resnet_1001¶
Overview¶
The resnet_1001 function constructs a ResNet-1001 model, an extremely deep residual network built with pre-activation bottleneck blocks, designed for large-scale and computationally intensive image classification tasks.
It uses PreActBottleneck as the building block and is designed for datasets with num_classes categories.
Total Parameters: 149,071,016
Function Signature¶
@register_model
def resnet_1001(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-1001 model.
Examples¶
Creating a ResNet-1001 model for 1000 classes:
model = resnet_1001(num_classes=1000)
print(model)
Note
ResNet-1001 uses a configuration of [3, 94, 94, 3] for its layers.
By default, it initializes weights internally unless specified otherwise through kwargs.