sglang
https://github.com/sgl-project/sglang
Python
SGLang is a high-performance serving framework for large language models and multimodal models.
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 supported1 Subscribers
Add a CodeTriage badge to sglang
Help out
- Issues
- [Fix] hicache: Recover storage worker collectives and expose storage liveness
- [Fix] hicache: Bound storage backend calls and keep the scheduler alive when a worker wedges
- [Perf] hicache: Add an opt-in admission throttle for the buffer-only write backlog
- [Fix] hicache: Flush the cache without dying when a storage worker cannot be stopped
- Step 5 Preview Support
- [Refactor] [NPU] diffusion: :recycle: Reuse shared MXFP8 linear execution
- [Feature] [RFC] Opt-in `toktier` tokenizer backend: exact incremental and chunk-parallel tokenization on the CPU
- [Feat] [CPU] cpu-graph: Add prefill torch compile graph and dynamic graph support
- [Bug] Chat returns tool_choice=required calls as plain content when no --tool-call-parser is set
- [Fix] spec-v2: pass pp_proxy_tensors through FrozenKVMTP/UNO workers, declare interface on BaseSpecWorker
- Docs
- Python not yet supported