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
- [Feature][Scheduler] Add adaptive token budget scheduler example
- [Bug]: API server crashes with IndexError in structured-decisions init when tokenizer files are missing (after engine fully loads)
- [Feature][MoE] Use explicit forward-context layouts for routed-expert capture
- [Bug]: [MRV2][Spec Decode] DSpark + enable_adaptive_verification fails at startup (device-side assert in drafter) when max_num_batched_tokens < max_num_seqs × draft query rows
- [Bugfix] deepseek_v32: honor the decoder's deferred MoE all-reduce on the naive all2all path
- [Feature][Elastic EP] Bootstrap external ranks from a shared launch template
- [Model] Serve Base Labs mammoth-1 / elephant-1 with their trained block-bidirectional mask
- [Spec Decode][MRV2]Add LongSpark
- [RFC]: Declared ownership and layout for WeightSource (NCCL broadcast PP + NCCL M2N)
- [Perf][Spec Decode] Warm up the speculative variant of the Qwen GDN recurrent update kernel
- Docs
- Python not yet supported