ikkuna.models

Contents

• ikkuna.models
  • Module contents

Module contents

class ikkuna.models.AlexNetMini(input_shape, num_classes=1000, exporter=None)

Bases: torch.nn.modules.module.Module

Reduced AlexNet (basically just a few conv layers with relu and max-pooling) which attempts to adapt to arbitrary input sizes, provided they are large enough to survive the strides and conv cutoffs.

features

Convolutional module, extracting features from the input

Type:torch.nn.Module

classifier

Classifier with relu and dropout

Type:torch.nn.Module

H_out

Output height of the classifier

Type:int

W_out

Output width of the classifier

Type:int

__init__(input_shape, num_classes=1000, exporter=None)

Parameters:input_shape (tuple) – H, W C (beware of channel order, is different from what you call the model with)

forward(x)

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

class ikkuna.models.DenseNet(input_shape, growth_rate=12, block_config=(16, 16, 16), compression=0.5, num_init_features=24, bn_size=4, drop_rate=0, num_classes=10, small_inputs=True, efficient=False, exporter=None)

Bases: torch.nn.modules.module.Module

Densenet-BC model class, based on “Densely Connected Convolutional Networks” <https://arxiv.org/pdf/1608.06993.pdf>

Parameters:

• growth_rate (int) – how many filters to add each layer (k in paper)
• block_config (list(int)) – 3 or 4 ints, how many layers in each pooling block
• num_init_features (int) – the number of filters to learn in the first convolution layer
• bn_size (int) – multiplicative factor for number of bottle neck layers (i.e. bn_size * k features in the bottleneck layer)
• drop_rate (float) – dropout rate after each dense layer
• num_classes (int) – number of classification classes
• small_inputs (bool) – set to True if images are 32x32 (e.g. CIFAR). Otherwise assumes images are larger.
• efficient (bool) – set to True to use checkpointing. Much more memory efficient, but slower.
• exporter (ikkuna.export.Exporter) – Optional exporter to use for monitoring

forward(x)

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

class ikkuna.models.ResNet(BlockType, num_blocks, num_classes=10, exporter=None)

Bases: torch.nn.modules.module.Module

forward(x)

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

class ikkuna.models.VGG(input_shape, num_classes=1000, exporter=None)

Bases: torch.nn.modules.module.Module

forward(x)

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

class ikkuna.models.FullyConnectedModel(input_shape, num_classes=1000, exporter=None)

Bases: torch.nn.modules.module.Module

forward(x)

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.