Repository Issue Activity (beta)

#1886 [FEATURE] Comet ML Integration

#68 AdvProp implementation

#871 [BUG] Training job cannot resume if LR scheduler is Plateau.

#2323 [BUG] inference.py expects labeled subfolders in the --data-dir folder

#1597 Pruned efficientnets don't respect the `in_chans` parameter

#2370 [BUG] feature_info.channels() is not match

#2542 [BUG] naflexvit_so400m_patch16_siglip has undocumented different default pos_embed_interp_mode of "bicubic" instead of "bilinear"

#2729 [FEATURE] Adding ISNet as new model

#2725 [BUG] `timm.create_model` failed when local path contain valid character "#"

#2711 [FEATURE] Add MogaNet (ICLR 2024) backbone

#2279 [FEATURE] LowFormer Support

#2690 [FEATURE] EUPE pretrained weights

#2419 Inference.py script is not returning the custom labels on test images

#2335 OmniVec:Learning robust representations with cross modal sharing

#2712 [BUG] reg_token excluded from group_matcher()/no_weight_decay() in Eva and VisionTransformer

#2177 [FEATURE] Add ViT weights: RADIO

#2363 [FEATURE] Support for RWKV and MAMBA architecture

#2700 [FEATURE] Add TIPSv2 model

#2693 [BUG] torch.compile applied at task level has no effect during validation and causes multi-node stalls

#2691 [BUG] The behaviour of `param_groups_layer_decay()` towards `requires_grad==False` parameters

#2687 [BUG] FX tracing fails for ViT due to dynamic is_causal passed to scaled_dot_product_attention

#2095 [FEATURE] License in csv

#1938 [FEATURE] Support for PyTorch-XLA 2.0

#2681 Potential mismatch with original DINOv3: timm models use avg pooling

#2551 [FEATURE] Support for DPT decoder