{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "## Download the model that you want to fine-tune from the Hugging Face model hub" ] }, { "cell_type": "code", "execution_count": 129, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "['densenet161', 'S', 'V3.pth']\n", "Model densenet161_S_V3.pth already exists in local directory. Skipping download.\n" ] } ], "source": [ "from huggingface_hub import hf_hub_download\n", "import os \n", "from mescnn.classification.inference.download import download_classifier\n", "\n", "# Replace this with the path to your local directory\n", "local_models_dir = \"/home/wfd/Desktop/Projet_M1/FineTuning/Models\"\n", "\n", "\n", "lesion = \"S\"\n", "\n", "# Download the densenet121_C_V3.pth model\n", "if (lesion == \"C\"):\n", " model_id = f\"mobilenetv2_{lesion}_V3.pth\" # For C lesion\n", "elif (lesion == \"S\"):\n", " model_id = f\"densenet161_{lesion}_V3.pth\" # For M & E lesion\n", "else:\n", " model_id = f\"efficientnetv2-m_{lesion}_V3.pth\" # For M & E lesion\n", "\n", "\n", "model_name = model_id.split(\"/\")[-1]\n", "splited_model_name = model_name.split(\"_\")\n", "print(splited_model_name)\n", "local_model_path = os.path.join(local_models_dir, model_name)\n", "\n", "\n", "if not os.path.exists(local_model_path):\n", " print(\"Downloading model to {}...\".format(local_model_path))\n", " downloaded_model = download_classifier(splited_model_name[0], splited_model_name[1], splited_model_name[2].split(\".\")[0])\n", " # Move the model from the download directory to the local directory\n", " os.rename(downloaded_model, local_model_path)\n", " print(\"Download complete.\")\n", "else:\n", " print(\"Model {} already exists in local directory. Skipping download.\".format(model_id))" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Fine-tuning the model " ] }, { "cell_type": "code", "execution_count": 130, "metadata": {}, "outputs": [], "source": [ "# Load the efficientnet C model\n", "import torch\n", "net = torch.load(local_model_path)" ] }, { "cell_type": "code", "execution_count": 131, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Sequential(\n", " (0): Sequential(\n", " (conv0): Conv2d(3, 96, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)\n", " (norm0): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu0): ReLU(inplace=True)\n", " (pool0): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n", " (denseblock1): _DenseBlock(\n", " (denselayer1): _DenseLayer(\n", " (norm1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(96, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer2): _DenseLayer(\n", " (norm1): BatchNorm2d(144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(144, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer3): _DenseLayer(\n", " (norm1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(192, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer4): _DenseLayer(\n", " (norm1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(240, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer5): _DenseLayer(\n", " (norm1): BatchNorm2d(288, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(288, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer6): _DenseLayer(\n", " (norm1): BatchNorm2d(336, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(336, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " )\n", " (transition1): _Transition(\n", " (norm): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu): ReLU(inplace=True)\n", " (conv): Conv2d(384, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (pool): AvgPool2d(kernel_size=2, stride=2, padding=0)\n", " )\n", " (denseblock2): _DenseBlock(\n", " (denselayer1): _DenseLayer(\n", " (norm1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(192, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer2): _DenseLayer(\n", " (norm1): BatchNorm2d(240, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(240, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer3): _DenseLayer(\n", " (norm1): BatchNorm2d(288, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(288, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer4): _DenseLayer(\n", " (norm1): BatchNorm2d(336, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(336, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer5): _DenseLayer(\n", " (norm1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(384, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer6): _DenseLayer(\n", " (norm1): BatchNorm2d(432, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(432, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer7): _DenseLayer(\n", " (norm1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(480, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer8): _DenseLayer(\n", " (norm1): BatchNorm2d(528, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(528, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer9): _DenseLayer(\n", " (norm1): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(576, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer10): _DenseLayer(\n", " (norm1): BatchNorm2d(624, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(624, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer11): _DenseLayer(\n", " (norm1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(672, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer12): _DenseLayer(\n", " (norm1): BatchNorm2d(720, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(720, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " )\n", " (transition2): _Transition(\n", " (norm): BatchNorm2d(768, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu): ReLU(inplace=True)\n", " (conv): Conv2d(768, 384, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (pool): AvgPool2d(kernel_size=2, stride=2, padding=0)\n", " )\n", " (denseblock3): _DenseBlock(\n", " (denselayer1): _DenseLayer(\n", " (norm1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(384, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer2): _DenseLayer(\n", " (norm1): BatchNorm2d(432, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(432, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer3): _DenseLayer(\n", " (norm1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(480, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer4): _DenseLayer(\n", " (norm1): BatchNorm2d(528, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(528, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer5): _DenseLayer(\n", " (norm1): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(576, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer6): _DenseLayer(\n", " (norm1): BatchNorm2d(624, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(624, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer7): _DenseLayer(\n", " (norm1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(672, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer8): _DenseLayer(\n", " (norm1): BatchNorm2d(720, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(720, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer9): _DenseLayer(\n", " (norm1): BatchNorm2d(768, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(768, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer10): _DenseLayer(\n", " (norm1): BatchNorm2d(816, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(816, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer11): _DenseLayer(\n", " (norm1): BatchNorm2d(864, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(864, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer12): _DenseLayer(\n", " (norm1): BatchNorm2d(912, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(912, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer13): _DenseLayer(\n", " (norm1): BatchNorm2d(960, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(960, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer14): _DenseLayer(\n", " (norm1): BatchNorm2d(1008, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1008, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer15): _DenseLayer(\n", " (norm1): BatchNorm2d(1056, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1056, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer16): _DenseLayer(\n", " (norm1): BatchNorm2d(1104, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1104, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer17): _DenseLayer(\n", " (norm1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer18): _DenseLayer(\n", " (norm1): BatchNorm2d(1200, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1200, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer19): _DenseLayer(\n", " (norm1): BatchNorm2d(1248, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1248, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer20): _DenseLayer(\n", " (norm1): BatchNorm2d(1296, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1296, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer21): _DenseLayer(\n", " (norm1): BatchNorm2d(1344, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1344, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer22): _DenseLayer(\n", " (norm1): BatchNorm2d(1392, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1392, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer23): _DenseLayer(\n", " (norm1): BatchNorm2d(1440, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1440, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer24): _DenseLayer(\n", " (norm1): BatchNorm2d(1488, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1488, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer25): _DenseLayer(\n", " (norm1): BatchNorm2d(1536, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1536, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer26): _DenseLayer(\n", " (norm1): BatchNorm2d(1584, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1584, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer27): _DenseLayer(\n", " (norm1): BatchNorm2d(1632, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1632, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer28): _DenseLayer(\n", " (norm1): BatchNorm2d(1680, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1680, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer29): _DenseLayer(\n", " (norm1): BatchNorm2d(1728, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1728, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer30): _DenseLayer(\n", " (norm1): BatchNorm2d(1776, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1776, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer31): _DenseLayer(\n", " (norm1): BatchNorm2d(1824, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1824, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer32): _DenseLayer(\n", " (norm1): BatchNorm2d(1872, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1872, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer33): _DenseLayer(\n", " (norm1): BatchNorm2d(1920, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1920, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer34): _DenseLayer(\n", " (norm1): BatchNorm2d(1968, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1968, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer35): _DenseLayer(\n", " (norm1): BatchNorm2d(2016, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(2016, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer36): _DenseLayer(\n", " (norm1): BatchNorm2d(2064, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(2064, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " )\n", " (transition3): _Transition(\n", " (norm): BatchNorm2d(2112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu): ReLU(inplace=True)\n", " (conv): Conv2d(2112, 1056, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (pool): AvgPool2d(kernel_size=2, stride=2, padding=0)\n", " )\n", " (denseblock4): _DenseBlock(\n", " (denselayer1): _DenseLayer(\n", " (norm1): BatchNorm2d(1056, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1056, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer2): _DenseLayer(\n", " (norm1): BatchNorm2d(1104, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1104, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer3): _DenseLayer(\n", " (norm1): BatchNorm2d(1152, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1152, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer4): _DenseLayer(\n", " (norm1): BatchNorm2d(1200, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1200, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer5): _DenseLayer(\n", " (norm1): BatchNorm2d(1248, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1248, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer6): _DenseLayer(\n", " (norm1): BatchNorm2d(1296, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1296, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer7): _DenseLayer(\n", " (norm1): BatchNorm2d(1344, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1344, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer8): _DenseLayer(\n", " (norm1): BatchNorm2d(1392, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1392, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer9): _DenseLayer(\n", " (norm1): BatchNorm2d(1440, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1440, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer10): _DenseLayer(\n", " (norm1): BatchNorm2d(1488, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1488, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer11): _DenseLayer(\n", " (norm1): BatchNorm2d(1536, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1536, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer12): _DenseLayer(\n", " (norm1): BatchNorm2d(1584, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1584, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer13): _DenseLayer(\n", " (norm1): BatchNorm2d(1632, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1632, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer14): _DenseLayer(\n", " (norm1): BatchNorm2d(1680, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1680, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer15): _DenseLayer(\n", " (norm1): BatchNorm2d(1728, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1728, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer16): _DenseLayer(\n", " (norm1): BatchNorm2d(1776, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1776, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer17): _DenseLayer(\n", " (norm1): BatchNorm2d(1824, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1824, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer18): _DenseLayer(\n", " (norm1): BatchNorm2d(1872, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1872, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer19): _DenseLayer(\n", " (norm1): BatchNorm2d(1920, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1920, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer20): _DenseLayer(\n", " (norm1): BatchNorm2d(1968, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(1968, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer21): _DenseLayer(\n", " (norm1): BatchNorm2d(2016, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(2016, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer22): _DenseLayer(\n", " (norm1): BatchNorm2d(2064, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(2064, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer23): _DenseLayer(\n", " (norm1): BatchNorm2d(2112, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(2112, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " (denselayer24): _DenseLayer(\n", " (norm1): BatchNorm2d(2160, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu1): ReLU(inplace=True)\n", " (conv1): Conv2d(2160, 192, kernel_size=(1, 1), stride=(1, 1), bias=False)\n", " (norm2): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " (relu2): ReLU(inplace=True)\n", " (conv2): Conv2d(192, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)\n", " )\n", " )\n", " (norm5): BatchNorm2d(2208, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n", " )\n", " (1): ReLU()\n", " (2): AdaptiveAvgPool2d(output_size=(1, 1))\n", " (3): Flatten(start_dim=1, end_dim=-1)\n", " (4): Linear(in_features=2208, out_features=3, bias=True)\n", ")\n" ] } ], "source": [ "# Show the last layers\n", "print(net)" ] }, { "cell_type": "code", "execution_count": 132, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "" ] }, "execution_count": 132, "metadata": {}, "output_type": "execute_result" } ], "source": [ "import torch\n", "import torch.nn as nn\n", "import torch.optim as optim\n", "from torch.optim import lr_scheduler\n", "import torch.backends.cudnn as cudnn\n", "import numpy as np\n", "import torchvision\n", "from torchvision import datasets, models, transforms\n", "import matplotlib.pyplot as plt\n", "import time\n", "import os\n", "from PIL import Image\n", "from tempfile import TemporaryDirectory\n", "\n", "cudnn.benchmark = True\n", "plt.ion() # interactive mode" ] }, { "cell_type": "code", "execution_count": 133, "metadata": {}, "outputs": [], "source": [ "learning_rate = 10**(-5)\n", "momentum = 0.8\n", "batch_size = 4\n", "epochs = 80\n", "augmentation = ['HFlip', 'VFlip', 'BtnsCtst']\n", "frozen = False" ] }, { "cell_type": "code", "execution_count": 134, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "{'train': Dataset ImageFolder\n", " Number of datapoints: 327\n", " Root location: /home/wfd/Desktop/Projet_M1/FineTuning/Data/Classification/S/train\n", " StandardTransform\n", "Transform: Compose(\n", " ToTensor()\n", " RandomHorizontalFlip(p=0.25)\n", " RandomVerticalFlip(p=0.25)\n", " RandomApply(\n", " p=0.25\n", " ColorJitter(brightness=[0.7, 1.3], contrast=[0.7, 1.3], saturation=None, hue=None)\n", " )\n", " ), 'val': Dataset ImageFolder\n", " Number of datapoints: 138\n", " Root location: /home/wfd/Desktop/Projet_M1/FineTuning/Data/Classification/S/val\n", " StandardTransform\n", "Transform: Compose(\n", " ToTensor()\n", " RandomHorizontalFlip(p=0.25)\n", " RandomVerticalFlip(p=0.25)\n", " RandomApply(\n", " p=0.25\n", " ColorJitter(brightness=[0.7, 1.3], contrast=[0.7, 1.3], saturation=None, hue=None)\n", " )\n", " )}\n", "['GGS', 'NoGS', 'SGS']\n", "{'GGS': 0, 'NoGS': 1, 'SGS': 2}\n" ] } ], "source": [ "data_transforms = {\n", " 'train': transforms.Compose([\n", " transforms.ToTensor(),\n", " transforms.RandomHorizontalFlip(p=0.25),\n", " transforms.RandomVerticalFlip(p=0.25),\n", " transforms.RandomApply([transforms.ColorJitter(brightness=0.3, contrast=0.3)], p=0.25),\n", " ]),\n", " 'val': transforms.Compose([\n", " transforms.ToTensor(),\n", " transforms.RandomHorizontalFlip(p=0.25),\n", " transforms.RandomVerticalFlip(p=0.25),\n", " transforms.RandomApply([transforms.ColorJitter(brightness=0.3, contrast=0.3)], p=0.25),\n", " ]),\n", "}\n", "\n", "data_dir = \"/home/wfd/Desktop/Projet_M1/FineTuning/Data/Classification/\" + lesion \n", "image_datasets = {x: datasets.ImageFolder(os.path.join(data_dir, x), data_transforms[x]) for x in ['train', 'val']}\n", "dataloaders = {x: torch.utils.data.DataLoader(image_datasets[x], batch_size=batch_size, shuffle=True) for x in ['train', 'val']}\n", "dataset_sizes = {x: len(image_datasets[x]) for x in ['train', 'val']}\n", "\n", "# Print the data, and each class, as well as the number associated with each class\n", "print(image_datasets)\n", "print(image_datasets['train'].classes)\n", "print(image_datasets['train'].class_to_idx)\n", "\n", "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")" ] }, { "cell_type": "code", "execution_count": 135, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Accuracy of the network on the 327 validation images: 89.29663608562691%\n" ] } ], "source": [ "# Test on all the data to get the accuracy before fine-tuning\n", "correct = 0\n", "total = 0\n", "with torch.no_grad():\n", " for data in dataloaders['train']:\n", " images, labels = data\n", " images, labels = images.to(device), labels.to(device)\n", " outputs = net(images)\n", " _, predicted = torch.max(outputs, 1)\n", " # print(labels, outputs, predicted)\n", " total += labels.size(0)\n", " correct += (predicted == labels).sum().item()\n", "print('Accuracy of the network on the {} validation images: {}%'.format(total, 100 * correct / total))" ] }, { "cell_type": "code", "execution_count": 136, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Accuracy of the network on the 138 validation images: 92.7536231884058%\n" ] } ], "source": [ "# Test on all the data to get the accuracy before fine-tuning\n", "correct = 0\n", "total = 0\n", "with torch.no_grad():\n", " for data in dataloaders['val']:\n", " images, labels = data\n", " images, labels = images.to(device), labels.to(device)\n", " outputs = net(images)\n", " _, predicted = torch.max(outputs, 1)\n", " # print(labels, outputs, predicted)\n", " total += labels.size(0)\n", " correct += (predicted == labels).sum().item()\n", "print('Accuracy of the network on the {} validation images: {}%'.format(total, 100 * correct / total))" ] }, { "cell_type": "code", "execution_count": 137, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# 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", "def imshow(inp, title=None):\n", " \"\"\"Imshow for Tensor.\"\"\"\n", " inp = inp.numpy().transpose((1, 2, 0))\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", " \n", "imshow(out, title=[image_datasets['train'].classes[x] for x in classes])" ] }, { "cell_type": "code", "execution_count": 138, "metadata": {}, "outputs": [], "source": [ "if frozen and lesion != \"S\":\n", " # Freeze the net layers except the final layer\n", " for param in net.parameters():\n", " param.requires_grad = False\n", "\n", " # Unfreeze the final layer\n", " for param in net.classifier.parameters():\n", " param.requires_grad = True\n", "\n", "net = net.to(device)\n", "\n", "criterion = nn.CrossEntropyLoss()\n", "optimizer = optim.SGD(net.parameters(), lr=learning_rate, momentum=momentum)" ] }, { "cell_type": "code", "execution_count": 139, "metadata": {}, "outputs": [], "source": [ "# for param in net.parameters():\n", "# print(param.requires_grad)" ] }, { "cell_type": "code", "execution_count": 140, "metadata": {}, "outputs": [ { "data": { "text/html": [ "Tracking run with wandb version 0.16.6" ], "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "Run data is saved locally in /home/wfd/Desktop/Projet_M1/wandb/run-20240416_145117-wcpiodox" ], "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "Syncing run jumping-moon-35 to Weights & Biases (docs)
" ], "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ " View project at https://wandb.ai/wlucet/projet_m1" ], "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ " View run at https://wandb.ai/wlucet/projet_m1/runs/wcpiodox" ], "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "" ], "text/plain": [ "" ] }, "execution_count": 140, "metadata": {}, "output_type": "execute_result" } ], "source": [ "import wandb\n", "\n", "# start a new wandb run to track this script\n", "wandb.init(\n", " # set the wandb project where this run will be logged\n", " project=\"projet_m1\",\n", "\n", " # track hyperparameters and run metadata\n", " config={\n", " \"learning_rate\": learning_rate,\n", " \"architecture\": model_name,\n", " \"dataset\": lesion,\n", " \"epochs\": epochs,\n", " \"batch_size\": batch_size,\n", " \"momentum\": momentum,\n", " \"augmentation\": augmentation,\n", " \"frozen\": frozen,\n", " }\n", ")" ] }, { "cell_type": "code", "execution_count": 141, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\n", "Epoch 1\n", "----------\n", "train Loss: 0.8158 Acc: 0.6697\n", "val Loss: 0.4256 Acc: 0.8406\n", "\n", "Epoch 2\n", "----------\n", "train Loss: 0.5582 Acc: 0.7829\n", "val Loss: 0.3179 Acc: 0.8768\n", "\n", "Epoch 3\n", "----------\n", "train Loss: 0.4802 Acc: 0.8165\n", "val Loss: 0.2736 Acc: 0.9058\n", "\n", "Epoch 4\n", "----------\n", "train Loss: 0.4496 Acc: 0.8287\n", "val Loss: 0.2296 Acc: 0.9420\n", "\n", "Epoch 5\n", "----------\n", "train Loss: 0.3829 Acc: 0.8471\n", "val Loss: 0.1755 Acc: 0.9420\n", "\n", "Epoch 6\n", "----------\n", "train Loss: 0.3816 Acc: 0.8624\n", "val Loss: 0.1508 Acc: 0.9493\n", "\n", "Epoch 7\n", "----------\n", "train Loss: 0.3356 Acc: 0.8654\n", "val Loss: 0.1866 Acc: 0.9493\n", "\n", "Epoch 8\n", "----------\n", "train Loss: 0.3601 Acc: 0.8593\n", "val Loss: 0.1615 Acc: 0.9493\n", "\n", "Epoch 9\n", "----------\n", "train Loss: 0.3379 Acc: 0.8838\n", "val Loss: 0.1554 Acc: 0.9275\n", "\n", "Epoch 10\n", "----------\n", "train Loss: 0.3691 Acc: 0.8807\n", "val Loss: 0.1846 Acc: 0.9420\n", "\n", "Epoch 11\n", "----------\n", "train Loss: 0.3038 Acc: 0.8930\n", "val Loss: 0.1660 Acc: 0.9493\n", "\n", "Epoch 12\n", "----------\n", "train Loss: 0.3558 Acc: 0.8716\n", "val Loss: 0.1774 Acc: 0.9348\n", "\n", "Epoch 13\n", "----------\n", "train Loss: 0.3224 Acc: 0.8746\n", "val Loss: 0.1682 Acc: 0.9420\n", "\n", "Epoch 14\n", "----------\n", "train Loss: 0.3202 Acc: 0.8869\n", "val Loss: 0.1844 Acc: 0.9348\n", "\n", "Epoch 15\n", "----------\n", "train Loss: 0.3543 Acc: 0.8869\n", "val Loss: 0.1633 Acc: 0.9348\n", "\n", "Epoch 16\n", "----------\n", "train Loss: 0.3385 Acc: 0.8777\n", "val Loss: 0.1677 Acc: 0.9348\n", "\n", "Epoch 17\n", "----------\n", "train Loss: 0.3163 Acc: 0.8807\n", "val Loss: 0.1536 Acc: 0.9565\n", "\n", "Epoch 18\n", "----------\n", "train Loss: 0.3256 Acc: 0.8960\n", "val Loss: 0.1616 Acc: 0.9638\n", "\n", "Epoch 19\n", "----------\n", "train Loss: 0.3030 Acc: 0.9052\n", "val Loss: 0.1586 Acc: 0.9420\n", "\n", "Epoch 20\n", "----------\n", "train Loss: 0.3020 Acc: 0.8899\n", "val Loss: 0.1522 Acc: 0.9348\n", "\n", "Epoch 21\n", "----------\n", "train Loss: 0.3114 Acc: 0.8777\n", "val Loss: 0.1952 Acc: 0.9348\n", "\n", "Epoch 22\n", "----------\n", "train Loss: 0.3061 Acc: 0.8991\n", "val Loss: 0.1970 Acc: 0.9275\n", "\n", "Epoch 23\n", "----------\n", "train Loss: 0.3240 Acc: 0.8807\n", "val Loss: 0.1716 Acc: 0.9565\n", "\n", "Epoch 24\n", "----------\n", "train Loss: 0.2539 Acc: 0.8991\n", "val Loss: 0.1346 Acc: 0.9565\n", "\n", "Epoch 25\n", "----------\n", "train Loss: 0.2918 Acc: 0.8869\n", "val Loss: 0.1492 Acc: 0.9493\n", "\n", "Epoch 26\n", "----------\n", "train Loss: 0.3409 Acc: 0.8746\n", "val Loss: 0.1675 Acc: 0.9275\n", "\n", "Epoch 27\n", "----------\n", "train Loss: 0.3085 Acc: 0.8899\n", "val Loss: 0.1793 Acc: 0.9058\n", "\n", "Epoch 28\n", "----------\n", "train Loss: 0.3007 Acc: 0.8777\n", "val Loss: 0.1605 Acc: 0.9348\n", "\n", "Epoch 29\n", "----------\n", "train Loss: 0.3341 Acc: 0.8838\n", "val Loss: 0.1468 Acc: 0.9638\n", "\n", "Epoch 30\n", "----------\n", "train Loss: 0.2675 Acc: 0.8869\n", "val Loss: 0.1803 Acc: 0.9275\n", "\n", "Epoch 31\n", "----------\n", "train Loss: 0.3322 Acc: 0.8746\n", "val Loss: 0.1532 Acc: 0.9348\n", "\n", "Epoch 32\n", "----------\n", "train Loss: 0.2753 Acc: 0.8899\n", "val Loss: 0.1527 Acc: 0.9420\n", "\n", "Epoch 33\n", "----------\n", "train Loss: 0.3657 Acc: 0.8593\n", "val Loss: 0.1663 Acc: 0.9420\n", "\n", "Epoch 34\n", "----------\n", "train Loss: 0.2498 Acc: 0.9052\n", "val Loss: 0.1540 Acc: 0.9493\n", "\n", "Epoch 35\n", "----------\n", "train Loss: 0.2769 Acc: 0.9235\n", "val Loss: 0.1479 Acc: 0.9493\n", "\n", "Epoch 36\n", "----------\n", "train Loss: 0.2954 Acc: 0.8930\n", "val Loss: 0.1723 Acc: 0.9348\n", "\n", "Epoch 37\n", "----------\n", "train Loss: 0.2928 Acc: 0.8869\n", "val Loss: 0.1598 Acc: 0.9565\n", "\n", "Epoch 38\n", "----------\n", "train Loss: 0.2861 Acc: 0.9021\n", "val Loss: 0.1474 Acc: 0.9638\n", "\n", "Epoch 39\n", "----------\n", "train Loss: 0.2632 Acc: 0.9052\n", "val Loss: 0.1868 Acc: 0.9348\n", "\n", "Epoch 40\n", "----------\n", "train Loss: 0.2804 Acc: 0.9021\n", "val Loss: 0.1507 Acc: 0.9493\n", "\n", "Epoch 41\n", "----------\n", "train Loss: 0.3137 Acc: 0.8899\n", "val Loss: 0.1382 Acc: 0.9420\n", "\n", "Epoch 42\n", "----------\n", "train Loss: 0.3013 Acc: 0.8869\n", "val Loss: 0.1582 Acc: 0.9348\n", "\n", "Epoch 43\n", "----------\n", "train Loss: 0.2627 Acc: 0.8930\n", "val Loss: 0.1740 Acc: 0.9420\n", "\n", "Epoch 44\n", "----------\n", "train Loss: 0.2745 Acc: 0.9052\n", "val Loss: 0.1435 Acc: 0.9493\n", "\n", "Epoch 45\n", "----------\n", "train Loss: 0.2612 Acc: 0.9113\n", "val Loss: 0.1604 Acc: 0.9493\n", "\n", "Epoch 46\n", "----------\n", "train Loss: 0.2486 Acc: 0.9174\n", "val Loss: 0.1733 Acc: 0.9275\n", "\n", "Epoch 47\n", "----------\n", "train Loss: 0.2790 Acc: 0.8960\n", "val Loss: 0.1615 Acc: 0.9493\n", "\n", "Epoch 48\n", "----------\n", "train Loss: 0.2825 Acc: 0.8899\n", "val Loss: 0.1579 Acc: 0.9420\n", "\n", "Epoch 49\n", "----------\n", "train Loss: 0.3128 Acc: 0.8960\n", "val Loss: 0.1336 Acc: 0.9638\n", "\n", "Epoch 50\n", "----------\n", "train Loss: 0.3000 Acc: 0.8777\n", "val Loss: 0.1702 Acc: 0.9493\n", "\n", "Epoch 51\n", "----------\n", "train Loss: 0.3090 Acc: 0.8838\n", "val Loss: 0.1593 Acc: 0.9348\n", "\n", "Epoch 52\n", "----------\n", "train Loss: 0.2512 Acc: 0.9083\n", "val Loss: 0.1521 Acc: 0.9420\n", "\n", "Epoch 53\n", "----------\n", "train Loss: 0.2611 Acc: 0.9174\n", "val Loss: 0.1711 Acc: 0.9493\n", "\n", "Epoch 54\n", "----------\n", "train Loss: 0.2845 Acc: 0.9083\n", "val Loss: 0.1647 Acc: 0.9420\n", "\n", "Epoch 55\n", "----------\n", "train Loss: 0.2652 Acc: 0.9144\n", "val Loss: 0.1746 Acc: 0.9348\n", "\n", "Epoch 56\n", "----------\n", "train Loss: 0.2507 Acc: 0.9083\n", "val Loss: 0.1584 Acc: 0.9348\n", "\n", "Epoch 57\n", "----------\n", "train Loss: 0.3042 Acc: 0.8899\n", "val Loss: 0.1664 Acc: 0.9275\n", "\n", "Epoch 58\n", "----------\n", "train Loss: 0.2511 Acc: 0.8991\n", "val Loss: 0.1560 Acc: 0.9348\n", "\n", "Epoch 59\n", "----------\n", "train Loss: 0.2728 Acc: 0.9052\n", "val Loss: 0.1519 Acc: 0.9493\n", "\n", "Epoch 60\n", "----------\n", "train Loss: 0.2925 Acc: 0.8838\n", "val Loss: 0.1765 Acc: 0.9203\n", "\n", "Epoch 61\n", "----------\n", "train Loss: 0.2563 Acc: 0.9052\n", "val Loss: 0.1704 Acc: 0.9348\n", "\n", "Epoch 62\n", "----------\n", "train Loss: 0.2903 Acc: 0.9021\n", "val Loss: 0.1430 Acc: 0.9420\n", "\n", "Epoch 63\n", "----------\n", "train Loss: 0.2632 Acc: 0.9083\n", "val Loss: 0.1427 Acc: 0.9565\n", "\n", "Epoch 64\n", "----------\n", "train Loss: 0.2891 Acc: 0.8960\n", "val Loss: 0.1385 Acc: 0.9565\n", "\n", "Epoch 65\n", "----------\n", "train Loss: 0.2404 Acc: 0.9083\n", "val Loss: 0.1348 Acc: 0.9565\n", "\n", "Epoch 66\n", "----------\n", "train Loss: 0.2496 Acc: 0.9021\n", "val Loss: 0.1845 Acc: 0.9420\n", "\n", "Epoch 67\n", "----------\n", "train Loss: 0.2659 Acc: 0.8960\n", "val Loss: 0.1618 Acc: 0.9565\n", "\n", "Epoch 68\n", "----------\n", "train Loss: 0.2617 Acc: 0.9174\n", "val Loss: 0.1512 Acc: 0.9420\n", "\n", "Epoch 69\n", "----------\n", "train Loss: 0.2599 Acc: 0.9205\n", "val Loss: 0.1617 Acc: 0.9275\n", "\n", "Epoch 70\n", "----------\n", "train Loss: 0.2744 Acc: 0.8899\n", "val Loss: 0.1598 Acc: 0.9493\n", "\n", "Epoch 71\n", "----------\n", "train Loss: 0.2638 Acc: 0.9021\n", "val Loss: 0.1440 Acc: 0.9565\n", "\n", "Epoch 72\n", "----------\n", "train Loss: 0.2540 Acc: 0.9021\n", "val Loss: 0.1318 Acc: 0.9638\n", "\n", "Epoch 73\n", "----------\n", "train Loss: 0.2253 Acc: 0.9235\n", "val Loss: 0.1403 Acc: 0.9638\n", "\n", "Epoch 74\n", "----------\n", "train Loss: 0.2287 Acc: 0.8991\n", "val Loss: 0.1477 Acc: 0.9493\n", "\n", "Epoch 75\n", "----------\n", "train Loss: 0.2410 Acc: 0.9174\n", "val Loss: 0.1617 Acc: 0.9420\n", "\n", "Epoch 76\n", "----------\n", "train Loss: 0.2680 Acc: 0.8960\n", "val Loss: 0.1510 Acc: 0.9638\n", "\n", "Epoch 77\n", "----------\n", "train Loss: 0.3136 Acc: 0.8930\n", "val Loss: 0.1411 Acc: 0.9420\n", "\n", "Epoch 78\n", "----------\n", "train Loss: 0.3218 Acc: 0.8807\n", "val Loss: 0.1681 Acc: 0.9275\n", "\n", "Epoch 79\n", "----------\n", "train Loss: 0.2416 Acc: 0.9205\n", "val Loss: 0.1644 Acc: 0.9348\n", "\n", "Epoch 80\n", "----------\n", "train Loss: 0.2248 Acc: 0.9144\n", "val Loss: 0.1396 Acc: 0.9565\n", "Training complete in 10m 25s\n", "Best val Acc: 0.9638\n" ] } ], "source": [ "import torch\n", "from torch.utils.data import DataLoader\n", "import time\n", "from tempfile import TemporaryDirectory\n", "import os\n", "\n", "def train_model(model, criterion, optimizer, dataloaders, dataset_sizes, device, num_epochs=25):\n", " since = time.time()\n", "\n", " # Make sure the device is set correctly\n", " model.to(device)\n", "\n", " # Create a temporary directory to save training checkpoints\n", " with TemporaryDirectory() as tempdir:\n", " best_model_params_path = os.path.join(tempdir, 'best_model_params.pt')\n", "\n", " # Initially save the current state of the model\n", " torch.save(model.state_dict(), best_model_params_path)\n", " best_acc = 0.0\n", "\n", " for epoch in range(num_epochs):\n", " print(f'\\nEpoch {epoch+1}')\n", " print('-' * 10)\n", "\n", " # Each epoch has a training and validation phase\n", " for phase in ['train', 'val']:\n", " if phase == 'train':\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 train\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", " # # Print outputs and labels to debug the model's predictions\n", " # if epoch == 0 and phase == 'train':\n", " # print(f\"First batch labels and predictions in training: {labels} {preds}\")\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", " if (phase == 'train'):\n", " wandb.log({\"train_acc\": epoch_acc, \"train_loss\": epoch_loss, \"epoch\": epoch})\n", " else:\n", " wandb.log({\"val_acc\": epoch_acc, \"val_loss\": epoch_loss, \"epoch\": epoch})\n", "\n", " print(f'{phase} Loss: {epoch_loss:.4f} Acc: {epoch_acc:.4f}')\n", "\n", " # deep copy the model\n", " if phase == 'val' and epoch_acc > best_acc:\n", " best_acc = epoch_acc\n", " wandb.log({\"best_val_acc\": best_acc})\n", " torch.save(model.state_dict(), best_model_params_path)\n", "\n", " time_elapsed = time.time() - since\n", " print(f'Training complete in {time_elapsed // 60:.0f}m {time_elapsed % 60:.0f}s')\n", " print(f'Best val Acc: {best_acc:.4f}')\n", "\n", " # load best model weights\n", " model.load_state_dict(torch.load(best_model_params_path))\n", " return best_acc, model\n", "\n", "# Example usage\n", "b_acc, net = train_model(net, criterion, optimizer, dataloaders, dataset_sizes, device, epochs)\n" ] }, { "cell_type": "code", "execution_count": 142, "metadata": {}, "outputs": [ { "data": { "text/html": [ "\n", "

Run history:


best_val_acc▁▃▅▇▇██
epoch▁▁▁▂▂▂▂▂▂▃▃▃▃▃▃▄▄▄▄▄▅▅▅▅▅▅▆▆▆▆▆▆▇▇▇▇▇███
train_acc▁▅▆▆▇▇▇▇▇▇▇▇▇▇▇▇▆█▇▇▇▇█▇▇▇██▇▇▇██▇█▇██▇█
train_loss█▄▃▂▂▂▂▃▂▂▂▂▂▂▂▂▃▂▂▁▂▁▁▂▂▂▁▁▂▂▁▁▁▁▁▁▁▁▂▁
val_acc▁▅▇▇▆▇▇▆█▇▆█▇▅█▆▇▇█▆▇▇▇▇█▆▇▆▆▇▆███▆██▇▇█
val_loss█▄▂▂▂▂▂▂▁▂▂▂▁▂▁▁▂▁▂▂▁▂▂▂▁▂▂▂▂▁▂▁▁▂▂▁▁▂▁▁

Run summary:


best_val_acc0.96377
epoch79
train_acc0.91437
train_loss0.22484
val_acc0.95652
val_loss0.13965

" ], "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ " View run jumping-moon-35 at: https://wandb.ai/wlucet/projet_m1/runs/wcpiodox
View project at: https://wandb.ai/wlucet/projet_m1
Synced 6 W&B file(s), 0 media file(s), 0 artifact file(s) and 0 other file(s)" ], "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/html": [ "Find logs at: ./wandb/run-20240416_145117-wcpiodox/logs" ], "text/plain": [ "" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# End wandb run\n", "wandb.finish()\n", "\n", "b_acc = round(b_acc.item(), 4)\n", "\n", "# Save the model\n", "torch.save(net, os.path.join(local_models_dir, f\"{model_name.split('.')[0]}_{b_acc}_fine_tuned.pth\"))" ] }, { "cell_type": "code", "execution_count": 143, "metadata": {}, "outputs": [], "source": [ "# # Test on all the data to get the accuracy before fine-tuning\n", "# correct = 0\n", "# total = 0\n", "# with torch.no_grad():\n", "# for data in dataloaders['train']:\n", "# images, labels = data\n", "# images, labels = images.to(device), labels.to(device)\n", "# outputs = net(images)\n", "# _, predicted = torch.max(outputs, 1)\n", "# total += labels.size(0)\n", "# correct += (predicted == labels).sum().item()\n", "# print('Accuracy of the network on the {} validation images: {}%'.format(total, 100 * correct / total))" ] }, { "cell_type": "code", "execution_count": 144, "metadata": {}, "outputs": [], "source": [ "# # Test on all the data to get the accuracy before fine-tuning\n", "# correct = 0\n", "# total = 0\n", "# with torch.no_grad():\n", "# for data in dataloaders['val']:\n", "# images, labels = data\n", "# images, labels = images.to(device), labels.to(device)\n", "# outputs = net(images)\n", "# _, predicted = torch.max(outputs, 1)\n", "# total += labels.size(0)\n", "# correct += (predicted == labels).sum().item()\n", "# print('Accuracy of the network on the {} validation images: {}%'.format(total, 100 * correct / total))" ] } ], "metadata": { "kernelspec": { "display_name": "segmentation", "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": "3.10.14" } }, "nbformat": 4, "nbformat_minor": 2 }