data
MobileNetV1Config
extends
ModelConfigMobileNetV1Config(num_classes: int = 1000, in_channels: int = 3, width_mult: float = 1.0, dropout: float = 0.001)Configuration for MobileNet v1.
width_mult — uniform channel multiplier (α in the paper).
1.0 → full model; 0.75 / 0.5 / 0.25 → slimmer variants.
resolution_mult — spatial resolution multiplier (ρ).
Typically applied to the input, not the architecture itself.
Kept here for documentation parity; not enforced inside the model.
dropout — classifier dropout (0.001 in the original paper).