vllm-project/vllm v0.22.1
v0.22.1 is a patch release on top of v0.22.0 with targeted bug fixes plus a couple of additions: new model support for JetBrains' Mellum v2, zentorch-accelerated quantized linear inference on AMD Zen CPUs, and fixes for multi-node Ray data-parallel serving, DeepSeek-V4 initialization, and a few model-loading regressions.
Key points
- This release features 8 commits from 6 contributors (1 new)!
- New model: JetBrains' Mellum v2, an open-weights Mixture-of-Experts code-generation model (#43992).
- Fix HyperCLOVAX loading after the upstream HuggingFace repo removed its remote code (now native in transformers >= 5.9.0): register the hyperclovax modeltype so vLLM uses its vendored config instead of the stale automap (#43860).
- Fix a deterministic hang in multi-node Ray data-parallel serving with numapiservers > 1 by excluding the Ray DP backend from the deferred (kernel-assigned) port allocation introduced in #42585 (#43864).
Sources (1)
- [1]vllm-project/vllm v0.22.1GitHub: vllm-project/vllm · Jun 5, 10:10 AM
v0.22.1 is a patch release on top of v0.22.0 with targeted bug fixes plus a couple of additions: new model support for JetBrains' Mellum v2, zentorch-accelerated quantized linear inference on AMD Zen CPUs, and fixes for multi-node Ray data-parallel serving, DeepSeek-V4 initialization, and a few model-loading regressions.
This release features 8 commits from 6 contributors (1 new)!
Extractive summary: sentences quoted from the sources.