inception_next_atto

lucid.models.inception_next_atto(num_classes: int = 1000, **kwargs) → InceptionNeXt

The inception_next_atto function instantiates an InceptionNeXt-Atto model, a lightweight variant of the InceptionNeXt architecture designed for efficient performance with minimal computational resources.

Total Parameters: 4,156,520

Function Signature

@register_model
def inception_next_atto(num_classes: int = 1000, **kwargs) -> InceptionNeXt

Parameters

  • num_classes (int, optional): Number of output classes for the final classification layer. Defaults to 1000.

  • kwargs (dict, optional): Additional arguments to customize the model configuration.

Returns

  • InceptionNeXt: An instance of the InceptionNeXt model configured as the Atto variant.

Example

from lucid.models import inception_next_atto

# Instantiate InceptionNeXt-Atto
model = inception_next_atto(num_classes=1000)

# Forward pass with a random input
input_tensor = lucid.random.randn(1, 3, 224, 224)
output = model(input_tensor)
print(output.shape)  # Output shape: (1, 1000)