Wednesday, Jul 29, 2026
shippedTwo releases across PEFT and LLM Compressor
Recoordinate shipped version 0.20.0 of huggingface/peft and version 0.10.0.3 of vllm-project/llm-compressor on Wednesday.
2 people shipped this
The PEFT release [1] added nine new parameter-efficient fine-tuning methods, including HiRA, and expanded the infrastructure with a new image generation benchmark for method comparison and restructured documentation. The LLM Compressor release [2] removed Sparse24 end-to-end tests from the release branch and bumped the upper bounds for pillow and requests dependencies.
python
ml-serving