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
- [Perf] [AMD] deepseek-v4: Fill a run of identity prefill requests in one batched write
- [Perf] kimi_k3: Batch expert weight transfers during initial loading
- [Perf] kimi_k3: Overlap kernel compilation with checkpoint loading
- [Perf] utils: Avoid per-element list allocation in flatten_hidden for flat float rows
- [Fix] disaggregation: Make the KV checksum digest each state component's own transfer payload
- [Fix] [AMD] GLM-5.3-Flash: Let breakable prefill graphs replay cached-prefix extends, with HIP KDA
- [Perf] qwen2.5-vl: Avoid empty vision windows
- [Perf] multimodal: Avoid CPU thread-pool fanout in SHM copy
- [Fix] multimodal: Preserve processor options when splitting requests
- [Fix] sampling: Validate parallel sample counts before batch expansion
- Docs
- Python not yet supported