{ "

Train a large model on CIFAR 10

\n

This trains a large model on CIFAR 10 for distillation.

\n": "

\u5728 CIFAR 10 \u4e0a\u8bad\u7ec3\u4e00\u4e2a\u5927\u578b\u6a21\u578b

\n

\u8fd9\u4f1a\u5728 CIFAR 10 \u4e0a\u8bad\u7ec3\u4e00\u4e2a\u7528\u4e8e\u84b8\u998f\u7684\u5927\u578b\u6a21\u578b\u3002

\n", "

Configurations

\n

We use _^_0_^_ which defines all the dataset related configurations, optimizer, and a training loop.

\n": "

\u914d\u7f6e

\n

\u6211\u4eec\u4f7f\u7528_^_0_^_\u5b83\u6765\u5b9a\u4e49\u6240\u6709\u4e0e\u6570\u636e\u96c6\u76f8\u5173\u7684\u914d\u7f6e\u3001\u4f18\u5316\u5668\u548c\u8bad\u7ec3\u5faa\u73af\u3002

\n", "

Create model

\n": "

\u521b\u5efa\u6a21\u578b

\n", "

VGG style model for CIFAR-10 classification

\n

This derives from the generic VGG style architecture.

\n": "

\u9002\u7528\u4e8e CIFAR-10 \u5206\u7c7b\u7684 VGG \u6837\u5f0f\u6a21\u578b

\n

\u8fd9\u6e90\u4e8e\u901a\u7528\u7684 VGG \u98ce\u683c\u67b6\u6784\u3002

\n", "

\n": "

\n", "

Create a convolution layer and the activations

\n": "

\u521b\u5efa\u5377\u79ef\u5c42\u548c\u6fc0\u6d3b

\n", "

Batch normalization

\n": "

\u6279\u91cf\u6807\u51c6\u5316

\n", "

Convolution layer

\n": "

\u5377\u79ef\u5c42

\n", "

Create a model with given convolution sizes (channels)

\n": "

\u4f7f\u7528\u7ed9\u5b9a\u7684\u5377\u79ef\u5927\u5c0f\uff08\u901a\u9053\uff09\u521b\u5efa\u6a21\u578b

\n", "

Create configurations

\n": "

\u521b\u5efa\u914d\u7f6e

\n", "

Create experiment

\n": "

\u521b\u5efa\u5b9e\u9a8c

\n", "

Dropout

\n": "

\u8f8d\u5b66

\n", "

Load configurations

\n": "

\u88c5\u8f7d\u914d\u7f6e

\n", "

Print number of parameters in the model

\n": "

\u6253\u5370\u6a21\u578b\u4e2d\u53c2\u6570\u7684\u6570\u91cf

\n", "

ReLU activation

\n": "

\u6fc0\u6d3b ReLU

\n", "

Set model for saving/loading

\n": "

\u8bbe\u7f6e\u4fdd\u5b58/\u52a0\u8f7d\u7684\u6a21\u578b

\n", "

Start the experiment and run the training loop

\n": "

\u5f00\u59cb\u5b9e\u9a8c\u5e76\u8fd0\u884c\u8bad\u7ec3\u5faa\u73af

\n", "Train a large model on CIFAR 10": "\u5728 CIFAR 10 \u4e0a\u8bad\u7ec3\u4e00\u4e2a\u5927\u578b\u6a21\u578b", "Train a large model on CIFAR 10 for distillation.": "\u5728 CIFAR 10 \u4e0a\u8bad\u7ec3\u4e00\u4e2a\u5927\u578b\u6a21\u578b\u8fdb\u884c\u84b8\u998f\u3002" }