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
- [Bug]: Mamba-hybrid models cannot start with speculative decoding. Mamba page padding is planned without the speculative widening
- [ROCm] Add native HIP RDNA3/RDNA4 custom all-reduce backend
- [Bug]: `prompt_embeds` + any penalty kills the engine with a device-side assert (`scatter gather kernel index out of bounds`)
- [Bug]: a malformed structured-output specification returns HTTP 500 with an empty error message, and a self-referential `$ref` passes validation and then kills the request mid-generation
- [Bug]: User-authored <|image_pad|> in text is misrecognized as image placeholder, causing image binding mismatch
- [XPU]Fix the accuracy issue when meet topk_ids=-1 on DP+EP scenarios
- [Bug]: GLM5.3-Flash does not support fp8 kv cache dtype on hopper
- [Bug][Reasoning] kimi_k3: structured output never engages when the completion skips the think channel
- [Scheduler] Prevent stop reason leakage across streaming continuations
- [Perf][GLM-5.3-Flash][SM90] Shape-aware launch configuration for recurrent KDA
- Docs
- Python not yet supported