resnet_269¶
Overview¶
The resnet_269 function constructs a ResNet-269 model, an ultra-deep residual network built with pre-activation bottleneck blocks, suitable for large-scale and complex image classification tasks.
It uses PreActBottleneck as the building block and is designed for datasets with num_classes categories.
Total Parameters: 102,069,416
Function Signature¶
@register_model
def resnet_269(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-269 model.
Examples¶
Creating a ResNet-269 model for 1000 classes:
model = resnet_269(num_classes=1000)
print(model)
Note
ResNet-269 uses a configuration of [3, 30, 48, 8] for its layers.
By default, it initializes weights internally unless specified otherwise through kwargs.