se_resnet_101

lucid.models.se_resnet_101(num_classes: int = 1000, **kwargs) → SENet

Overview

The se_resnet_101 function constructs an SE-ResNet-101 model, a deep residual network with SE blocks for adaptive feature recalibration, designed for large-scale image classification tasks.

Total Parameters: 49,326,872

Function Signature

@register_model
def se_resnet_101(num_classes: int = 1000, **kwargs) -> SENet:

Parameters

  • num_classes (int, optional): Number of output classes for the classification task. Default is 1000.

  • kwargs: Additional keyword arguments to customize the model.

Returns

  • SENet: An instance of the SE-ResNet-101 model.

Examples

Creating an SE-ResNet-101 model for 1000 classes:

model = se_resnet_101(num_classes=1000)
print(model)

Note

  • SE-ResNet-101 uses a configuration of [3, 4, 23, 3] for its layers.

  • Incorporates _SEResNetBottleneck for SE operations.

  • Initializes weights internally unless specified otherwise through kwargs.