experimental_logistic_regression.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "2c2313fd",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import torch\n",
    "from tqdm import tqdm\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "21a1ffeb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([2., 3., 1.], device='cuda:0')\n"
     ]
    }
   ],
   "source": [
    "assert torch.cuda.is_available(), 'CUDA is not correctly installed!!'\n",
    "print(torch.tensor([2.0, 3.0, 1.0]).cuda())\n",
    "property = torch.cuda.get_device_properties(0)\n",
    "\n",
    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "cd0dc67b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "====== Your GPU info ======\n",
      "name:\t\t Tesla T4\n",
      "capability:\t v7.5\n",
      "memory:\t\t 16 Gb\n",
      "processors:\t 40\n"
     ]
    }
   ],
   "source": [
    "print('====== Your GPU info ======')\n",
    "print('name:\\t\\t', property.name)\n",
    "print('capability:\\t', 'v{}.{}'.format(property.major, property.minor))\n",
    "print('memory:\\t\\t', round(property.total_memory / 1e9), 'Gb')\n",
    "print('processors:\\t', property.multi_processor_count)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "684d7bd3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(2462, 15)\n",
      "(2462, 16)\n",
      "(2462,)\n",
      "[0 0 1 ... 0 1 0]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.datasets import make_classification\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "# LOAD AND SPLIT DATASET HERE\n",
    "\n",
    "# Load headers\n",
    "csv_path = \"dataset_v2.csv\"\n",
    "\n",
    "with open(csv_path, 'r', newline='') as csv_fh:\n",
    "    headers = csv_fh.readline().strip().split(',')\n",
    "    \n",
    "    \n",
    "label_col = \"Price\"\n",
    "date_col = \"Date(UTC)\"\n",
    "# Load features and labels\n",
    "x_cols = [i for i in range(len(headers)) if (headers[i] != label_col and headers[i] != date_col)]\n",
    "l_cols = [i for i in range(len(headers)) if headers[i] == label_col]\n",
    "inputs = np.loadtxt(csv_path, delimiter=',', skiprows=1, usecols=x_cols)\n",
    "prices = np.loadtxt(csv_path, delimiter=',', skiprows=1, usecols=l_cols)\n",
    "\n",
    "inputs = np.delete(inputs, -1, axis=0)\n",
    "print(inputs.shape)\n",
    "\n",
    "# add intercept\n",
    "new_x = np.zeros((inputs.shape[0], inputs.shape[1] + 1), dtype=inputs.dtype)\n",
    "new_x[:, 0] = 1\n",
    "new_x[:, 1:] = inputs\n",
    "\n",
    "inputs = new_x\n",
    "print(inputs.shape)\n",
    "\n",
    "diff = np.diff(prices)\n",
    "labels = np.where(diff > 0, 1, 0)\n",
    "print(labels.shape)\n",
    "print(labels)\n",
    "\n",
    "# Feature engineering\n",
    "norm = MinMaxScaler().fit(inputs)\n",
    "inputs = norm.transform(inputs)\n",
    "\n",
    "# apply standardization on numerical features\n",
    "for i in range(inputs.shape[1]):\n",
    "    \n",
    "    # fit on training data column\n",
    "    scale = StandardScaler().fit(inputs[[i]])\n",
    "    \n",
    "    # transform the training data column\n",
    "    inputs[i] = scale.transform(inputs[[i]])\n",
    "\n",
    "\n",
    "X_train, X_test, y_train,  y_test = train_test_split(\n",
    "    inputs, labels, test_size=0.1, random_state=42)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "143fd703",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.nn.functional as F\n",
    "\n",
    "class LogisticRegression(torch.nn.Module):\n",
    "    def __init__(self, input_dim, output_dim):\n",
    "        super(LogisticRegression, self).__init__()\n",
    "        self.layer1 = torch.nn.Linear(input_dim,50)\n",
    "        self.layer2 = torch.nn.Linear(50, 10)\n",
    "        self.layer3 = torch.nn.Linear(10, output_dim)\n",
    "        \n",
    "    def forward(self, x):\n",
    "        x = torch.sigmoid(self.layer1(x))\n",
    "        x = torch.sigmoid(self.layer2(x))\n",
    "        x = torch.sigmoid(self.layer3(x)) # To check with the loss function\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "fbe9b849",
   "metadata": {},
   "outputs": [],
   "source": [
    "epochs = 1000000\n",
    "input_dim = inputs.shape[1] # features \n",
    "output_dim = 1 # price\n",
    "learning_rate = 0.005\n",
    "\n",
    "model = LogisticRegression(input_dim,output_dim).to(device)\n",
    "\n",
    "criterion = torch.nn.BCELoss()\n",
    "\n",
    "optimizer = torch.optim.SGD(model.parameters(), lr=learning_rate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "9ed4fbd7",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Convert data to tensors, move onto GPU\n",
    "\n",
    "X_train, X_test = torch.Tensor(X_train).to(device), torch.Tensor(X_test).to(device)\n",
    "y_train, y_test = torch.Tensor(y_train).to(device), torch.Tensor(y_test).to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "ef2f6b4d",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:   0%|                  | 226/1000000 [00:00<14:46, 1127.52it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 0. \n",
      "Test - Loss: 0.6974973678588867. Accuracy: 48.582995951417004\n",
      "Train -  Loss: 0.6961884498596191. Accuracy: 49.52595936794582\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:   1%|▏               | 10171/1000000 [00:08<14:31, 1136.31it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 10000. \n",
      "Test - Loss: 0.6929652094841003. Accuracy: 51.417004048582996\n",
      "Train -  Loss: 0.6930874586105347. Accuracy: 50.47404063205418\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:   2%|▎               | 20152/1000000 [00:17<14:20, 1138.80it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 20000. \n",
      "Test - Loss: 0.6929425001144409. Accuracy: 51.417004048582996\n",
      "Train -  Loss: 0.693055272102356. Accuracy: 50.47404063205418\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:   3%|▌                | 30115/1000000 [00:27<20:26, 790.49it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 30000. \n",
      "Test - Loss: 0.6929199695587158. Accuracy: 51.417004048582996\n",
      "Train -  Loss: 0.6930225491523743. Accuracy: 50.47404063205418\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:   4%|▋               | 40143/1000000 [00:36<14:07, 1133.13it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 40000. \n",
      "Test - Loss: 0.6928967833518982. Accuracy: 51.417004048582996\n",
      "Train -  Loss: 0.6929885149002075. Accuracy: 50.47404063205418\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:   5%|▊               | 50141/1000000 [00:45<14:04, 1124.27it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 50000. \n",
      "Test - Loss: 0.6928727030754089. Accuracy: 51.417004048582996\n",
      "Train -  Loss: 0.692952573299408. Accuracy: 50.47404063205418\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:   6%|▉               | 60198/1000000 [00:54<14:00, 1118.02it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 60000. \n",
      "Test - Loss: 0.6928471326828003. Accuracy: 51.417004048582996\n",
      "Train -  Loss: 0.6929139494895935. Accuracy: 50.47404063205418\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:   7%|█               | 70125/1000000 [01:03<13:42, 1130.26it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 70000. \n",
      "Test - Loss: 0.6928195953369141. Accuracy: 51.417004048582996\n",
      "Train -  Loss: 0.6928720474243164. Accuracy: 50.47404063205418\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:   8%|█▎              | 80176/1000000 [01:12<13:31, 1134.11it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 80000. \n",
      "Test - Loss: 0.6927896738052368. Accuracy: 51.417004048582996\n",
      "Train -  Loss: 0.692825973033905. Accuracy: 50.519187358916476\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:   9%|█▍              | 90216/1000000 [01:21<13:21, 1135.48it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 90000. \n",
      "Test - Loss: 0.6927570700645447. Accuracy: 51.417004048582996\n",
      "Train -  Loss: 0.692775309085846. Accuracy: 50.42889390519187\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  10%|█▌             | 100131/1000000 [01:31<12:56, 1158.55it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 100000. \n",
      "Test - Loss: 0.6927213668823242. Accuracy: 51.012145748987855\n",
      "Train -  Loss: 0.6927192807197571. Accuracy: 50.42889390519187\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  11%|█▋             | 110149/1000000 [01:40<13:09, 1126.83it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 110000. \n",
      "Test - Loss: 0.692682147026062. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6926573514938354. Accuracy: 52.50564334085779\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  12%|█▊             | 120162/1000000 [01:49<12:49, 1142.91it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 120000. \n",
      "Test - Loss: 0.6926394701004028. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6925891041755676. Accuracy: 51.963882618510155\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  13%|█▉             | 130198/1000000 [01:58<12:47, 1132.65it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 130000. \n",
      "Test - Loss: 0.6925932168960571. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.692514181137085. Accuracy: 51.46726862302483\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  14%|██             | 140165/1000000 [02:07<12:33, 1141.23it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 140000. \n",
      "Test - Loss: 0.6925437450408936. Accuracy: 54.25101214574899\n",
      "Train -  Loss: 0.6924329400062561. Accuracy: 51.78329571106095\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  15%|██▎            | 150208/1000000 [02:16<12:39, 1118.62it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 150000. \n",
      "Test - Loss: 0.692491888999939. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6923457384109497. Accuracy: 51.78329571106095\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  16%|██▍            | 160126/1000000 [02:24<12:52, 1087.30it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 160000. \n",
      "Test - Loss: 0.6924384236335754. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6922536492347717. Accuracy: 51.918735891647856\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  17%|██▌            | 170227/1000000 [02:35<12:10, 1136.19it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 170000. \n",
      "Test - Loss: 0.6923847794532776. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6921582221984863. Accuracy: 51.82844243792325\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  18%|██▋            | 180176/1000000 [02:43<11:57, 1143.06it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 180000. \n",
      "Test - Loss: 0.6923326849937439. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6920614242553711. Accuracy: 51.64785553047404\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  19%|██▊            | 190123/1000000 [02:52<11:58, 1126.72it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 190000. \n",
      "Test - Loss: 0.692284107208252. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6919660568237305. Accuracy: 51.64785553047404\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  20%|███            | 200120/1000000 [03:01<11:31, 1155.95it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 200000. \n",
      "Test - Loss: 0.6922410130500793. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6918744444847107. Accuracy: 51.64785553047404\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  21%|███▏           | 210191/1000000 [03:10<11:53, 1106.72it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 210000. \n",
      "Test - Loss: 0.6922048926353455. Accuracy: 52.63157894736842\n",
      "Train -  Loss: 0.6917893886566162. Accuracy: 51.918735891647856\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  22%|███▎           | 220183/1000000 [03:19<11:42, 1110.09it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 220000. \n",
      "Test - Loss: 0.6921765804290771. Accuracy: 52.63157894736842\n",
      "Train -  Loss: 0.6917125582695007. Accuracy: 52.18961625282167\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  23%|███▋            | 230094/1000000 [03:29<17:00, 754.46it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 230000. \n",
      "Test - Loss: 0.6921564340591431. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6916452050209045. Accuracy: 52.41534988713318\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  24%|███▌           | 240185/1000000 [03:38<11:05, 1140.91it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 240000. \n",
      "Test - Loss: 0.6921437978744507. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.691587507724762. Accuracy: 52.27990970654628\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  25%|███▊           | 250186/1000000 [03:47<11:07, 1123.49it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 250000. \n",
      "Test - Loss: 0.6921378374099731. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6915389895439148. Accuracy: 52.37020316027088\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  26%|███▉           | 260209/1000000 [03:56<10:44, 1147.49it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 260000. \n",
      "Test - Loss: 0.6921367645263672. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6914985775947571. Accuracy: 52.59593679458239\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  27%|████           | 270196/1000000 [04:04<10:41, 1138.43it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 270000. \n",
      "Test - Loss: 0.6921390891075134. Accuracy: 53.84615384615385\n",
      "Train -  Loss: 0.6914646625518799. Accuracy: 52.86681715575621\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  28%|████▏          | 280124/1000000 [04:13<10:28, 1144.84it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 280000. \n",
      "Test - Loss: 0.692143440246582. Accuracy: 53.84615384615385\n",
      "Train -  Loss: 0.6914357542991638. Accuracy: 52.957110609480814\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  29%|████▎          | 290223/1000000 [04:22<10:32, 1121.92it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 290000. \n",
      "Test - Loss: 0.6921486258506775. Accuracy: 53.84615384615385\n",
      "Train -  Loss: 0.6914106607437134. Accuracy: 52.8216704288939\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  30%|████▌          | 300136/1000000 [04:32<10:38, 1096.74it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 300000. \n",
      "Test - Loss: 0.692153811454773. Accuracy: 53.84615384615385\n",
      "Train -  Loss: 0.691387951374054. Accuracy: 52.8216704288939\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  31%|████▋          | 310165/1000000 [04:41<10:04, 1140.96it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 310000. \n",
      "Test - Loss: 0.6921584010124207. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6913670301437378. Accuracy: 52.91196388261851\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  32%|████▊          | 320125/1000000 [04:50<09:55, 1141.61it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 320000. \n",
      "Test - Loss: 0.692162036895752. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6913469433784485. Accuracy: 52.957110609480814\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  33%|████▉          | 330133/1000000 [04:59<10:47, 1034.28it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 330000. \n",
      "Test - Loss: 0.6921648979187012. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6913273334503174. Accuracy: 52.86681715575621\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  34%|█████          | 340167/1000000 [05:07<09:45, 1126.14it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 340000. \n",
      "Test - Loss: 0.6921666264533997. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6913079619407654. Accuracy: 52.91196388261851\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  35%|█████▎         | 350156/1000000 [05:16<09:25, 1148.80it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 350000. \n",
      "Test - Loss: 0.6921674609184265. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.691288411617279. Accuracy: 52.957110609480814\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  36%|█████▊          | 360054/1000000 [05:25<11:16, 946.51it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 360000. \n",
      "Test - Loss: 0.6921675801277161. Accuracy: 53.84615384615385\n",
      "Train -  Loss: 0.6912686824798584. Accuracy: 52.91196388261851\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  37%|█████▌         | 370123/1000000 [05:35<09:13, 1138.92it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 370000. \n",
      "Test - Loss: 0.6921666264533997. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6912485957145691. Accuracy: 52.91196388261851\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  38%|█████▋         | 380189/1000000 [05:44<09:36, 1074.76it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 380000. \n",
      "Test - Loss: 0.6921650171279907. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6912281513214111. Accuracy: 53.04740406320542\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  39%|█████▊         | 390223/1000000 [05:53<08:51, 1146.69it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 390000. \n",
      "Test - Loss: 0.692162811756134. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6912070512771606. Accuracy: 52.91196388261851\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  40%|██████         | 400152/1000000 [06:01<08:44, 1142.96it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 400000. \n",
      "Test - Loss: 0.6921600699424744. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6911852955818176. Accuracy: 52.8216704288939\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  41%|██████▏        | 410225/1000000 [06:10<08:31, 1152.27it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 410000. \n",
      "Test - Loss: 0.6921567320823669. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6911631226539612. Accuracy: 52.776523702031604\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  42%|██████▎        | 420162/1000000 [06:19<08:34, 1126.91it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 420000. \n",
      "Test - Loss: 0.6921527981758118. Accuracy: 52.63157894736842\n",
      "Train -  Loss: 0.691140353679657. Accuracy: 52.86681715575621\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  43%|██████▉         | 430179/1000000 [06:29<10:09, 935.41it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 430000. \n",
      "Test - Loss: 0.6921484470367432. Accuracy: 52.22672064777328\n",
      "Train -  Loss: 0.6911169290542603. Accuracy: 52.91196388261851\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  44%|██████▌        | 440187/1000000 [06:38<08:13, 1134.34it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 440000. \n",
      "Test - Loss: 0.6921436190605164. Accuracy: 52.22672064777328\n",
      "Train -  Loss: 0.6910927295684814. Accuracy: 52.7313769751693\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  45%|██████▊        | 450152/1000000 [06:47<08:00, 1144.48it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 450000. \n",
      "Test - Loss: 0.6921384334564209. Accuracy: 52.22672064777328\n",
      "Train -  Loss: 0.691067636013031. Accuracy: 52.776523702031604\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  46%|██████▉        | 460162/1000000 [06:55<07:46, 1158.41it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 460000. \n",
      "Test - Loss: 0.6921324729919434. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6910420656204224. Accuracy: 52.91196388261851\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  47%|███████        | 470169/1000000 [07:04<07:46, 1134.87it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 470000. \n",
      "Test - Loss: 0.6921260952949524. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6910154819488525. Accuracy: 52.86681715575621\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  48%|███████▏       | 480124/1000000 [07:13<07:30, 1154.07it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 480000. \n",
      "Test - Loss: 0.6921194195747375. Accuracy: 53.84615384615385\n",
      "Train -  Loss: 0.6909881830215454. Accuracy: 52.86681715575621\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  49%|███████▎       | 490131/1000000 [07:22<07:23, 1149.74it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 490000. \n",
      "Test - Loss: 0.6921122670173645. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6909599304199219. Accuracy: 52.91196388261851\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  50%|███████▌       | 500137/1000000 [07:32<07:33, 1102.73it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 500000. \n",
      "Test - Loss: 0.692104697227478. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6909307241439819. Accuracy: 53.00225733634311\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  51%|███████▋       | 510198/1000000 [07:41<07:07, 1145.12it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 510000. \n",
      "Test - Loss: 0.6920963525772095. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6909007430076599. Accuracy: 53.00225733634311\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  52%|███████▊       | 520133/1000000 [07:49<07:03, 1133.72it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 520000. \n",
      "Test - Loss: 0.6920877695083618. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6908696889877319. Accuracy: 52.957110609480814\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  53%|███████▉       | 530197/1000000 [07:58<06:54, 1132.75it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 530000. \n",
      "Test - Loss: 0.6920783519744873. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6908377408981323. Accuracy: 53.04740406320542\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  54%|████████       | 540142/1000000 [08:07<06:41, 1144.17it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 540000. \n",
      "Test - Loss: 0.6920685768127441. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6908047795295715. Accuracy: 53.09255079006772\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  55%|████████▎      | 550137/1000000 [08:16<06:30, 1150.93it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 550000. \n",
      "Test - Loss: 0.6920581459999084. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.69077068567276. Accuracy: 53.04740406320542\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  56%|████████▍      | 560213/1000000 [08:24<06:34, 1115.27it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 560000. \n",
      "Test - Loss: 0.692047119140625. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6907355785369873. Accuracy: 53.182844243792324\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  57%|████████▌      | 570203/1000000 [08:35<06:31, 1097.04it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 570000. \n",
      "Test - Loss: 0.6920351982116699. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6906996369361877. Accuracy: 53.27313769751693\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  58%|████████▋      | 580191/1000000 [08:44<06:15, 1117.62it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 580000. \n",
      "Test - Loss: 0.6920229196548462. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6906625628471375. Accuracy: 53.182844243792324\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  59%|████████▊      | 590191/1000000 [08:52<05:57, 1144.77it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 590000. \n",
      "Test - Loss: 0.6920100450515747. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6906245350837708. Accuracy: 53.22799097065463\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  60%|█████████      | 600231/1000000 [09:01<05:44, 1159.42it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 600000. \n",
      "Test - Loss: 0.6919968724250793. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6905854344367981. Accuracy: 53.09255079006772\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  61%|█████████▏     | 610118/1000000 [09:10<05:40, 1146.06it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 610000. \n",
      "Test - Loss: 0.6919829249382019. Accuracy: 52.63157894736842\n",
      "Train -  Loss: 0.690545380115509. Accuracy: 53.137697516930025\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  62%|█████████▎     | 620165/1000000 [09:19<05:30, 1149.48it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 620000. \n",
      "Test - Loss: 0.6919682621955872. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6905044913291931. Accuracy: 53.00225733634311\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  63%|██████████      | 630103/1000000 [09:28<07:49, 787.87it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 630000. \n",
      "Test - Loss: 0.6919528841972351. Accuracy: 52.22672064777328\n",
      "Train -  Loss: 0.6904626488685608. Accuracy: 53.182844243792324\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  64%|█████████▌     | 640102/1000000 [09:37<05:23, 1111.39it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 640000. \n",
      "Test - Loss: 0.6919368505477905. Accuracy: 51.821862348178136\n",
      "Train -  Loss: 0.6904199123382568. Accuracy: 53.182844243792324\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  65%|█████████▊     | 650124/1000000 [09:46<05:04, 1150.13it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 650000. \n",
      "Test - Loss: 0.6919204592704773. Accuracy: 52.22672064777328\n",
      "Train -  Loss: 0.6903764605522156. Accuracy: 53.27313769751693\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  66%|█████████▉     | 660118/1000000 [09:55<04:56, 1147.81it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 660000. \n",
      "Test - Loss: 0.6919035911560059. Accuracy: 52.63157894736842\n",
      "Train -  Loss: 0.6903322339057922. Accuracy: 53.22799097065463\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  67%|██████████     | 670122/1000000 [10:04<04:47, 1149.32it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 670000. \n",
      "Test - Loss: 0.6918862462043762. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6902874112129211. Accuracy: 53.318284424379236\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  68%|██████████▏    | 680151/1000000 [10:13<04:37, 1153.32it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 680000. \n",
      "Test - Loss: 0.6918681263923645. Accuracy: 52.63157894736842\n",
      "Train -  Loss: 0.6902420520782471. Accuracy: 53.318284424379236\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  69%|██████████▎    | 690136/1000000 [10:22<04:31, 1140.65it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 690000. \n",
      "Test - Loss: 0.6918493509292603. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6901965737342834. Accuracy: 53.589164785553045\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  70%|██████████▌    | 700224/1000000 [10:32<04:19, 1153.61it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 700000. \n",
      "Test - Loss: 0.6918300986289978. Accuracy: 53.441295546558706\n",
      "Train -  Loss: 0.6901506185531616. Accuracy: 53.589164785553045\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  71%|██████████▋    | 710226/1000000 [10:41<04:18, 1122.75it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 710000. \n",
      "Test - Loss: 0.6918107271194458. Accuracy: 53.036437246963565\n",
      "Train -  Loss: 0.6901042461395264. Accuracy: 53.589164785553045\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  72%|██████████▊    | 720163/1000000 [10:49<04:07, 1130.18it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 720000. \n",
      "Test - Loss: 0.6917911767959595. Accuracy: 53.84615384615385\n",
      "Train -  Loss: 0.6900578737258911. Accuracy: 53.408577878103834\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  73%|██████████▉    | 730164/1000000 [10:58<03:59, 1127.45it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 730000. \n",
      "Test - Loss: 0.691771388053894. Accuracy: 53.84615384615385\n",
      "Train -  Loss: 0.6900113821029663. Accuracy: 53.45372460496614\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  74%|███████████    | 740172/1000000 [11:07<03:49, 1133.62it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 740000. \n",
      "Test - Loss: 0.6917517185211182. Accuracy: 53.84615384615385\n",
      "Train -  Loss: 0.689965009689331. Accuracy: 53.137697516930025\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  75%|███████████▎   | 750121/1000000 [11:16<03:36, 1151.57it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 750000. \n",
      "Test - Loss: 0.6917320489883423. Accuracy: 53.84615384615385\n",
      "Train -  Loss: 0.6899188756942749. Accuracy: 53.182844243792324\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  76%|███████████▍   | 760145/1000000 [11:24<03:28, 1149.90it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 760000. \n",
      "Test - Loss: 0.6917124390602112. Accuracy: 53.84615384615385\n",
      "Train -  Loss: 0.6898730397224426. Accuracy: 53.09255079006772\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  77%|███████████▌   | 770217/1000000 [11:35<03:23, 1128.81it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 770000. \n",
      "Test - Loss: 0.6916929483413696. Accuracy: 53.84615384615385\n",
      "Train -  Loss: 0.6898276209831238. Accuracy: 53.09255079006772\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  78%|███████████▋   | 780176/1000000 [11:43<03:14, 1132.34it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 780000. \n",
      "Test - Loss: 0.691673755645752. Accuracy: 53.84615384615385\n",
      "Train -  Loss: 0.6897826194763184. Accuracy: 53.137697516930025\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  79%|███████████▊   | 790175/1000000 [11:52<03:02, 1147.55it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 790000. \n",
      "Test - Loss: 0.6916550993919373. Accuracy: 54.65587044534413\n",
      "Train -  Loss: 0.6897381544113159. Accuracy: 53.00225733634311\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  80%|████████████   | 800199/1000000 [12:01<02:53, 1150.03it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 800000. \n",
      "Test - Loss: 0.6916362047195435. Accuracy: 54.65587044534413\n",
      "Train -  Loss: 0.6896945238113403. Accuracy: 52.91196388261851\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  81%|████████████▏  | 810174/1000000 [12:10<02:44, 1152.43it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 810000. \n",
      "Test - Loss: 0.6916177868843079. Accuracy: 54.25101214574899\n",
      "Train -  Loss: 0.689651608467102. Accuracy: 52.776523702031604\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  82%|████████████▎  | 820210/1000000 [12:19<02:36, 1148.89it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 820000. \n",
      "Test - Loss: 0.6915996670722961. Accuracy: 54.25101214574899\n",
      "Train -  Loss: 0.6896098256111145. Accuracy: 52.91196388261851\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  83%|█████████████▎  | 830098/1000000 [12:28<03:23, 834.63it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 830000. \n",
      "Test - Loss: 0.6915825605392456. Accuracy: 54.65587044534413\n",
      "Train -  Loss: 0.689568817615509. Accuracy: 53.04740406320542\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  84%|████████████▌  | 840134/1000000 [12:37<02:18, 1151.93it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 840000. \n",
      "Test - Loss: 0.6915661096572876. Accuracy: 54.65587044534413\n",
      "Train -  Loss: 0.6895285248756409. Accuracy: 53.318284424379236\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  85%|████████████▊  | 850143/1000000 [12:46<02:16, 1097.82it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 850000. \n",
      "Test - Loss: 0.6915497183799744. Accuracy: 54.65587044534413\n",
      "Train -  Loss: 0.6894892454147339. Accuracy: 53.318284424379236\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  86%|████████████▉  | 860188/1000000 [12:55<02:01, 1152.41it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 860000. \n",
      "Test - Loss: 0.691534161567688. Accuracy: 54.65587044534413\n",
      "Train -  Loss: 0.6894509196281433. Accuracy: 53.182844243792324\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  87%|█████████████  | 870133/1000000 [13:04<01:54, 1137.94it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 870000. \n",
      "Test - Loss: 0.6915192008018494. Accuracy: 54.65587044534413\n",
      "Train -  Loss: 0.6894137859344482. Accuracy: 53.22799097065463\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  88%|█████████████▏ | 880226/1000000 [13:12<01:42, 1168.28it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 880000. \n",
      "Test - Loss: 0.6915053725242615. Accuracy: 55.06072874493927\n",
      "Train -  Loss: 0.6893774271011353. Accuracy: 53.22799097065463\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  89%|█████████████▎ | 890218/1000000 [13:21<01:37, 1123.92it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 890000. \n",
      "Test - Loss: 0.6914921998977661. Accuracy: 55.87044534412956\n",
      "Train -  Loss: 0.6893421411514282. Accuracy: 53.09255079006772\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  90%|█████████████▌ | 900113/1000000 [13:31<01:33, 1069.50it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 900000. \n",
      "Test - Loss: 0.6914798021316528. Accuracy: 55.465587044534416\n",
      "Train -  Loss: 0.6893079280853271. Accuracy: 53.04740406320542\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  91%|█████████████▋ | 910137/1000000 [13:40<01:22, 1084.14it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 910000. \n",
      "Test - Loss: 0.6914674639701843. Accuracy: 55.465587044534416\n",
      "Train -  Loss: 0.6892750263214111. Accuracy: 52.91196388261851\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  92%|█████████████▊ | 920173/1000000 [13:49<01:13, 1085.73it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 920000. \n",
      "Test - Loss: 0.6914556622505188. Accuracy: 55.465587044534416\n",
      "Train -  Loss: 0.6892431378364563. Accuracy: 53.04740406320542\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  93%|█████████████▉ | 930127/1000000 [13:58<01:02, 1110.11it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 930000. \n",
      "Test - Loss: 0.6914448142051697. Accuracy: 55.87044534412956\n",
      "Train -  Loss: 0.6892120838165283. Accuracy: 53.09255079006772\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  94%|██████████████ | 940212/1000000 [14:07<00:53, 1124.79it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 940000. \n",
      "Test - Loss: 0.6914350986480713. Accuracy: 55.465587044534416\n",
      "Train -  Loss: 0.6891818642616272. Accuracy: 53.22799097065463\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  95%|██████████████▎| 950168/1000000 [14:16<00:46, 1070.46it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 950000. \n",
      "Test - Loss: 0.6914246082305908. Accuracy: 55.465587044534416\n",
      "Train -  Loss: 0.6891528964042664. Accuracy: 53.22799097065463\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  96%|██████████████▍| 960135/1000000 [14:24<00:34, 1149.55it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 960000. \n",
      "Test - Loss: 0.6914145350456238. Accuracy: 55.465587044534416\n",
      "Train -  Loss: 0.6891250014305115. Accuracy: 53.09255079006772\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  97%|██████████████▌| 970169/1000000 [14:34<00:25, 1150.86it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 970000. \n",
      "Test - Loss: 0.6914052963256836. Accuracy: 55.465587044534416\n",
      "Train -  Loss: 0.6890980005264282. Accuracy: 53.09255079006772\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  98%|██████████████▋| 980146/1000000 [14:43<00:17, 1124.30it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 980000. \n",
      "Test - Loss: 0.6913957595825195. Accuracy: 55.06072874493927\n",
      "Train -  Loss: 0.689071774482727. Accuracy: 53.182844243792324\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs:  99%|██████████████▊| 990230/1000000 [14:52<00:08, 1160.10it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteration: 990000. \n",
      "Test - Loss: 0.6913856863975525. Accuracy: 55.06072874493927\n",
      "Train -  Loss: 0.6890462636947632. Accuracy: 53.137697516930025\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Training Epochs: 100%|██████████████| 1000000/1000000 [15:00<00:00, 1109.90it/s]\n"
     ]
    }
   ],
   "source": [
    "# ----------------- TRAIN ------------------\n",
    "losses = []\n",
    "losses_test = []\n",
    "acc = []\n",
    "acc_test =[]\n",
    "Iterations = []\n",
    "iter = 0\n",
    "step = 10000\n",
    "for epoch in tqdm(range(int(epochs)),desc='Training Epochs'):\n",
    "    x = X_train\n",
    "    labels = y_train\n",
    "    optimizer.zero_grad() # Setting our stored gradients equal to zero\n",
    "    outputs = model(X_train)\n",
    "    loss = criterion(torch.squeeze(outputs), labels) \n",
    "    \n",
    "    loss.backward() # Computes the gradient of the given tensor w.r.t. the weights/bias\n",
    "    \n",
    "    optimizer.step() # Updates weights and biases with the optimizer (SGD)\n",
    "    \n",
    "    # Print out benchmarking\n",
    "    if iter%step==0:\n",
    "        with torch.no_grad():\n",
    "            # Calculating the loss and accuracy for the test dataset\n",
    "            correct_test = 0\n",
    "            total_test = 0\n",
    "            outputs_test = torch.squeeze(model(X_test))\n",
    "            loss_test = criterion(outputs_test, y_test)\n",
    "            \n",
    "            predicted_test = outputs_test.cpu().round().detach().numpy()\n",
    "            total_test += y_test.size(0)\n",
    "            correct_test += np.sum(predicted_test == y_test.cpu().detach().numpy())\n",
    "            accuracy_test = 100 * correct_test/total_test\n",
    "            acc_test.append(accuracy_test)\n",
    "            losses_test.append(loss_test.item())\n",
    "            \n",
    "            # Calculating the loss and accuracy for the train dataset\n",
    "            total = 0\n",
    "            correct = 0\n",
    "            total += y_train.size(0)\n",
    "            correct += np.sum(torch.squeeze(outputs).cpu().round().detach().numpy() == y_train.cpu().detach().numpy())\n",
    "            accuracy = 100 * correct/total\n",
    "            acc.append(accuracy)\n",
    "            losses.append(loss.item())\n",
    "            Iterations.append(iter)\n",
    "            \n",
    "            print(f\"Iteration: {iter}. \\nTest - Loss: {loss_test.item()}. Accuracy: {accuracy_test}\")\n",
    "            print(f\"Train -  Loss: {loss.item()}. Accuracy: {accuracy}\\n\")\n",
    "        \n",
    "    iter+=1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "011a43cf",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f11cb865b10>"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# ------------------- PLOT RESULTS -------------------\n",
    "fig, (ax1, ax2) = plt.subplots(2, 1)\n",
    "\n",
    "t = np.arange(epochs, step=step)\n",
    "\n",
    "ax1.plot(t, losses,'r', label='train')\n",
    "ax1.plot(t, losses_test, 'b', label='test')\n",
    "ax1.set_xlabel('epochs')\n",
    "ax1.set_ylabel('loss')\n",
    "ax1.legend()\n",
    "\n",
    "ax2.plot(t, acc,'r', label='train')\n",
    "ax2.plot(t, acc_test, 'b', label='test')\n",
    "ax2.set_xlabel('epochs')\n",
    "ax2.set_ylabel('accuracy')\n",
    "ax2.legend()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d423e7d1",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}