resnet_34¶
Overview¶
The resnet_34 function constructs a ResNet-34 model, a deeper residual network suitable for image classification tasks with more complex datasets.
It uses the preset ResNetConfig(block=”basic”, layers=[3, 4, 6, 3]) and accepts additional ResNetConfig keyword overrides such as in_channels, stem_type, stem_width, avg_down, channels, and block_args.
Total Parameters: 21,797,672
Function Signature¶
@register_model
def resnet_34(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 forwarded to ResNetConfig, excluding the preset block and layers fields.
Returns¶
ResNet: An instance of the ResNet-34 model.
Examples¶
Creating a ResNet-34 model for 1000 classes:
model = resnet_34(num_classes=1000)
print(model)
Note
ResNet-34 uses BasicBlock with a stage configuration of [3, 4, 6, 3].
The returned model is equivalent to ResNet(ResNetConfig(…)) with the preset values above.