Skip to content

Instantly share code, notes, and snippets.

@sagorbrur

sagorbrur/simple_pytorch.ipynb

Created September 24, 2021 17:34
Show Gist options
  • Save sagorbrur/95a2c2370a18a1b3fae337dd06f65b9b to your computer and use it in GitHub Desktop.
Save sagorbrur/95a2c2370a18a1b3fae337dd06f65b9b to your computer and use it in GitHub Desktop.
Download ZIP
simple_pytorch.ipynb
Display the source blob
Display the rendered blob
Raw
simple_pytorch.ipynb
{
"nbformat": 4,
"nbformat_minor": 0,
"metadata": {
"colab": {
"name": "simple_pytorch.ipynb",
"provenance": [],
"collapsed_sections": [],
"authorship_tag": "ABX9TyNU1iA/twF8+dcyHE5G2TsM",
"include_colab_link": true
},
"kernelspec": {
"name": "python3",
"display_name": "Python 3"
},
"language_info": {
"name": "python"
},
"accelerator": "GPU"
},
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "view-in-github",
"colab_type": "text"
},
"source": [
"<a href=\"https://colab.research.google.com/gist/sagorbrur/95a2c2370a18a1b3fae337dd06f65b9b/simple_pytorch.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "Morez129upER"
},
"source": [
"## Import Necessaries"
]
},
{
"cell_type": "code",
"metadata": {
"id": "kfuKotgBsvnC"
},
"source": [
"import torch\n",
"import torch.nn as nn\n",
"import torch.optim as optim\n",
"from torch.utils.data import DataLoader"
],
"execution_count": 1,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "NeHXixMHusPf"
},
"source": [
"## Set Device"
]
},
{
"cell_type": "code",
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "JIBpIOtxs1en",
"outputId": "19c2ffa7-1e3d-4b5f-b4b0-c2343dd25aab"
},
"source": [
"device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
"print(device)"
],
"execution_count": 2,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"cuda\n"
]
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "_HLHIGHzuuBo"
},
"source": [
"## Defince and Prepare Data"
]
},
{
"cell_type": "code",
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "Zkbe-vYCs5Xs",
"outputId": "abb4bdd1-841a-4008-943e-5e7510da76c6"
},
"source": [
"# device a sample trainset\n",
"traindata = torch.tensor([[1, 2, 3], [2, 3, 4], [5, 7, 4], [9, 1, 4]], dtype=torch.float32)\n",
"# map traindata in cuda device or cpu\n",
"traindata = traindata.to(device)\n",
"# device train label\n",
"trainlabel = torch.LongTensor([0, 1, 0, 1])\n",
"# map label to cuda or cpu\n",
"trainlabel = trainlabel.to(device)\n",
"print(traindata)\n",
"print(trainlabel)\n",
"\n",
"# define sample test set\n",
"testdata = torch.tensor([[1, 2, 3], [2, 3, 4], [5, 7, 4], [9, 1, 4]], dtype=torch.float32)\n",
"# map testdata in cuda device or cpu\n",
"testdata = testdata.to(device)\n",
"# device test label\n",
"testlabel = torch.LongTensor([0, 1, 0, 1])\n",
"# map label to cuda or cpu\n",
"testlabel = testlabel.to(device)\n",
"\n",
"# convert traindata and label to example\n",
"def generate_data_example(data, label):\n",
" examples = []\n",
" for d, l in zip(data, label):\n",
" data_set = (d, l)\n",
" examples.append(data_set)\n",
" return examples\n",
" \n",
"train_examples = generate_data_example(traindata, trainlabel)\n",
"test_examples = generate_data_example(testdata, testlabel)\n",
"\n",
"# fit in dataloader\n",
"batch_size = 1\n",
"train_dataloader = DataLoader(train_examples, batch_size=batch_size)\n",
"test_dataloader = DataLoader(test_examples, batch_size=batch_size)\n",
"print(f\"train data: {len(train_dataloader)} and test data: {len(test_dataloader)}\")"
],
"execution_count": 3,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"tensor([[1., 2., 3.],\n",
" [2., 3., 4.],\n",
" [5., 7., 4.],\n",
" [9., 1., 4.]], device='cuda:0')\n",
"tensor([0, 1, 0, 1], device='cuda:0')\n",
"train data: 4 and test data: 4\n"
]
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "LZL-CVMAuwqX"
},
"source": [
"## Define Model"
]
},
{
"cell_type": "code",
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "aTzsotOBtFj4",
"outputId": "2efba92c-8fbc-46f2-fa26-54a35bff6dde"
},
"source": [
"class Model(nn.Module):\n",
" def __init__(self, input_size, output_size):\n",
" super(Model, self).__init__()\n",
" self.input_size = input_size\n",
" self.output_size = output_size\n",
" self.linear = nn.Linear(input_size, output_size)\n",
"\n",
" def forward(self, x):\n",
" out = self.linear(x)\n",
" return out\n",
" \n",
"input_size = traindata.shape\n",
"print(input_size[1])\n",
"model = Model(input_size[1], 2)\n",
"model.to(device)\n",
"print(model)"
],
"execution_count": 4,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"3\n",
"Model(\n",
" (linear): Linear(in_features=3, out_features=2, bias=True)\n",
")\n"
]
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "ZNz9LHYGu0Ha"
},
"source": [
"## Define Loss"
]
},
{
"cell_type": "code",
"metadata": {
"id": "sd4CUF28tmIg"
},
"source": [
"criterion = nn.CrossEntropyLoss()"
],
"execution_count": 5,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "2aId5ff7u1u_"
},
"source": [
"## Define Optimizer"
]
},
{
"cell_type": "code",
"metadata": {
"id": "d4L9wP1Xto_B"
},
"source": [
"optimizer = optim.Adam(model.parameters(), lr=0.001)"
],
"execution_count": 6,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "o-KMlwNNu3iS"
},
"source": [
"## Define Train Function"
]
},
{
"cell_type": "code",
"metadata": {
"id": "Qrpw2kUotsM4"
},
"source": [
"def train(dataloader, model, loss_fn, optimizer):\n",
" size = len(dataloader.dataset)\n",
" model.train()\n",
" for batch, (X, y) in enumerate(dataloader):\n",
" X, y = X.to(device), y.to(device)\n",
"\n",
" # Compute prediction error\n",
" pred = model(X)\n",
" loss = loss_fn(pred, y)\n",
"\n",
" # Backpropagation\n",
" optimizer.zero_grad()\n",
" loss.backward()\n",
" optimizer.step()\n",
"\n",
" if batch % 1 == 0:\n",
" loss, current = loss.item(), batch * len(X)\n",
" print(f\"loss: {loss:>7f} [{current:>5d}/{size:>5d}]\")"
],
"execution_count": 7,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "MdZDBavCu6f7"
},
"source": [
"## Define Test Function"
]
},
{
"cell_type": "code",
"metadata": {
"id": "scrbT9eUtzCf"
},
"source": [
"def test(dataloader, model, loss_fn):\n",
" size = len(dataloader.dataset)\n",
" num_batches = len(dataloader)\n",
" model.eval()\n",
" test_loss, correct = 0, 0\n",
" with torch.no_grad():\n",
" for X, y in dataloader:\n",
" X, y = X.to(device), y.to(device)\n",
" pred = model(X)\n",
" test_loss += loss_fn(pred, y).item()\n",
" correct += (pred.argmax(1) == y).type(torch.float).sum().item()\n",
" test_loss /= num_batches\n",
" correct /= size\n",
" print(f\"Test Error: \\n Accuracy: {(100*correct):>0.1f}%, Avg loss: {test_loss:>8f} \\n\")"
],
"execution_count": 8,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "SDKdp3Ymu8xQ"
},
"source": [
"## Train and Test"
]
},
{
"cell_type": "code",
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "eiEF-pIRt1zq",
"outputId": "52df69df-051b-4926-bf61-4cd6757828d4"
},
"source": [
"epochs = 5\n",
"for t in range(epochs):\n",
" print(f\"Epoch {t+1}\\n-------------------------------\")\n",
" train(train_dataloader, model, criterion, optimizer)\n",
" test(test_dataloader, model, criterion)\n",
"print(\"Done!\")"
],
"execution_count": 9,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"Epoch 1\n",
"-------------------------------\n",
"loss: 0.468978 [ 0/ 4]\n",
"loss: 1.550761 [ 1/ 4]\n",
"loss: 0.021035 [ 2/ 4]\n",
"loss: 6.216814 [ 3/ 4]\n",
"Test Error: \n",
" Accuracy: 50.0%, Avg loss: 2.054861 \n",
"\n",
"Epoch 2\n",
"-------------------------------\n",
"loss: 0.471171 [ 0/ 4]\n",
"loss: 1.522439 [ 1/ 4]\n",
"loss: 0.022363 [ 2/ 4]\n",
"loss: 6.155248 [ 3/ 4]\n",
"Test Error: \n",
" Accuracy: 50.0%, Avg loss: 2.034846 \n",
"\n",
"Epoch 3\n",
"-------------------------------\n",
"loss: 0.481420 [ 0/ 4]\n",
"loss: 1.491253 [ 1/ 4]\n",
"loss: 0.023894 [ 2/ 4]\n",
"loss: 6.090882 [ 3/ 4]\n",
"Test Error: \n",
" Accuracy: 50.0%, Avg loss: 2.014044 \n",
"\n",
"Epoch 4\n",
"-------------------------------\n",
"loss: 0.492393 [ 0/ 4]\n",
"loss: 1.459543 [ 1/ 4]\n",
"loss: 0.025564 [ 2/ 4]\n",
"loss: 6.025702 [ 3/ 4]\n",
"Test Error: \n",
" Accuracy: 50.0%, Avg loss: 1.993133 \n",
"\n",
"Epoch 5\n",
"-------------------------------\n",
"loss: 0.503710 [ 0/ 4]\n",
"loss: 1.427929 [ 1/ 4]\n",
"loss: 0.027358 [ 2/ 4]\n",
"loss: 5.960329 [ 3/ 4]\n",
"Test Error: \n",
" Accuracy: 50.0%, Avg loss: 1.972324 \n",
"\n",
"Done!\n"
]
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "rrJ2Izm3vAFv"
},
"source": [
"## Save Model"
]
},
{
"cell_type": "code",
"metadata": {
"id": "7S3deFI5t4yR"
},
"source": [
"torch.save(model.state_dict(), 'model.pth')"
],
"execution_count": 10,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "8OGgoCxbvCQ2"
},
"source": [
"## Load and Inference"
]
},
{
"cell_type": "code",
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "FnxQin0FuIV6",
"outputId": "8a6a6881-fa37-4c6a-8b6d-772be24e2ede"
},
"source": [
"load_model = model.load_state_dict(torch.load('model.pth'))\n",
"test_input = torch.tensor([[1, 2, 3]], dtype=torch.float32).to(device)\n",
"predict = model(test_input).data.max(1, keepdim=True)[1]\n",
"print(predict)"
],
"execution_count": 11,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"tensor([[0]], device='cuda:0')\n"
]
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "uALWJA6svEcG"
},
"source": [
"## Inspect Model"
]
},
{
"cell_type": "code",
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/"
},
"id": "inm_nhNZuLlI",
"outputId": "88b55d17-9935-4cd3-d4ec-294cfec832a7"
},
"source": [
"# inspect parameters\n",
"params = model.parameters()\n",
"for param in params:\n",
" print(param)\n",
"modules = model.named_modules()\n",
"for idx, m in enumerate(modules):\n",
" print(idx, m)"
],
"execution_count": 12,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"Parameter containing:\n",
"tensor([[ 0.1549, 0.1563, -0.2369],\n",
" [-0.5547, 0.0815, -0.1797]], device='cuda:0', requires_grad=True)\n",
"Parameter containing:\n",
"tensor([0.2673, 0.5606], device='cuda:0', requires_grad=True)\n",
"0 ('', Model(\n",
" (linear): Linear(in_features=3, out_features=2, bias=True)\n",
"))\n",
"1 ('linear', Linear(in_features=3, out_features=2, bias=True))\n"
]
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "CFyZhzW9uekJ",
"colab": {
"base_uri": "https://localhost:8080/"
},
"outputId": "d937a950-a270-47c9-a192-06e129305dc7"
},
"source": [
"# Print model's state_dict\n",
"print(\"Model's state_dict:\")\n",
"for param_tensor in model.state_dict():\n",
" print(param_tensor, \"\\t\", model.state_dict()[param_tensor].size())\n",
"\n",
"# Print optimizer's state_dict\n",
"print(\"Optimizer's state_dict:\")\n",
"for var_name in optimizer.state_dict():\n",
" print(var_name, \"\\t\", optimizer.state_dict()[var_name])"
],
"execution_count": 13,
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"Model's state_dict:\n",
"linear.weight \t torch.Size([2, 3])\n",
"linear.bias \t torch.Size([2])\n",
"Optimizer's state_dict:\n",
"state \t {0: {'step': 20, 'exp_avg': tensor([[ 2.5106, 0.5475, 1.4160],\n",
" [-2.5106, -0.5475, -1.4160]], device='cuda:0'), 'exp_avg_sq': tensor([[0.4126, 0.0347, 0.1331],\n",
" [0.4126, 0.0347, 0.1331]], device='cuda:0')}, 1: {'step': 20, 'exp_avg': tensor([ 0.3360, -0.3360], device='cuda:0'), 'exp_avg_sq': tensor([0.0086, 0.0086], device='cuda:0')}}\n",
"param_groups \t [{'lr': 0.001, 'betas': (0.9, 0.999), 'eps': 1e-08, 'weight_decay': 0, 'amsgrad': False, 'params': [0, 1]}]\n"
]
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "fsNDlygO3yV3"
},
"source": [
""
],
"execution_count": null,
"outputs": []
}
]
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment