A class that keeps all conversation history.

MobileOne building block. This block has a multi-branched architecture at train-time and plain-CNN style architecture at inference time F

Convolutional FFN Module.

Fake file-like stream object that redirects writes to a logger instance.

Implementation of metaformer block with MHSA as token mixer. For more details on Metaformer structure, please refer to: `MetaFormer Is Actual

Collate examples for supervised fine-tuning.

Collate examples for supervised fine-tuning.

This class implements `FastViT architecture <https://arxiv.org/pdf/2303.14189.pdf>`_

This class implements global pooling with linear projection.

Dataset for supervised fine-tuning.

Dataset for supervised fine-tuning.

r""" Sampler that samples indices in a way that groups together features of the dataset of roughly the same length while keeping a bit of rand

Multi-headed Self Attention module. Source modified from: https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/vision_tr

Convolutional patch embedding layer.

Reparameterizable token mixer. For more details, please refer to our paper: `FastViT: A Fast Hybrid Vision Transformer using Structural Repar

Implementation of Metaformer block with RepMixer as token mixer. For more details on Metaformer structure, please refer to: `MetaFormer Is Ac

Building Block of RepLKNet This class defines overparameterized large kernel conv block introduced in `RepLKNet <https://arxiv.org/abs/2203.0

Squeeze and Excite module. Pytorch implementation of `Squeeze-and-Excitation Networks` - https://arxiv.org/pdf/1709.01507.pdf

LayerNorm only for Channel Dimension. Input: tensor in shape [B, C, H, W]

This class implements `MCi Models <https://arxiv.org/pdf/2311.17049.pdf>`_

Implementation of conditional positional encoding. For more details refer to paper: `Conditional Positional Encodings for Vision Transformers

Different separator style.