resnet_101¶
Overview¶
The resnet_101 function constructs a ResNet-101 model, a deep residual network suitable for large-scale image classification tasks.
It uses Bottleneck as the building block and is designed for datasets with num_classes categories.
Total Parameters: 44,549,160
Function Signature¶
@register_model
def resnet_101(num_classes: int = 1000, **kwargs) -> ResNet:
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¶
ResNet: An instance of the ResNet-101 model.
Examples¶
Creating a ResNet-101 model for 1000 classes:
model = resnet_101(num_classes=1000)
print(model)
Note
ResNet-101 uses a configuration of [3, 4, 23, 3] for its layers.
By default, it initializes weights internally unless specified otherwise through kwargs.