Creates an EfficientNet-EdgeTPU model Ref impl: https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet/edgetpu

Creates an EfficientNet-V2 Small model Ref impl: https://github.com/google/automl/tree/master/efficientnetv2 Paper: `EfficientNetV2: Smaller

q,k: [B, N, C]（N 必须等于 H*W） 返回：旋转后的 q,k（形状不变）

Remove CLS token.

3x3 convolution + batch norm

Creates an EfficientNet-CondConv model. Ref impl: https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet/condconv

Creates a hardcorenas model Ref impl: https://github.com/Alibaba-MIIL/HardCoReNAS Paper: https://arxiv.org/abs/2102.11646

Creates a MixNet Medium-Large model. Ref impl: https://github.com/tensorflow/tpu/tree/master/models/official/mnasnet/mixnet Paper: https://ar

Generate bbox and apply lambda correction.

Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks). This is the same as the DropConnect impl I created for E

Revert normalized depth in [0, 1] back to metric depth in [0, `max_depth`].

Performs a single optimization step. Args: closure (callable, optional): A closure that reevaluates the model and returns the los

Parameters ---------- block (nn.Module): Bottleneck class. - For SENet154: SEBottleneck - For SE-ResN

With 50% prob, negate the value

Create a 1D, 2D, or 3D convolution module.

Modified based on https://github.com/mit-han-lab/torchsparse/blob/462dea4a701f87a7545afb3616bf2cf53dd404f3/torchsparse/utils/quantize.py