mnist-classification-minimal.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0, Loss: 0.1344318389892578\n",
      "Epoch: 1, Loss: 0.10877793282270432\n",
      "Epoch: 2, Loss: 0.1007857620716095\n",
      "Epoch: 3, Loss: 0.06776227056980133\n",
      "Epoch: 4, Loss: 0.03420461341738701\n",
      "Epoch: 5, Loss: 0.11107035726308823\n",
      "Epoch: 6, Loss: 0.028283704072237015\n",
      "Epoch: 7, Loss: 0.01456095464527607\n",
      "Epoch: 8, Loss: 0.0457158200442791\n",
      "Epoch: 9, Loss: 0.030669715255498886\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "tensor(0.9742)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import torch\n",
    "import torchvision as tv\n",
    "\n",
    "t = tv.transforms.ToTensor()\n",
    "\n",
    "mnist_training = tv.datasets.MNIST(\n",
    "    root='/tmp/mnist', \n",
    "    train=True, \n",
    "    download=True, \n",
    "    transform=t\n",
    ")\n",
    "\n",
    "mnist_val = tv.datasets.MNIST(\n",
    "    root='/tmp/mnist', \n",
    "    train=False, \n",
    "    download=True, \n",
    "    transform=t\n",
    ")\n",
    "\n",
    "model = torch.nn.Sequential(\n",
    "    torch.nn.Linear(28*28, 128),\n",
    "    torch.nn.ReLU(),\n",
    "    torch.nn.Linear(128, 10)\n",
    ")\n",
    "\n",
    "opt = torch.optim.Adam(params=model.parameters(), lr=0.01)\n",
    "\n",
    "loss_fn = torch.nn.CrossEntropyLoss()\n",
    "\n",
    "loader = torch.utils.data.DataLoader(\n",
    "    mnist_training, \n",
    "    batch_size=500, \n",
    "    shuffle=True\n",
    ")\n",
    "\n",
    "for epoch in range(10):\n",
    "    for imgs, labels in loader:\n",
    "        n = len(imgs)\n",
    "        imgs = imgs.view(n, -1)\n",
    "        predictions = model(imgs)  \n",
    "        loss = loss_fn(predictions, labels) \n",
    "        opt.zero_grad()\n",
    "        loss.backward()\n",
    "        opt.step()\n",
    "    print(f\"Epoch: {epoch}, Loss: {float(loss)}\")\n",
    "\n",
    "n = 10000\n",
    "loader = torch.utils.data.DataLoader(mnist_val, batch_size=n)\n",
    "images, labels = iter(loader).next()\n",
    "\n",
    "predictions = model(images.view(n, -1))\n",
    "\n",
    "predicted_labels = predictions.argmax(dim=1)\n",
    "torch.sum(predicted_labels == labels) / n"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "0b567db37d44c4d7108c6daa0f9e0ef92bd52d42f7ebb7e5d1452f34c4acbe1f"
  },
  "kernelspec": {
   "display_name": "Python 3.8.10 ('.venv': venv)",
   "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": "3.8.10"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}