{
"cells": [
{
"cell_type": "markdown",
"id": "f4cf78e1",
"metadata": {
"slideshow": {
"slide_type": "-"
}
},
"source": [
"# Densely Connected Networks (DenseNet)\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "898d7836",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T19:47:25.812288Z",
"iopub.status.busy": "2023-08-18T19:47:25.811221Z",
"iopub.status.idle": "2023-08-18T19:47:28.844635Z",
"shell.execute_reply": "2023-08-18T19:47:28.843703Z"
},
"origin_pos": 3,
"tab": [
"pytorch"
]
},
"outputs": [],
"source": [
"import torch\n",
"from torch import nn\n",
"from d2l import torch as d2l"
]
},
{
"cell_type": "markdown",
"id": "dcb6db8a",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"Dense Blocks"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "805394e3",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T19:47:28.856547Z",
"iopub.status.busy": "2023-08-18T19:47:28.855981Z",
"iopub.status.idle": "2023-08-18T19:47:28.862956Z",
"shell.execute_reply": "2023-08-18T19:47:28.861783Z"
},
"origin_pos": 13,
"tab": [
"pytorch"
]
},
"outputs": [],
"source": [
"def conv_block(num_channels):\n",
" return nn.Sequential(\n",
" nn.LazyBatchNorm2d(), nn.ReLU(),\n",
" nn.LazyConv2d(num_channels, kernel_size=3, padding=1))\n",
"\n",
"class DenseBlock(nn.Module):\n",
" def __init__(self, num_convs, num_channels):\n",
" super(DenseBlock, self).__init__()\n",
" layer = []\n",
" for i in range(num_convs):\n",
" layer.append(conv_block(num_channels))\n",
" self.net = nn.Sequential(*layer)\n",
"\n",
" def forward(self, X):\n",
" for blk in self.net:\n",
" Y = blk(X)\n",
" X = torch.cat((X, Y), dim=1)\n",
" return X"
]
},
{
"cell_type": "markdown",
"id": "ff6dae74",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"Define a `DenseBlock` instance"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "e369c1ad",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T19:47:28.867108Z",
"iopub.status.busy": "2023-08-18T19:47:28.866407Z",
"iopub.status.idle": "2023-08-18T19:47:28.909936Z",
"shell.execute_reply": "2023-08-18T19:47:28.908954Z"
},
"origin_pos": 17,
"tab": [
"pytorch"
]
},
"outputs": [
{
"data": {
"text/plain": [
"torch.Size([4, 23, 8, 8])"
]
},
"execution_count": 4,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"blk = DenseBlock(2, 10)\n",
"X = torch.randn(4, 3, 8, 8)\n",
"Y = blk(X)\n",
"Y.shape"
]
},
{
"cell_type": "markdown",
"id": "dc32fbb1",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"Transition Layers"
]
},
{
"cell_type": "code",
"execution_count": 5,
"id": "6160cc48",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T19:47:28.915937Z",
"iopub.status.busy": "2023-08-18T19:47:28.914796Z",
"iopub.status.idle": "2023-08-18T19:47:28.920281Z",
"shell.execute_reply": "2023-08-18T19:47:28.919184Z"
},
"origin_pos": 21,
"tab": [
"pytorch"
]
},
"outputs": [],
"source": [
"def transition_block(num_channels):\n",
" return nn.Sequential(\n",
" nn.LazyBatchNorm2d(), nn.ReLU(),\n",
" nn.LazyConv2d(num_channels, kernel_size=1),\n",
" nn.AvgPool2d(kernel_size=2, stride=2))"
]
},
{
"cell_type": "markdown",
"id": "9f88387c",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"Apply a transition layer"
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "fc0cacfc",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T19:47:28.924597Z",
"iopub.status.busy": "2023-08-18T19:47:28.924323Z",
"iopub.status.idle": "2023-08-18T19:47:28.938373Z",
"shell.execute_reply": "2023-08-18T19:47:28.937285Z"
},
"origin_pos": 26,
"tab": [
"pytorch"
]
},
"outputs": [
{
"data": {
"text/plain": [
"torch.Size([4, 10, 4, 4])"
]
},
"execution_count": 6,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"blk = transition_block(10)\n",
"blk(Y).shape"
]
},
{
"cell_type": "markdown",
"id": "e0ea910e",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"DenseNet Model"
]
},
{
"cell_type": "code",
"execution_count": 8,
"id": "58137883",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T19:47:28.950027Z",
"iopub.status.busy": "2023-08-18T19:47:28.949746Z",
"iopub.status.idle": "2023-08-18T19:47:28.958660Z",
"shell.execute_reply": "2023-08-18T19:47:28.957444Z"
},
"origin_pos": 33,
"tab": [
"pytorch"
]
},
"outputs": [],
"source": [
"class DenseNet(d2l.Classifier):\n",
" def b1(self):\n",
" return nn.Sequential(\n",
" nn.LazyConv2d(64, kernel_size=7, stride=2, padding=3),\n",
" nn.LazyBatchNorm2d(), nn.ReLU(),\n",
" nn.MaxPool2d(kernel_size=3, stride=2, padding=1))\n",
"\n",
"@d2l.add_to_class(DenseNet)\n",
"def __init__(self, num_channels=64, growth_rate=32, arch=(4, 4, 4, 4),\n",
" lr=0.1, num_classes=10):\n",
" super(DenseNet, self).__init__()\n",
" self.save_hyperparameters()\n",
" self.net = nn.Sequential(self.b1())\n",
" for i, num_convs in enumerate(arch):\n",
" self.net.add_module(f'dense_blk{i+1}', DenseBlock(num_convs,\n",
" growth_rate))\n",
" num_channels += num_convs * growth_rate\n",
" if i != len(arch) - 1:\n",
" num_channels //= 2\n",
" self.net.add_module(f'tran_blk{i+1}', transition_block(\n",
" num_channels))\n",
" self.net.add_module('last', nn.Sequential(\n",
" nn.LazyBatchNorm2d(), nn.ReLU(),\n",
" nn.AdaptiveAvgPool2d((1, 1)), nn.Flatten(),\n",
" nn.LazyLinear(num_classes)))\n",
" self.net.apply(d2l.init_cnn)"
]
},
{
"cell_type": "markdown",
"id": "ab4d03b5",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"Training"
]
},
{
"cell_type": "code",
"execution_count": 9,
"id": "ef87c44e",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T19:47:28.963624Z",
"iopub.status.busy": "2023-08-18T19:47:28.962964Z",
"iopub.status.idle": "2023-08-18T19:50:01.060105Z",
"shell.execute_reply": "2023-08-18T19:50:01.059052Z"
},
"origin_pos": 36,
"tab": [
"pytorch"
]
},
"outputs": [
{
"data": {
"image/svg+xml": [
"\n",
"\n",
"\n"
],
"text/plain": [
""
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"model = DenseNet(lr=0.01)\n",
"trainer = d2l.Trainer(max_epochs=10, num_gpus=1)\n",
"data = d2l.FashionMNIST(batch_size=128, resize=(96, 96))\n",
"trainer.fit(model, data)"
]
}
],
"metadata": {
"celltoolbar": "Slideshow",
"language_info": {
"name": "python"
},
"required_libs": [],
"rise": {
"autolaunch": true,
"enable_chalkboard": true,
"overlay": "",
"scroll": true
}
},
"nbformat": 4,
"nbformat_minor": 5
}