      0.33      0.50         3\n",
+      "          13       0.00      0.00      0.00         3\n",
+      "          14       0.00      0.00      0.00         3\n",
+      "          15       0.00      0.00      0.00         3\n",
+      "          16       0.00      0.00      0.00         3\n",
+      "\n",
+      "    accuracy                           0.10        48\n",
+      "   macro avg       0.10      0.10      0.06        48\n",
+      "weighted avg       0.10      0.10      0.06        48\n",
+      "\n",
+      "CPU times: user 662 ms, sys: 196 ms, total: 858 ms\n",
+      "Wall time: 627 ms\n"
+     ]
     }
    ],
    "source": [
+    "%%time\n",
+    "\n",
     "from sklearn.metrics import confusion_matrix\n",
     "import seaborn as sn\n",
     "\n",
-    "set_digits = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16}\n",
+    "from sklearn.metrics import classification_report\n",
+    "\n",
+    "set_digits = { 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16 }\n",
     "\n",
     "train_cm = confusion_matrix(ltrain, ptrain, normalize='true')\n",
     "test_cm = confusion_matrix(ltest, ptest, normalize='true')\n",
@@ -787,13 +11287,44 @@
     "sn_plot = sn.heatmap(df_cm, annot=True, cmap=\"Greys\")\n",
     "plt.ylabel(\"True Label\")\n",
     "plt.xlabel(\"Predicted Label\")\n",
-    "plt.show()"
+    "plt.show()\n",
+    "\n",
+    "print(classification_report(ltest, ptest, zero_division=0))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 29,
+   "id": "01ee17bc",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "cenario: SYY\n",
+      "win_sz: 10\n",
+      "stride_sz: 10\n",
+      "dense_steps: 3\n",
+      "layer_count: 7\n",
+      "drop_count: 0.3\n"
+     ]
+    }
+   ],
+   "source": [
+    "print(f'cenario: {cenario}')\n",
+    "print(f'win_sz: {win_sz}')\n",
+    "print(f'stride_sz: {stride_sz}')\n",
+    "print(f'dense_steps: {dense_steps}')\n",
+    "print(f'layer_count: {layer_count}')\n",
+    "print(f'drop_count: {drop_count}')\n",
+    "\n"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "9ad253a7",
+   "id": "2fa548e7",
    "metadata": {},
    "outputs": [],
    "source": []
diff --git a/2-second-project/tdt/SYY35 1010/checkpoint b/2-second-project/tdt/SYY35 1010/checkpoint
new file mode 100644
index 0000000..a3ed6e4
--- /dev/null
+++ b/2-second-project/tdt/SYY35 1010/checkpoint	
@@ -0,0 +1,2 @@
+model_checkpoint_path: "goat.weights"
+all_model_checkpoint_paths: "goat.weights"
diff --git a/2-second-project/tdt/SYY35 1010/goat.weights.data-00000-of-00001 b/2-second-project/tdt/SYY35 1010/goat.weights.data-00000-of-00001
new file mode 100644
index 0000000..da6ea09
Binary files /dev/null and b/2-second-project/tdt/SYY35 1010/goat.weights.data-00000-of-00001 differ
diff --git a/2-second-project/tdt/SYY35 1010/goat.weights.index b/2-second-project/tdt/SYY35 1010/goat.weights.index
new file mode 100644
index 0000000..626ca28
Binary files /dev/null and b/2-second-project/tdt/SYY35 1010/goat.weights.index differ
diff --git a/2-second-project/tdt/checkpoint b/2-second-project/tdt/checkpoint
new file mode 100644
index 0000000..a3ed6e4
--- /dev/null
+++ b/2-second-project/tdt/checkpoint
@@ -0,0 +1,2 @@
+model_checkpoint_path: "goat.weights"
+all_model_checkpoint_paths: "goat.weights"
diff --git a/2-second-project/tdt/goat.weights.data-00000-of-00001 b/2-second-project/tdt/goat.weights.data-00000-of-00001
new file mode 100644
index 0000000..58a10a8
Binary files /dev/null and b/2-second-project/tdt/goat.weights.data-00000-of-00001 differ
diff --git a/2-second-project/tdt/goat.weights.index b/2-second-project/tdt/goat.weights.index
new file mode 100644
index 0000000..074419c
Binary files /dev/null and b/2-second-project/tdt/goat.weights.index differ
diff --git a/2-second-project/tdt/goat.weights/keras_metadata.pb b/2-second-project/tdt/goat.weights/keras_metadata.pb
new file mode 100644
index 0000000..412c0e9
--- /dev/null
+++ b/2-second-project/tdt/goat.weights/keras_metadata.pb
@@ -0,0 +1,15 @@
+
+�Aroot"_tf_keras_network*�@{"name": "model", "trainable": true, "expects_training_arg": true, "dtype": "float32", "batch_input_shape": null, "must_restore_from_config": false, "class_name": "Functional", "config": {"name": "model", "layers": [{"class_name": "InputLayer", "config": {"batch_input_shape": {"class_name": "__tuple__", "items": [null, 10, 338]}, "dtype": "float32", "sparse": false, "ragged": false, "name": "input_1"}, "name": "input_1", "inbound_nodes": []}, {"class_name": "Flatten", "config": {"name": "flatten", "trainable": true, "dtype": "float32", "data_format": "channels_last"}, "name": "flatten", "inbound_nodes": [[["input_1", 0, 0, {}]]]}, {"class_name": "Dropout", "config": {"name": "dropout", "trainable": true, "dtype": "float32", "rate": 0.1, "noise_shape": null, "seed": null}, "name": "dropout", "inbound_nodes": [[["flatten", 0, 0, {}]]]}, {"class_name": "Dense", "config": {"name": "dense", "trainable": true, "dtype": "float32", "units": 500, "activation": "relu", "use_bias": true, "kernel_initializer": {"class_name": "GlorotUniform", "config": {"seed": null}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "kernel_regularizer": null, "bias_regularizer": null, "activity_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "name": "dense", "inbound_nodes": [[["dropout", 0, 0, {}]]]}, {"class_name": "Dropout", "config": {"name": "dropout_1", "trainable": true, "dtype": "float32", "rate": 0.2, "noise_shape": null, "seed": null}, "name": "dropout_1", "inbound_nodes": [[["dense", 0, 0, {}]]]}, {"class_name": "Dense", "config": {"name": "dense_1", "trainable": true, "dtype": "float32", "units": 500, "activation": "relu", "use_bias": true, "kernel_initializer": {"class_name": "GlorotUniform", "config": {"seed": null}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "kernel_regularizer": null, "bias_regularizer": null, "activity_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "name": "dense_1", "inbound_nodes": [[["dropout_1", 0, 0, {}]]]}, {"class_name": "Dropout", "config": {"name": "dropout_2", "trainable": true, "dtype": "float32", "rate": 0.2, "noise_shape": null, "seed": null}, "name": "dropout_2", "inbound_nodes": [[["dense_1", 0, 0, {}]]]}, {"class_name": "Dense", "config": {"name": "dense_2", "trainable": true, "dtype": "float32", "units": 500, "activation": "relu", "use_bias": true, "kernel_initializer": {"class_name": "GlorotUniform", "config": {"seed": null}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "kernel_regularizer": null, "bias_regularizer": null, "activity_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "name": "dense_2", "inbound_nodes": [[["dropout_2", 0, 0, {}]]]}, {"class_name": "Dropout", "config": {"name": "dropout_3", "trainable": true, "dtype": "float32", "rate": 0.3, "noise_shape": null, "seed": null}, "name": "dropout_3", "inbound_nodes": [[["dense_2", 0, 0, {}]]]}, {"class_name": "Dense", "config": {"name": "dense_3", "trainable": true, "dtype": "float32", "units": 16, "activation": "softmax", "use_bias": true, "kernel_initializer": {"class_name": "GlorotUniform", "config": {"seed": null}}, "bias_initializer": {"class_name": "Zeros", "config": {}}, "kernel_regularizer": null, "bias_regularizer": null, "activity_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "name": "dense_3", "inbound_nodes": [[["dropout_3", 0, 0, {}]]]}], "input_layers": [["input_1", 0, 0]], "output_layers": [["dense_3", 0, 0]]}, "shared_object_id": 18, "input_spec": [{"class_name": "InputSpec", "config": {"dtype": null, "shape": {"class_name": "__tuple__", "items": [null, 10, 338]}, "ndim": 3, "max_ndim": null, "min_ndim": null, "axes": {}}}], "build_input_shape": {"class_name": "TensorShape", "items": [null, 10, 338]}, "is_graph_network": true, "save_spec": {"class_name": "TypeSpec", "type_spec": "tf.TensorSpec", "serialized": [{"class_name": "TensorShape", "items": [null, 10, 338]}, "float32", "input_1"]}, "keras_version": "2.5.0", "backend": "tensorflow", "model_config": {"class_name": "Functional", "config": {"name": "model", "layers": [{"class_name": "InputLayer", "config": {"batch_input_shape": {"class_name": "__tuple__", "items": [null, 10, 338]}, "dtype": "float32", "sparse": false, "ragged": false, "name": "input_1"}, "name": "input_1", "inbound_nodes": [], "shared_object_id": 0}, {"class_name": "Flatten", "config": {"name": "flatten", "trainable": true, "dtype": "float32", "data_format": "channels_last"}, "name": "flatten", "inbound_nodes": [[["input_1", 0, 0, {}]]], "shared_object_id": 1}, {"class_name": "Dropout", "config": {"name": "dropout", "trainable": true, "dtype": "float32", "rate": 0.1, "noise_shape": null, "seed": null}, "name": "dropout", "inbound_nodes": [[["flatten", 0, 0, {}]]], "shared_object_id": 2}, {"class_name": "Dense", "config": {"name": "dense", "trainable": true, "dtype": "float32", "units": 500, "activation": "relu", "use_bias": true, "kernel_initializer": {"class_name": "GlorotUniform", "config": {"seed": null}, "shared_object_id": 3}, "bias_initializer": {"class_name": "Zeros", "config": {}, "shared_object_id": 4}, "kernel_regularizer": null, "bias_regularizer": null, "activity_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "name": "dense", "inbound_nodes": [[["dropout", 0, 0, {}]]], "shared_object_id": 5}, {"class_name": "Dropout", "config": {"name": "dropout_1", "trainable": true, "dtype": "float32", "rate": 0.2, "noise_shape": null, "seed": null}, "name": "dropout_1", "inbound_nodes": [[["dense", 0, 0, {}]]], "shared_object_id": 6}, {"class_name": "Dense", "config": {"name": "dense_1", "trainable": true, "dtype": "float32", "units": 500, "activation": "relu", "use_bias": true, "kernel_initializer": {"class_name": "GlorotUniform", "config": {"seed": null}, "shared_object_id": 7}, "bias_initializer": {"class_name": "Zeros", "config": {}, "shared_object_id": 8}, "kernel_regularizer": null, "bias_regularizer": null, "activity_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "name": "dense_1", "inbound_nodes": [[["dropout_1", 0, 0, {}]]], "shared_object_id": 9}, {"class_name": "Dropout", "config": {"name": "dropout_2", "trainable": true, "dtype": "float32", "rate": 0.2, "noise_shape": null, "seed": null}, "name": "dropout_2", "inbound_nodes": [[["dense_1", 0, 0, {}]]], "shared_object_id": 10}, {"class_name": "Dense", "config": {"name": "dense_2", "trainable": true, "dtype": "float32", "units": 500, "activation": "relu", "use_bias": true, "kernel_initializer": {"class_name": "GlorotUniform", "config": {"seed": null}, "shared_object_id": 11}, "bias_initializer": {"class_name": "Zeros", "config": {}, "shared_object_id": 12}, "kernel_regularizer": null, "bias_regularizer": null, "activity_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "name": "dense_2", "inbound_nodes": [[["dropout_2", 0, 0, {}]]], "shared_object_id": 13}, {"class_name": "Dropout", "config": {"name": "dropout_3", "trainable": true, "dtype": "float32", "rate": 0.3, "noise_shape": null, "seed": null}, "name": "dropout_3", "inbound_nodes": [[["dense_2", 0, 0, {}]]], "shared_object_id": 14}, {"class_name": "Dense", "config": {"name": "dense_3", "trainable": true, "dtype": "float32", "units": 16, "activation": "softmax", "use_bias": true, "kernel_initializer": {"class_name": "GlorotUniform", "config": {"seed": null}, "shared_object_id": 15}, "bias_initializer": {"class_name": "Zeros", "config": {}, "shared_object_id": 16}, "kernel_regularizer": null, "bias_regularizer": null, "activity_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "name": "dense_3", "inbound_nodes": [[["dropout_3", 0, 0, {}]]], "shared_object_id": 17}], "input_layers": [["input_1", 0, 0]], "output_layers": [["dense_3", 0, 0]]}}, "training_config": {"loss": "categorical_crossentropy", "metrics": [[{"class_name": "MeanMetricWrapper", "config": {"name": "accuracy", "dtype": "float32", "fn": "categorical_accuracy"}, "shared_object_id": 20}]], "weighted_metrics": null, "loss_weights": null, "optimizer_config": {"class_name": "Adadelta", "config": {"name": "Adadelta", "learning_rate": 0.0010000000474974513, "decay": 0.0, "rho": 0.949999988079071, "epsilon": 1e-07}}}}2


+�
root.layer-0"_tf_keras_input_layer*�{"class_name": "InputLayer", "name": "input_1", "dtype": "float32", "sparse": false, "ragged": false, "batch_input_shape": {"class_name": "__tuple__", "items": [null, 10, 338]}, "config": {"batch_input_shape": {"class_name": "__tuple__", "items": [null, 10, 338]}, "dtype": "float32", "sparse": false, "ragged": false, "name": "input_1"}}2


+�root.layer-1"_tf_keras_layer*�{"name": "flatten", "trainable": true, "expects_training_arg": false, "dtype": "float32", "batch_input_shape": null, "stateful": false, "must_restore_from_config": false, "class_name": "Flatten", "config": {"name": "flatten", "trainable": true, "dtype": "float32", "data_format": "channels_last"}, "inbound_nodes": [[["input_1", 0, 0, {}]]], "shared_object_id": 1, "input_spec": {"class_name": "InputSpec", "config": {"dtype": null, "shape": null, "ndim": null, "max_ndim": null, "min_ndim": 1, "axes": {}}, "shared_object_id": 21}}2


+�root.layer-2"_tf_keras_layer*�{"name": "dropout", "trainable": true, "expects_training_arg": true, "dtype": "float32", "batch_input_shape": null, "stateful": false, "must_restore_from_config": false, "class_name": "Dropout", "config": {"name": "dropout", "trainable": true, "dtype": "float32", "rate": 0.1, "noise_shape": null, "seed": null}, "inbound_nodes": [[["flatten", 0, 0, {}]]], "shared_object_id": 2}2


+�root.layer_with_weights-0"_tf_keras_layer*�{"name": "dense", "trainable": true, "expects_training_arg": false, "dtype": "float32", "batch_input_shape": null, "stateful": false, "must_restore_from_config": false, "class_name": "Dense", "config": {"name": "dense", "trainable": true, "dtype": "float32", "units": 500, "activation": "relu", "use_bias": true, "kernel_initializer": {"class_name": "GlorotUniform", "config": {"seed": null}, "shared_object_id": 3}, "bias_initializer": {"class_name": "Zeros", "config": {}, "shared_object_id": 4}, "kernel_regularizer": null, "bias_regularizer": null, "activity_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "inbound_nodes": [[["dropout", 0, 0, {}]]], "shared_object_id": 5, "input_spec": {"class_name": "InputSpec", "config": {"dtype": null, "shape": null, "ndim": null, "max_ndim": null, "min_ndim": 2, "axes": {"-1": 3380}}, "shared_object_id": 22}, "build_input_shape": {"class_name": "TensorShape", "items": [null, 3380]}}2


+�root.layer-4"_tf_keras_layer*�{"name": "dropout_1", "trainable": true, "expects_training_arg": true, "dtype": "float32", "batch_input_shape": null, "stateful": false, "must_restore_from_config": false, "class_name": "Dropout", "config": {"name": "dropout_1", "trainable": true, "dtype": "float32", "rate": 0.2, "noise_shape": null, "seed": null}, "inbound_nodes": [[["dense", 0, 0, {}]]], "shared_object_id": 6}2


+�root.layer_with_weights-1"_tf_keras_layer*�{"name": "dense_1", "trainable": true, "expects_training_arg": false, "dtype": "float32", "batch_input_shape": null, "stateful": false, "must_restore_from_config": false, "class_name": "Dense", "config": {"name": "dense_1", "trainable": true, "dtype": "float32", "units": 500, "activation": "relu", "use_bias": true, "kernel_initializer": {"class_name": "GlorotUniform", "config": {"seed": null}, "shared_object_id": 7}, "bias_initializer": {"class_name": "Zeros", "config": {}, "shared_object_id": 8}, "kernel_regularizer": null, "bias_regularizer": null, "activity_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "inbound_nodes": [[["dropout_1", 0, 0, {}]]], "shared_object_id": 9, "input_spec": {"class_name": "InputSpec", "config": {"dtype": null, "shape": null, "ndim": null, "max_ndim": null, "min_ndim": 2, "axes": {"-1": 500}}, "shared_object_id": 23}, "build_input_shape": {"class_name": "TensorShape", "items": [null, 500]}}2


+�root.layer-6"_tf_keras_layer*�{"name": "dropout_2", "trainable": true, "expects_training_arg": true, "dtype": "float32", "batch_input_shape": null, "stateful": false, "must_restore_from_config": false, "class_name": "Dropout", "config": {"name": "dropout_2", "trainable": true, "dtype": "float32", "rate": 0.2, "noise_shape": null, "seed": null}, "inbound_nodes": [[["dense_1", 0, 0, {}]]], "shared_object_id": 10}2


+�root.layer_with_weights-2"_tf_keras_layer*�{"name": "dense_2", "trainable": true, "expects_training_arg": false, "dtype": "float32", "batch_input_shape": null, "stateful": false, "must_restore_from_config": false, "class_name": "Dense", "config": {"name": "dense_2", "trainable": true, "dtype": "float32", "units": 500, "activation": "relu", "use_bias": true, "kernel_initializer": {"class_name": "GlorotUniform", "config": {"seed": null}, "shared_object_id": 11}, "bias_initializer": {"class_name": "Zeros", "config": {}, "shared_object_id": 12}, "kernel_regularizer": null, "bias_regularizer": null, "activity_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "inbound_nodes": [[["dropout_2", 0, 0, {}]]], "shared_object_id": 13, "input_spec": {"class_name": "InputSpec", "config": {"dtype": null, "shape": null, "ndim": null, "max_ndim": null, "min_ndim": 2, "axes": {"-1": 500}}, "shared_object_id": 24}, "build_input_shape": {"class_name": "TensorShape", "items": [null, 500]}}2


+�	root.layer-8"_tf_keras_layer*�{"name": "dropout_3", "trainable": true, "expects_training_arg": true, "dtype": "float32", "batch_input_shape": null, "stateful": false, "must_restore_from_config": false, "class_name": "Dropout", "config": {"name": "dropout_3", "trainable": true, "dtype": "float32", "rate": 0.3, "noise_shape": null, "seed": null}, "inbound_nodes": [[["dense_2", 0, 0, {}]]], "shared_object_id": 14}2


+�
+root.layer_with_weights-3"_tf_keras_layer*�{"name": "dense_3", "trainable": true, "expects_training_arg": false, "dtype": "float32", "batch_input_shape": null, "stateful": false, "must_restore_from_config": false, "class_name": "Dense", "config": {"name": "dense_3", "trainable": true, "dtype": "float32", "units": 16, "activation": "softmax", "use_bias": true, "kernel_initializer": {"class_name": "GlorotUniform", "config": {"seed": null}, "shared_object_id": 15}, "bias_initializer": {"class_name": "Zeros", "config": {}, "shared_object_id": 16}, "kernel_regularizer": null, "bias_regularizer": null, "activity_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "inbound_nodes": [[["dropout_3", 0, 0, {}]]], "shared_object_id": 17, "input_spec": {"class_name": "InputSpec", "config": {"dtype": null, "shape": null, "ndim": null, "max_ndim": null, "min_ndim": 2, "axes": {"-1": 500}}, "shared_object_id": 25}, "build_input_shape": {"class_name": "TensorShape", "items": [null, 500]}}2


+�
sroot.keras_api.metrics.0"_tf_keras_metric*�
{"class_name": "Mean", "name": "loss", "dtype": "float32", "config": {"name": "loss", "dtype": "float32"}, "shared_object_id": 26}2


+�
troot.keras_api.metrics.1"_tf_keras_metric*�
{"class_name": "MeanMetricWrapper", "name": "accuracy", "dtype": "float32", "config": {"name": "accuracy", "dtype": "float32", "fn": "categorical_accuracy"}, "shared_object_id": 20}2


\ No newline at end of file
diff --git a/2-second-project/tdt/goat.weights/saved_model.pb b/2-second-project/tdt/goat.weights/saved_model.pb
new file mode 100644
index 0000000..1f9a1a2
Binary files /dev/null and b/2-second-project/tdt/goat.weights/saved_model.pb differ
diff --git a/2-second-project/tdt/goat.weights/variables/variables.data-00000-of-00001 b/2-second-project/tdt/goat.weights/variables/variables.data-00000-of-00001
new file mode 100644
index 0000000..1a650d1
Binary files /dev/null and b/2-second-project/tdt/goat.weights/variables/variables.data-00000-of-00001 differ
diff --git a/2-second-project/tdt/goat.weights/variables/variables.index b/2-second-project/tdt/goat.weights/variables/variables.index
new file mode 100644
index 0000000..e614ec1
Binary files /dev/null and b/2-second-project/tdt/goat.weights/variables/variables.index differ