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
- [Bug][ROCm] GLM-5.3-Flash: recent-context pools not guaranteed in kpool top-k selection (always_select_tail covers only the incomplete pool)
- [Feature][ROCm]: GLM-5.3-Flash prefill checkpoints
- [MoE Refactor] Call monolithic MoE kernels from one place in FusedMoEMethodBase
- [KVConnector][NIXL] Pair push-mode requests by router transfer_id
- [ROCm] Add defensive group-size validation for W4A16 kernels
- [Bug][CPU] apply_top_k_top_p_triton drops tokens that top-p must keep (kernel logic; reproduces end to end on V1 and V2)
- [Frontend][Responses] Support client-executed tool_search
- [Bug]: LoRA adapters with PEFT rank_pattern / alpha_pattern are served with the wrong scaling
- [Bug]: Engine fails to start with `--enable-expert-parallel` and `--load-format ipc_cache`; the disk fallback fails too
- [Bugfix][Sampler] Keep tokens top-p must keep in Triton top-k/top-p search
- Docs
- Python not yet supported