#

Train a small model on CIFAR 10

This trains a small model on CIFAR 10 to test how much distillation benefits.

13import torch.nn as nn
14
15from labml import experiment, logger
16from labml.configs import option
17from labml_nn.experiments.cifar10 import CIFAR10Configs, CIFAR10VGGModel
18from labml_nn.normalization.batch_norm import BatchNorm

#

Configurations

We use CIFAR10Configs which defines all the dataset related configurations, optimizer, and a training loop.

21class Configs(CIFAR10Configs):

#

28    pass

#

VGG style model for CIFAR-10 classification

This derives from the generic VGG style architecture.

31class SmallModel(CIFAR10VGGModel):

#

Create a convolution layer and the activations

38    def conv_block(self, in_channels, out_channels) -> nn.Module:

#

42        return nn.Sequential(

#

Convolution layer

44            nn.Conv2d(in_channels, out_channels, kernel_size=3, padding=1),

#

Batch normalization

46            BatchNorm(out_channels, track_running_stats=False),

#

ReLU activation

48            nn.ReLU(inplace=True),
49        )

#

51    def __init__(self):

#

Create a model with given convolution sizes (channels)

53        super().__init__([[32, 32], [64, 64], [128], [128], [128]])

#

Create model

56@option(Configs.model)
57def _small_model(c: Configs):

#

61    return SmallModel().to(c.device)

#

64def main():

#

Create experiment

66    experiment.create(name='cifar10', comment='small model')

#

Create configurations

68    conf = Configs()

#

Load configurations

70    experiment.configs(conf, {
71        'optimizer.optimizer': 'Adam',
72        'optimizer.learning_rate': 2.5e-4,
73    })

#

Set model for saving/loading

75    experiment.add_pytorch_models({'model': conf.model})

#

Print number of parameters in the model

77    logger.inspect(params=(sum(p.numel() for p in conf.model.parameters() if p.requires_grad)))

#

Start the experiment and run the training loop

79    with experiment.start():
80        conf.run()

#

84if __name__ == '__main__':
85    main()