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][Core] Keep cache-registered Mamba states out of retirement
- [Perf][DSV4.1] Project sequence shards before all-gather
- [Bugfix] Run DeepGEMM warmup before memory profiling
- [Perf][Hybrid] Reserve Mamba speculative-decoding scratch per running request instead of per KV-concurrency slot
- [Frontend] Skip empty generate stream outputs only while prompt metadata is pending
- [Bug]: Hybrid Mamba2 + prefix caching with an explicit `--mamba-block-size`, `conv_ssm_forward` writes SSM state from an out-of-range strided slice and kills EngineCore.
- [Performance] Enable CUDA BF16 GQA RoPE KV-cache fusion
- [Model] Register and implement Glm4MoeLitePlusPlus model
- [Bugfix][EPD] Preserve raw fallback identity and transfer lifecycle
- [Bug]: MRV2不支持anthropic?
- Docs
- Python not yet supported