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 supported61 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [RFC]: PD disaggregation + RL lifecycle (pause, sleep, abort) tracking
- [Bugfix] Fix M-RoPE positions stride assertion in compile_sizes graphs
- [Bug]: Mamba2 hybrid models crash once the prompt is longer than block_size (regression from #57329)
- [Bug]: sharded_state can't round-trip models with attention/KV-cache scale params (Missing keys ...attn.q_scale, k_scale, v_scale, prob_scale)
- [Bugfix][Frontend] Fix missing input logging for systemone
- [Bugfix][Model Runner V2] Read main_stream at step time, not at first use
- [Bugfix][Attention] Chunk sparse indexer decode logits by request
- [Perf] Emit the DSA indexer weights in BF16 on SM100
- [Perf][Spec Decode] Capture every small decode request count by default
- [Perf][Parser] Find the common tool argument prefix with slice comparisons (32 KB Llama 3 JSON call: 9.8 s to 4.9 s)
- Docs
- Python not yet supported