-
- Notifications
You must be signed in to change notification settings - Fork 5.1k
Pull requests: huggingface/pytorch-image-models
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Fixup casting issues for weights/bias in fp32 norm layers
#2539 by rwightman was merged Jul 7, 2025 Loading…
Add stem_bias option to MNV5. Resolve the norm layer so can pass string.
#2538 by rwightman was merged Jul 6, 2025 Loading…
Add a min layer-decay scale clamp, and no optimization threshold to exclude groups from optimization
#2537 by rwightman was merged Jul 5, 2025 Loading…
Add flag to enable float32 computation for normalization (norm + affine)
#2536 by rwightman was merged Jul 6, 2025 Loading…
For completeness sake, pass pad_type to final UIB block in MSFA for MNV5, minor impact
#2535 by rwightman was closed Jul 4, 2025 Loading…
Update no_grad usage to inference_mode if possible
#2534 by GuillaumeErhard was merged Jul 4, 2025 Loading…
fix: mnv5 conv_stem bias and GELU with approximate=tanh
#2533 by RyanMullins was merged Jul 7, 2025 Loading…
Fix layer decay to work as intended with optimizers
#2532 by Jookare was closed Jul 5, 2025 Loading…
Update README: add references for additional supported models
#2526 by brianhou0208 was merged Jun 24, 2025 Loading…
Add initial weights for my first 3 naflexvit_base models
#2523 by rwightman was merged Jun 20, 2025 Loading…
Fix
head_dim reference in AttentionRope class of attention.py #2519 by amorehead was merged Jun 15, 2025 Loading…
Refactor patch and pos embed resampling based on feedback from https://github.com/stas-sl
#2518 by rwightman was merged Jun 19, 2025 Loading…
Update legacy AdamW impl so it has a multi-tensor impl like NAdamW (n…
#2517 by rwightman was merged Jun 15, 2025 Loading…
Add corrected_weight decay to several optimizers
#2511 by rwightman was merged Jun 11, 2025 Loading…
Throw exception on non-directory path for pretrained weights
#2510 by emmanuel-ferdman was merged Jun 10, 2025 Loading…
Forgot to compact attention pool branches after verifying
#2507 by rwightman was merged Jun 5, 2025 Loading…
Doing some Claude enabled docstring, type annotation and other cleanup
#2504 by rwightman was merged Jun 12, 2025 Loading…
A cleaned up beit3 remap onto vision_transformer.py vit
#2503 by rwightman was merged May 30, 2025 Loading…
Support gradient checkpointing in
forward_intermediates() #2501 by brianhou0208 was merged Jun 21, 2025 Loading…
ProTip! Follow long discussions with comments:>50.