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
- [Unified Cache] Prevent misaligned Mamba checkpoints after speculative truncation
- Fix cgroup in namespace on hicache init
- [Bug] CUDA graphs stay enabled for `torch_native` attention when the platform fallback selects it
- [Bug] parse_arguments turns comma-containing values into Python tuples
- [Bug] PoolsideV1Detector silently retypes nullable/union string arguments
- [Bug] Qwen3CoderDetector streaming emits unterminated JSON arguments when the model omits `</function>`
- [Bug] Qwen3CoderDetector streaming parser fabricates tool calls from plain text, emitting `tool_index=-1
- [Bug] sglang.srt.parser.conversation LLAMA2 style derives role tags from message index parity instead of the message's own role
- [Bug] `get_rope_config` fabricates `rope_theta=10000` for transformers-v5 per-layer-type rope configs
- [Bug] sglang.Engine `schedule_policy="priority"` is an advertised choice that always kills the scheduler
- Docs
- Python not yet supported