diff --git a/TD2 Deep Learning.ipynb b/TD2 Deep Learning.ipynb
index 48543b667566028121d970d5b10f580f839f0ff7..2194e766254ad2dca2af828099b149926d889bf2 100644
--- a/TD2 Deep Learning.ipynb
+++ b/TD2 Deep Learning.ipynb
@@ -2145,6 +2145,365 @@
"Apply ther quantization (post and quantization aware) and evaluate impact on model size and accuracy."
]
},
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "# Modified code \n",
+ "import copy\n",
+ "import os\n",
+ "import time\n",
+ "\n",
+ "import matplotlib.pyplot as plt\n",
+ "import numpy as np\n",
+ "import torch\n",
+ "import torch.nn as nn\n",
+ "import torch.optim as optim\n",
+ "import torchvision\n",
+ "from torch.optim import lr_scheduler\n",
+ "from torchvision import datasets, transforms\n",
+ "\n",
+ "# Data augmentation and normalization for training\n",
+ "# Just normalization for validation\n",
+ "data_transforms = {\n",
+ " \"train\": transforms.Compose(\n",
+ " [\n",
+ " transforms.RandomResizedCrop(224),\n",
+ " transforms.RandomHorizontalFlip(),\n",
+ " transforms.ToTensor(),\n",
+ " transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),\n",
+ " ]\n",
+ " ),\n",
+ " \"val\": transforms.Compose(\n",
+ " [\n",
+ " transforms.Resize(256),\n",
+ " transforms.CenterCrop(224),\n",
+ " transforms.ToTensor(),\n",
+ " transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),\n",
+ " ]\n",
+ " ),\n",
+ " \"test\": transforms.Compose(\n",
+ " [\n",
+ " transforms.Resize(256),\n",
+ " transforms.CenterCrop(224),\n",
+ " transforms.ToTensor(),\n",
+ " transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),\n",
+ " ]\n",
+ " ),\n",
+ "}\n",
+ "\n",
+ "data_dir = \"hymenoptera_data\"\n",
+ "# Create train and validation datasets and loaders\n",
+ "image_datasets = {\n",
+ " x: datasets.ImageFolder(os.path.join(data_dir, x), data_transforms[x])\n",
+ " for x in [\"train\", \"val\",\"test\"]\n",
+ "}\n",
+ "dataloaders = {\n",
+ " x: torch.utils.data.DataLoader(\n",
+ " image_datasets[x], batch_size=4, shuffle=True, num_workers=4\n",
+ " )\n",
+ " for x in [\"train\", \"val\",\"test\"]\n",
+ "}\n",
+ "dataset_sizes = {x: len(image_datasets[x]) for x in [\"train\", \"val\", \"test\"]}\n",
+ "class_names = image_datasets[\"train\"].classes\n",
+ "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
+ "\n",
+ "# Helper function for displaying images\n",
+ "def imshow(inp, title=None):\n",
+ " \"\"\"Imshow for Tensor.\"\"\"\n",
+ " inp = inp.numpy().transpose((1, 2, 0))\n",
+ " mean = np.array([0.485, 0.456, 0.406])\n",
+ " std = np.array([0.229, 0.224, 0.225])\n",
+ "\n",
+ " # Un-normalize the images\n",
+ " inp = std * inp + mean\n",
+ " # Clip just in case\n",
+ " inp = np.clip(inp, 0, 1)\n",
+ " plt.imshow(inp)\n",
+ " if title is not None:\n",
+ " plt.title(title)\n",
+ " plt.pause(0.001) # pause a bit so that plots are updated\n",
+ " plt.show()\n",
+ "\n",
+ "\n",
+ "# Get a batch of training data\n",
+ "# inputs, classes = next(iter(dataloaders['train']))\n",
+ "\n",
+ "# Make a grid from batch\n",
+ "# out = torchvision.utils.make_grid(inputs)\n",
+ "\n",
+ "# imshow(out, title=[class_names[x] for x in classes])\n",
+ "# training\n",
+ "\n",
+ "\n",
+ "def train_model(model, criterion, optimizer, scheduler, num_epochs=25):\n",
+ " since = time.time()\n",
+ "\n",
+ " best_model_wts = copy.deepcopy(model.state_dict())\n",
+ " best_acc = 0.0\n",
+ "\n",
+ " epoch_time = [] # we'll keep track of the time needed for each epoch\n",
+ "\n",
+ " for epoch in range(num_epochs):\n",
+ " epoch_start = time.time()\n",
+ " print(\"Epoch {}/{}\".format(epoch + 1, num_epochs))\n",
+ " print(\"-\" * 10)\n",
+ "\n",
+ " # Each epoch has a training and validation phase\n",
+ " for phase in [\"train\", \"val\"]:\n",
+ " if phase == \"train\":\n",
+ " scheduler.step()\n",
+ " model.train() # Set model to training mode\n",
+ " else:\n",
+ " model.eval() # Set model to evaluate mode\n",
+ "\n",
+ " running_loss = 0.0\n",
+ " running_corrects = 0\n",
+ "\n",
+ " # Iterate over data.\n",
+ " for inputs, labels in dataloaders[phase]:\n",
+ " inputs = inputs.to(device)\n",
+ " labels = labels.to(device)\n",
+ "\n",
+ " # zero the parameter gradients\n",
+ " optimizer.zero_grad()\n",
+ "\n",
+ " # Forward\n",
+ " # Track history if only in training phase\n",
+ " with torch.set_grad_enabled(phase == \"train\"):\n",
+ " outputs = model(inputs)\n",
+ " _, preds = torch.max(outputs, 1)\n",
+ " loss = criterion(outputs, labels)\n",
+ "\n",
+ " # backward + optimize only if in training phase\n",
+ " if phase == \"train\":\n",
+ " loss.backward()\n",
+ " optimizer.step()\n",
+ "\n",
+ " # Statistics\n",
+ " running_loss += loss.item() * inputs.size(0)\n",
+ " running_corrects += torch.sum(preds == labels.data)\n",
+ "\n",
+ " epoch_loss = running_loss / dataset_sizes[phase]\n",
+ " epoch_acc = running_corrects.double() / dataset_sizes[phase]\n",
+ "\n",
+ " print(\"{} Loss: {:.4f} Acc: {:.4f}\".format(phase, epoch_loss, epoch_acc))\n",
+ "\n",
+ " # Deep copy the model\n",
+ " if phase == \"val\" and epoch_acc > best_acc:\n",
+ " best_acc = epoch_acc\n",
+ " best_model_wts = copy.deepcopy(model.state_dict())\n",
+ "\n",
+ " # Add the epoch time\n",
+ " t_epoch = time.time() - epoch_start\n",
+ " epoch_time.append(t_epoch)\n",
+ " print()\n",
+ "\n",
+ " time_elapsed = time.time() - since\n",
+ " print(\n",
+ " \"Training complete in {:.0f}m {:.0f}s\".format(\n",
+ " time_elapsed // 60, time_elapsed % 60\n",
+ " )\n",
+ " )\n",
+ " print(\"Best val Acc: {:4f}\".format(best_acc))\n",
+ "\n",
+ " # Load best model weights\n",
+ " model.load_state_dict(best_model_wts)\n",
+ "\n",
+ " return model, epoch_time\n",
+ "\n",
+ "\n",
+ "# Download a pre-trained ResNet18 model and freeze its weights\n",
+ "model = torchvision.models.resnet18(pretrained=True)\n",
+ "for param in model.parameters():\n",
+ " param.requires_grad = False\n",
+ "\n",
+ "# Replace the final fully connected layer\n",
+ "# Parameters of newly constructed modules have requires_grad=True by default\n",
+ "num_ftrs = model.fc.in_features\n",
+ "model.fc = nn.Linear(num_ftrs, 2)\n",
+ "# Send the model to the GPU\n",
+ "model = model.to(device)\n",
+ "# Set the loss function\n",
+ "criterion = nn.CrossEntropyLoss()\n",
+ "\n",
+ "# Observe that only the parameters of the final layer are being optimized\n",
+ "optimizer_conv = optim.SGD(model.fc.parameters(), lr=0.001, momentum=0.9)\n",
+ "exp_lr_scheduler = lr_scheduler.StepLR(optimizer_conv, step_size=7, gamma=0.1)\n",
+ "model, epoch_time = train_model(\n",
+ " model, criterion, optimizer_conv, exp_lr_scheduler, num_epochs=10\n",
+ ")\n",
+ "\n",
+ "# Evaluating the model \n"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "\n",
+ "\n",
+ "# track test loss\n",
+ "test_loss = 0.0\n",
+ "class_correct = list(0.0 for i in range(10))\n",
+ "class_total = list(0.0 for i in range(10))\n",
+ "\n",
+ "model.eval()\n",
+ "# iterate over test data\n",
+ "for data, target in image_datasets[\"test\"]:\n",
+ " # move tensors to GPU if CUDA is available\n",
+ " if train_on_gpu:\n",
+ " data, target = data.cuda(), target.cuda()\n",
+ " # forward pass: compute predicted outputs by passing inputs to the model\n",
+ " output = model(data)\n",
+ " # calculate the batch loss\n",
+ " loss = criterion(output, target)\n",
+ " # update test loss\n",
+ " test_loss += loss.item() * data.size(0)\n",
+ " # convert output probabilities to predicted class\n",
+ " _, pred = torch.max(output, 1)\n",
+ " # compare predictions to true label\n",
+ " correct_tensor = pred.eq(target.data.view_as(pred))\n",
+ " correct = (\n",
+ " np.squeeze(correct_tensor.numpy())\n",
+ " if not train_on_gpu\n",
+ " else np.squeeze(correct_tensor.cpu().numpy())\n",
+ " )\n",
+ " # calculate test accuracy for each object class\n",
+ " for i in range(batch_size):\n",
+ " label = target.data[i]\n",
+ " class_correct[label] += correct[i].item()\n",
+ " class_total[label] += 1\n",
+ "\n",
+ "# average test loss\n",
+ "test_loss = test_loss / len(test_loader)\n",
+ "print(\"Test Loss: {:.6f}\\n\".format(test_loss))\n",
+ "\n",
+ "for i in range(10):\n",
+ " if class_total[i] > 0:\n",
+ " print(\n",
+ " \"Test Accuracy of %5s: %2d%% (%2d/%2d)\"\n",
+ " % (\n",
+ " classes[i],\n",
+ " 100 * class_correct[i] / class_total[i],\n",
+ " np.sum(class_correct[i]),\n",
+ " np.sum(class_total[i]),\n",
+ " )\n",
+ " )\n",
+ " else:\n",
+ " print(\"Test Accuracy of %5s: N/A (no training examples)\" % (classes[i]))\n",
+ "\n",
+ "print(\n",
+ " \"\\nTest Accuracy (Overall): %2d%% (%2d/%2d)\"\n",
+ " % (\n",
+ " 100.0 * np.sum(class_correct) / np.sum(class_total),\n",
+ " np.sum(class_correct),\n",
+ " np.sum(class_total),\n",
+ " )\n",
+ ")"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "import torch.quantization\n",
+ "\n",
+ "# Post quantization\n",
+ "\n",
+ "quantized_model = torch.quantization.quantize_dynamic(model, dtype=torch.qint8)\n",
+ "print_size_of_model(quantized_model, \"int8\")"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "# Aware quantization\n",
+ "my_model = model\n",
+ "my_model.qconfig = torch.ao.quantization.get_default_qat_qconfig('fbgemm')\n",
+ "my_model = torch.ao.quantization.prepare_qat(my_model)\n",
+ "epochquantized_model=torch.quantization.convert(my_model.eval(), inplace=False)\n",
+ "\n",
+ "import torch.optim as optim\n",
+ "\n",
+ "criterion = nn.CrossEntropyLoss() # specify loss function\n",
+ "optimizer = optim.SGD(my_model.parameters(), lr=0.01) # specify optimizer\n",
+ "\n",
+ "n_epochs = 30 # number of epochs to train the model\n",
+ "train_loss_list = [] # list to store loss to visualize\n",
+ "valid_loss_min = np.Inf # track change in validation loss\n",
+ "valid_loss_list = []\n",
+ "\n",
+ "for epoch in range(n_epochs):\n",
+ " # Keep track of training and validation loss\n",
+ " train_loss = 0.0\n",
+ " valid_loss = 0.0\n",
+ "\n",
+ " # Train the model\n",
+ " my_model.train()\n",
+ " for data, target in train_loader:\n",
+ " # Move tensors to GPU if CUDA is available\n",
+ " if train_on_gpu:\n",
+ " data, target = data.cuda(), target.cuda()\n",
+ " # Clear the gradients of all optimized variables\n",
+ " optimizer.zero_grad()\n",
+ " # Forward pass: compute predicted outputs by passing inputs to the model\n",
+ " output = my_model(data)\n",
+ " # Calculate the batch loss\n",
+ " loss = criterion(output, target)\n",
+ " # Backward pass: compute gradient of the loss with respect to model parameters\n",
+ " loss.backward()\n",
+ " # Perform a single optimization step (parameter update)\n",
+ " optimizer.step()\n",
+ " # Update training loss\n",
+ " train_loss += loss.item() * data.size(0)\n",
+ "\n",
+ " # Validate the model\n",
+ " my_model.eval()\n",
+ " for data, target in valid_loader:\n",
+ " # Move tensors to GPU if CUDA is available\n",
+ " if train_on_gpu:\n",
+ " data, target = data.cuda(), target.cuda()\n",
+ " # Forward pass: compute predicted outputs by passing inputs to the model\n",
+ " output = my_model(data)\n",
+ " # Calculate the batch loss\n",
+ " loss = criterion(output, target)\n",
+ " # Update average validation loss\n",
+ " valid_loss += loss.item() * data.size(0)\n",
+ "\n",
+ " # Calculate average losses\n",
+ " train_loss = train_loss / len(train_loader)\n",
+ " valid_loss = valid_loss / len(valid_loader)\n",
+ " train_loss_list.append(train_loss)\n",
+ " valid_loss_list.append(valid_loss)\n",
+ "\n",
+ " # Print training/validation statistics\n",
+ " print(\n",
+ " \"Epoch: {} \\tTraining Loss: {:.6f} \\tValidation Loss: {:.6f}\".format(\n",
+ " epoch, train_loss, valid_loss\n",
+ " )\n",
+ " )\n",
+ "\n",
+ " # Save model if validation loss has decreased\n",
+ " if valid_loss <= valid_loss_min:\n",
+ " print(\n",
+ " \"Validation loss decreased ({:.6f} --> {:.6f}). Saving model ...\".format(\n",
+ " valid_loss_min, valid_loss\n",
+ " )\n",
+ " )\n",
+ " torch.save(my_model.state_dict(), \"my_model_cifar.pt\")\n",
+ " valid_loss_min = valid_loss"
+ ]
+ },
{
"cell_type": "markdown",
"id": "04a263f0",