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
- [Bugfix][KV Connector] Cap MooncakeStore lookup at num_tokens - 1
- [Kernel] persistent_topk: deterministic select, faster than the exact-topk workaround
- [Bug][ROCm] Sparse-MLA indexer reserves a 32 GiB decode-logits workspace during profiling (max_num_batched_tokens rows), costing 21 % of the KV cache
- [Bug]: PP performance drop when MRV2 is set by default.
- [Bugfix][Parser] GLM-4.7: stray arg tags must not leak into tool-call argument keys
- [Bug]: tool_choice='none' silently deletes tool-call-shaped content
- [Kernel][HY-V4] iHC: head op, custom-op registration, tensor-core path for large batches, warmup and tests
- [ROCm] Fuse MLA dual RMSNorm + FP8 group quant for DeepSeek-R1
- [Bugfix] Clamp memory budget to free memory when GPU is shared with another process
- [Perf][GLM-5.3-Flash] Shard long-context indexer prefill rows across TP
- Docs
- Python not yet supported