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
- MiniMax-H3: blocky moving face in pre-encode RGB at 32 steps / short_edge 768 (evidence pack; exact prompt withheld)
- [DeepSeek-V4.1] Run encoder SWA bounded replay under the breakable prefill CUDA graph
- [Bug] `--enable-deterministic-inference` does not make gpt-oss-20b deterministic: `test_deterministic --test-mode prefix` returns several outputs for one prompt
- [Bug] `--enable-deterministic-inference` + one request with `repetition_penalty` kills the scheduler on granite-4.0-h (torch.compile `InternalTorchDynamoError` in `apply_scaling_penalties`)
- [DeepSeek-V4.1] Sequence-parallel mHC boundary and fused Engram gather/seam
- [Bug] GPU JPEG decode can overwrite memory that queued GPU work still reads
- num_matched_prefix_tokens stays 0 for cache-agnostic policies (fcfs, lof, random, routing-key)
- Agentic Cache Scheduling System Roadmap (Q4)
- [DSv4] Rebuild the evicted SWA window on a cache hit instead of prefilling again
- [Spec] Fix DFLASH verify mask policy for wrapped attention backends
- Docs
- Python not yet supported