Repository Issue Activity (beta)

#123 Add FNet as a optional implementation of ViT

#362 perf: Fix-03 fractal_vit.py .tolist() 同步点（搁置 — 需数据结构重构）

#358 NaViT token dropout incompatible with torch.use_deterministic_algorithms(True)

#359 NaViT token dropout: device mismatch between keep_indices and sequence tensor

#340 How to train？

#357 About DINO hyper-parameter

#347 NaViT is slower than expected

#352 [Bug] Incorrect dimension ordering in ViT3D Rearrange patch embedding

#351 Tensor size mismatch: mae incompatible with encoder with pool = 'mean'

#349 I hope the author can incorporate the implementations of FastViT and FastViTHD.

#348 `pytest-runner` deprecation

#344 Possible redundant parameter `mlp_dim` in `class Transformer` definition?

#320 SimpleViT misleading summary

#296 Problems regarding training 3D Vision transformer : model does not converge

#185 where is the train.py? tks

#255 Apply Tanh activation function to ViT - MLP Head

#337 How can I use a 1D Vision Transformer?

#335 Why not always project out from Attention block?

#336 Allow to use classification token in CvT?

#332 train navit nest 3d error when backward

#214 issue VIT + Loss function

#330 RegionViT - Local token embedding

#329 The Total params: and Params size (MB) of the model printed by summary are different from the bit_base model in timm library. Theoretically, the same settings should be the same. What is the reason?

#328 MS-COCO training from Imagenet pretrained checkpoint

#13 how to get train.zip.