Sourced from timm's releases.

Release v1.0.25

Feb 23, 2026

• Add token distillation training support to distillation task wrappers
• Remove some torch.jit usage in prep for official deprecation
• Caution added to AdamP optimizer
• Call reset_parameters() even if meta-device init so that buffers get init w/ hacks like init_empty_weights
• Tweak Muon optimizer to work with DTensor/FSDP2 (clamp_ instead of clamp_min_, alternate NS branch for DTensor)
• Release 1.0.25

Jan 21, 2026

• Compat Break: Fix oversight w/ QKV vs MLP bias in ParallelScalingBlock (& DiffParallelScalingBlock)
  • Does not impact any trained timm models but could impact downstream use.

What's Changed

• Token distill task & distill task refactoring by @​rwightman in huggingface/pytorch-image-models#2647
• Fix distilled head dropout using wrong token in PiT forward_head by @​hassonofer in huggingface/pytorch-image-models#2649
• Fix #2653, no models with weights impacted so just a clean fix by @​rwightman in huggingface/pytorch-image-models#2654
• Add the cautious optimizer to AdamP. by @​Yuan-Jinghui in huggingface/pytorch-image-models#2657
• Enhance the numerical stability of the Cautious Optimizer by @​Yuan-Jinghui in huggingface/pytorch-image-models#2658
• Some misc fixes for torch.jit deprecation and meta device init by @​rwightman in huggingface/pytorch-image-models#2664
• fix(optim): replace bare except with Exception in Lion optimizer by @​llukito in huggingface/pytorch-image-models#2666
• Change clamp_min_ to clamp_(min=) as former doesn't work with DTensor / FSDP2 by @​rwightman in huggingface/pytorch-image-models#2668
• Add DTensor compatible NS impl for Muon by @​rwightman in huggingface/pytorch-image-models#2669

New Contributors

• @​Yuan-Jinghui made their first contribution in huggingface/pytorch-image-models#2657
• @​llukito made their first contribution in huggingface/pytorch-image-models#2666

Full Changelog: huggingface/pytorch-image-models@v1.0.24...v1.0.25

• 9326ff2 Release 1.0.25
• 75684fa Add DTensor compatible NS impl for Muon
• feed46c Change clamp_min_ to clamp_(min=) as former doesn't work with DTensor / FSDP2...
• 56467ad fix(optim): replace bare except with Exception in Lion optimizer
• 41754f2 Fix #2661 ... don't skip reset_parameters/init when meta device detected as i...
• bdab30b Remove torch.jit calls except for .ignore and .is_scripting to avoid 2.11 dep...
• 9171d82 Enhance the numerical stability of the Cautious Optimizer
• 17b9764 Add the test of cadamp
• ba85080 Add cadamp optimizer to optimization factory
• a3643d2 Adding the cautious optimizer and the spherical cautious optimizer
• Additional commits viewable in compare view

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)