Image to Patch Embedding

Vision Transformer with support for patch or hybrid CNN input stage

Set the learning rate of each parameter group using a cosine annealing schedule, where :math:`\eta_{max}` is set to the initial lr and :math

Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).

Base class for bottlenecks that implements `forward()` method.

Bottleneck for SENet154.

ResNeXt bottleneck type C with a Squeeze-and-Excitation module.