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 supported59 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Feature][LoRA] Allow DP for the TRTLLM MXFP4 MoE-LoRA experts
- [Bugfix][LoRA] Keep the adapters on an idle DP rank's dummy batch
- [Bugfix][LoRA][Spec Decode] Build the DSpark drafter without the target's LoRA config
- [Bug]: MoE oracle picks LoRA-aware experts from the global lora_config, ignoring the per-layer target_modules filter
- [Performance]: Whisper encoder runs eagerly; compiling it gives ~+10% transcription throughput
- [Bugfix][Benchmark] Isolate probe generation length from workload extra_body (#66246)
- [Bugfix][Benchmark] Preserve fractional workload sweep rates
- [Bug]: UVA CPU offload silently undone when replace_parameter rebinds .data (marker survives, re-offload skipped)
- [Bugfix][Offload] Re-offload UVA params whose data replace_parameter rebound
- [Transformers v5] Restore PaddleOCR-VL generation tests with a native HF reference
- Docs
- Python not yet supported