{
"cells": [
{
"cell_type": "markdown",
"id": "741aaf07",
"metadata": {
"slideshow": {
"slide_type": "-"
}
},
"source": [
"# Multi-Branch Networks (GoogLeNet)\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "066f8e89",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T20:12:43.017176Z",
"iopub.status.busy": "2023-08-18T20:12:43.016624Z",
"iopub.status.idle": "2023-08-18T20:12:46.230431Z",
"shell.execute_reply": "2023-08-18T20:12:46.228850Z"
},
"origin_pos": 3,
"tab": [
"pytorch"
]
},
"outputs": [],
"source": [
"import torch\n",
"from torch import nn\n",
"from torch.nn import functional as F\n",
"from d2l import torch as d2l"
]
},
{
"cell_type": "markdown",
"id": "e6c8822d",
"metadata": {
"slideshow": {
"slide_type": "-"
}
},
"source": [
"Inception Blocks"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "7deda9e0",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T20:12:46.235999Z",
"iopub.status.busy": "2023-08-18T20:12:46.235477Z",
"iopub.status.idle": "2023-08-18T20:12:46.245899Z",
"shell.execute_reply": "2023-08-18T20:12:46.244271Z"
},
"origin_pos": 8,
"tab": [
"pytorch"
]
},
"outputs": [],
"source": [
"class Inception(nn.Module):\n",
" def __init__(self, c1, c2, c3, c4, **kwargs):\n",
" super(Inception, self).__init__(**kwargs)\n",
" self.b1_1 = nn.LazyConv2d(c1, kernel_size=1)\n",
" self.b2_1 = nn.LazyConv2d(c2[0], kernel_size=1)\n",
" self.b2_2 = nn.LazyConv2d(c2[1], kernel_size=3, padding=1)\n",
" self.b3_1 = nn.LazyConv2d(c3[0], kernel_size=1)\n",
" self.b3_2 = nn.LazyConv2d(c3[1], kernel_size=5, padding=2)\n",
" self.b4_1 = nn.MaxPool2d(kernel_size=3, stride=1, padding=1)\n",
" self.b4_2 = nn.LazyConv2d(c4, kernel_size=1)\n",
"\n",
" def forward(self, x):\n",
" b1 = F.relu(self.b1_1(x))\n",
" b2 = F.relu(self.b2_2(F.relu(self.b2_1(x))))\n",
" b3 = F.relu(self.b3_2(F.relu(self.b3_1(x))))\n",
" b4 = F.relu(self.b4_2(self.b4_1(x)))\n",
" return torch.cat((b1, b2, b3, b4), dim=1)"
]
},
{
"cell_type": "markdown",
"id": "5f988986",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"GoogLeNet Model"
]
},
{
"cell_type": "code",
"execution_count": 8,
"id": "fe47c47d",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T20:12:46.306521Z",
"iopub.status.busy": "2023-08-18T20:12:46.305540Z",
"iopub.status.idle": "2023-08-18T20:12:46.313597Z",
"shell.execute_reply": "2023-08-18T20:12:46.312191Z"
},
"origin_pos": 23,
"tab": [
"pytorch"
]
},
"outputs": [],
"source": [
"class GoogleNet(d2l.Classifier):\n",
" def b1(self):\n",
" return nn.Sequential(\n",
" nn.LazyConv2d(64, kernel_size=7, stride=2, padding=3),\n",
" nn.ReLU(), nn.MaxPool2d(kernel_size=3, stride=2, padding=1))\n",
"\n",
"@d2l.add_to_class(GoogleNet)\n",
"def b2(self):\n",
" return nn.Sequential(\n",
" nn.LazyConv2d(64, kernel_size=1), nn.ReLU(),\n",
" nn.LazyConv2d(192, kernel_size=3, padding=1), nn.ReLU(),\n",
" nn.MaxPool2d(kernel_size=3, stride=2, padding=1))\n",
"\n",
"@d2l.add_to_class(GoogleNet)\n",
"def b3(self):\n",
" return nn.Sequential(Inception(64, (96, 128), (16, 32), 32),\n",
" Inception(128, (128, 192), (32, 96), 64),\n",
" nn.MaxPool2d(kernel_size=3, stride=2, padding=1))\n",
"\n",
"@d2l.add_to_class(GoogleNet)\n",
"def b4(self):\n",
" return nn.Sequential(Inception(192, (96, 208), (16, 48), 64),\n",
" Inception(160, (112, 224), (24, 64), 64),\n",
" Inception(128, (128, 256), (24, 64), 64),\n",
" Inception(112, (144, 288), (32, 64), 64),\n",
" Inception(256, (160, 320), (32, 128), 128),\n",
" nn.MaxPool2d(kernel_size=3, stride=2, padding=1))\n",
"\n",
"@d2l.add_to_class(GoogleNet)\n",
"def b5(self):\n",
" return nn.Sequential(Inception(256, (160, 320), (32, 128), 128),\n",
" Inception(384, (192, 384), (48, 128), 128),\n",
" nn.AdaptiveAvgPool2d((1,1)), nn.Flatten())\n",
"\n",
"@d2l.add_to_class(GoogleNet)\n",
"def __init__(self, lr=0.1, num_classes=10):\n",
" super(GoogleNet, self).__init__()\n",
" self.save_hyperparameters()\n",
" self.net = nn.Sequential(self.b1(), self.b2(), self.b3(), self.b4(),\n",
" self.b5(), nn.LazyLinear(num_classes))\n",
" self.net.apply(d2l.init_cnn)"
]
},
{
"cell_type": "markdown",
"id": "a6886825",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"Reduce the input height and width from 224 to 96\n",
"to have a reasonable training time on Fashion-MNIST"
]
},
{
"cell_type": "code",
"execution_count": 9,
"id": "83b695b7",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T20:12:46.317969Z",
"iopub.status.busy": "2023-08-18T20:12:46.316995Z",
"iopub.status.idle": "2023-08-18T20:12:46.501717Z",
"shell.execute_reply": "2023-08-18T20:12:46.500827Z"
},
"origin_pos": 25,
"tab": [
"pytorch"
]
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Sequential output shape:\t torch.Size([1, 64, 24, 24])\n",
"Sequential output shape:\t torch.Size([1, 192, 12, 12])\n",
"Sequential output shape:\t torch.Size([1, 480, 6, 6])\n",
"Sequential output shape:\t torch.Size([1, 832, 3, 3])\n",
"Sequential output shape:\t torch.Size([1, 1024])\n",
"Linear output shape:\t torch.Size([1, 10])\n"
]
}
],
"source": [
"model = GoogleNet().layer_summary((1, 1, 96, 96))"
]
},
{
"cell_type": "markdown",
"id": "ed1f8ec6",
"metadata": {
"slideshow": {
"slide_type": "slide"
}
},
"source": [
"Training"
]
},
{
"cell_type": "code",
"execution_count": 10,
"id": "d52cffee",
"metadata": {
"execution": {
"iopub.execute_input": "2023-08-18T20:12:46.505565Z",
"iopub.status.busy": "2023-08-18T20:12:46.504978Z",
"iopub.status.idle": "2023-08-18T20:16:06.908422Z",
"shell.execute_reply": "2023-08-18T20:16:06.907202Z"
},
"origin_pos": 28,
"tab": [
"pytorch"
]
},
"outputs": [
{
"data": {
"image/svg+xml": [
"\n",
"\n",
"\n"
],
"text/plain": [
""
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"model = GoogleNet(lr=0.01)\n",
"trainer = d2l.Trainer(max_epochs=10, num_gpus=1)\n",
"data = d2l.FashionMNIST(batch_size=128, resize=(96, 96))\n",
"model.apply_init([next(iter(data.get_dataloader(True)))[0]], d2l.init_cnn)\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
}