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
- TokenizerManager: hoist weight_version lookup out of request loop
- memoize stop-string tail decoding within a step
- Detokenizer: call the tokenizers backend decode directly instead of HF batch_decode for stock fast tokenizers
- [Bugfix] dsv4: append assistant prefix for system/reminder-terminated conversations and parse markdown-style pseudo tool calls
- [AMD][DSv4] Tuned decode moe kernels for mori-EP backend
- [Fix] Honor KDA safe-gate backend capabilities
- [Fix] Unpack KDA prefill output regardless of state tracking
- [AMD] Support Qwen3.8-Flash-Next MXFP4 on ROCm with new kernels
- penalties use one step stale history under the overlap scheduler
- [diffusion] Make warmup preferred preload components and headroom configurable
- Docs
- Python not yet supported