vllm
https://github.com/vllm-project/vllm
Python
A high-throughput and memory-efficient inference and serving engine for LLMs
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.
Python not yet supported50 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [KV offload] Add max_recompute_pending_tokens to skip waiting on short in-flight offloads
- [Feature]: [ROCm][AITER][Hy4] Enable gfx950 MXFP8 MoE and dense linear backends
- [Model] Add AliceAI support
- [Bug]: v0.28.0 — repeated illegal memory access / illegal instruction under production traffic with compressed-tensors per-tensor static FP8 (W8A8) on SM120 (RTX PRO 6000 Blackwell); v0.24.0 stable on the same host and checkpoint
- [KVConnector] Add priority-aware transfer ordering
- [Bugfix][Reasoning] Say why thinking_token_budget is unavailable
- [Bugfix][RL] Fail a weight update without taking the engine down
- [Bugfix] Account for workspace retained by graph profiling before KV sizing
- [Feature][KVConnector][DSv4.1] Encoder-only prefill with Mooncake P/D
- [CPU][Zen] Prepack quantized weights and add DA8W4 for zentorch on Zen CPUs
- Docs
- Python not yet supported