Swin-Transformer · Issues· 202 open

KeyError: 'LOCAL_RANK'

Question about `relative_coords_table` buffer in Swin V2.

Use of pretrained_window_size

3D Implementation

can i use negative sample of images /background images for training does it take

The model can not converge

Inconsistent Accuracy Jump on Resume – Potential Scaler Issue?

How do I use swin-transformer to predict the category of images in my own folder and show size mismatch when I load the model

How to input custom image size data, such as 64×64, when training my own dataset using the Swin-TransformerV2?

Clarification on Speed Improvement with `fused_window_process` and Its Necessity for Small-Scale Tasks

CLS token

Why do you use different weight_decay ?

swin_transformer_v2.py RuntimeError Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu! (when checking argument for argument max in method wrapper_CUDA_clamp_Tensor)

Structure of SwinT.

Equivalence between Patch Merging and Conv.