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 supported61 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Core] Update PyTorch to 2.15.0, torchvision to 0.29.1, triton to 3.9.0 (test channel, cu132)
- [Bugfix][Spec Decode] Warm up adaptive verification budget kernel at startup
- [Bugfix][Responses API] Return 400 when streaming retrieval has no stored events
- [Bugfix][KVConnector] Drop deferred notifications for evicted remote engines
- [Bug][ROCm] Two or more vLLM processes on one MI300X (SR-IOV VF) hang within minutes: queue head dispatch accepted, no waves, GFX 100%
- [ROCm][Perf][Qwen4Exp] Tile QSA indexer scoring and skip invisible columns
- [ROCm][Perf][Qwen4Exp] Tune sparse GQA prefill profile on gfx942 and gfx950
- fix: remove misleading `--quantization gptq` hint from Marlin shape errors
- [ROCm][Perf] prefix_prefill: run mixed batches on all XCDs and skip empty query blocks
- [Bug]: `cudagraph_mode=FULL` + FA3 split-KV returns wrong logits: the combine kernel's tile scheduler is chosen from the capture-time dummy batch
- Docs
- Python not yet supported