peft
https://github.com/huggingface/peft
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
not yet supported1 Subscribers
Add a CodeTriage badge to peft
Help out
- Issues
- `load_adapter` infers `cuda` for adapter weights even when the base model is on CPU — breaks ZeroGPU Spaces with safetensors >= 0.8
- FEAT Add a 'healthcheck' method to PEFT models
- FIX safe_merge leaving the base layer modified, and merging several adapters at once on torchao
- Hotswapping an adapter can zero another adapter with a shared name prefix
- FIX MultiheadAttention LoRA merge() applies out_proj delta twice
- add_weighted_adapter with SVD based combination types fails for LoRA on Conv1d and Conv3d layers
- Pin UTF-8 encoding when reading/writing adapter config JSON
- Two test-infrastructure bugs found while adding LoHa/LoKr/LN Tuning coverage to decoder-model tests
- add_weighted_adapter silently produces a wrong adapter for use_rslora, lora_bias, DoRA and KaSA
- fix(bnb): apply scaling to lora_bias merge/unmerge on 8-bit and 4-bit
- Docs
- not yet supported