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][Spec Decode] Llama 4 / Mistral Large 3 EAGLE drafters crash at startup with a multimodal target: no attribute `_embed_text_input_ids`
- [Bug]: collect-env is not supported in MAC
- [Bug]: Task cancellation during _commit_scale_down_elastic_ep corrupts cluster state without rollback
- [Bug]: With a KV connector, prompt_logprobs requests are counted as full external prefix cache misses
- kv-cache: honor scheduler lookahead in mamba align-mode allocation
- [Bug]: Sliding-window prefix cache misses entirely when a request diverges inside the last alignment block of a cached prompt (sparse retention keeps a tail exactly `need` blocks long)
- [Bug]: The nested-RoPE validator patch is gated by model type only at install time, then applies to every later config in the process
- [Bugfix][Metrics] Do not count skip_reading_prefix_cache requests as external prefix cache misses
- [Security] Inject Cache Salt By Header
- [Bug]: In version 0.26.0, when scheduler initialization fails, the worker process will not proactively exit or release resources.
- Docs
- Python not yet supported