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 supported62 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Bugfix][CPU] Reuse selected memory node for KV cache sizing
- [Bugfix][Spec Decode] Trim optimistic spec-decode tokens on every PP rank
- [Bugfix][MooncakeStore] Keep connector state for requests preempted and re-admitted in the same scheduler step
- [Bug]: Official MiniCPM-V-4.6 GPTQ/AWQ checkpoints fail to load (vision tower built quantized)
- [Feature]: Support /v1/systemone endpoint for models like convaiinnovations/laya
- updated
- [Perf][Attention] Avoid unused host-length sync for non-causal XQA decode
- [Bug]: bench serve drops or mis-buckets several result fields, including all E2EL metrics for pooling
- [Bug]: bench throughput prints prompt/output token counts and output tok/s but omits them from --output-json
- [Bug]: bench startup --output-json records no model identifier
- Docs
- Python not yet supported