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 supported60 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Bugfix] Fix MLA accuracy drop under DCP with RoPE + KV-cache fusion (GSM8K 77% to 1.5%)
- [CPU] Warm up the remaining sampling kernels before serving
- [Bugfix][Model Runner V2] Embed randomized dummy input ids for inputs_embeds models
- [Perf] Zero bubble draft tokens handoff for structured outputs
- [ROCm][Kimi-K3] Attention kernel cleanup for mixed Prefill + Spec decode steps Kimi K3
- [Bugfix][Config] Preserve Ray DP master address at DP=1
- [Perf][Attention][DCP] Gather the MLA DCP context rows by index instead of reorg_kvcache
- [Model] Nemotron-H: skip layers and override MoE top-k via hf_overrides
- [ROCm][Perf][MLA] Fuse the decode W_UK absorb, q concat + fp8 quant and head pad via AITER
- [Bugfix][MXFP4] Fix MXFP4 CuTeDSL MoE-EP megakernel weight registration
- Docs
- Python not yet supported