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 supported51 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Feature]: Recency-based progressive mixed-precision KV cache
- [Bugfix] Fix MiniMax M3 index_topk kernel for non-power-of-2 num_idx_heads (#49157)
- [Bug Fix] DiffusionGemma get different outputs compared with transformers version (mismatch sliding window attention)
- [Bug]: Qwen3.6-35B-A3B (GDN hybrid) intermittently livelocks under load on GB10/SM121 — GPU 96% util, 0 tok/s, no crash, no Xid
- [Bug]: Engine core livelock (100% CPU, no crash) with MTP speculative decoding + xgrammar structured outputs - regression from v0.24.0
- [Bug]: Speculative decoding with a hybrid draft model (LFM2/LFM2.5 short_conv) fails — drafter requires all draft layers in one KV cache group
- [Perf] Shape-adaptive tile config for batch-invariant matmul on SM80
- [Kernel][Attention] Pipeline Triton prefill-attention K/V loads for long BF16/FP16 prefills
- [DSv4 Perf] Right-size gather-k workers to the reachable token count and use num_…
- [DSv4 Perf] Stream sparse-attn compress softmax to cut register pressure and lift…
- Docs
- Python not yet supported