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
- [Feat] [CPU] Model: Enable DeepSeek V4.1 Flash on xeon cpu
- [Feat] dflash: Serve SpecForge KDA drafts through the target KDA layer
- [Perf] disaggregation: Keep prefill KV-send index readback off the scheduler under CP and KV sharding
- [Fix] openai: Reject unk_token_id and colliding IDs in serving_rerank._get_yes_no_token_ids
- [Bug] S3 model URI overrides explicit remote_instance + modelexpress with runai_streamer
- [Perf] [AMD] aiter: FP8 Q for EAGLE target_verify and draft_extend attention with an FP8 KV cache
- [Fix] [AMD] quark: Match exclude entries that name a parameter (.weight) by module
- [Perf] [AMD] aiter/spec: Remove two host syncs from extend metadata and spec M-RoPE positions
- [Perf] [AMD] gdn: ReplaySSM verify uses one pipeline stage at every batch size on gfx950
- [Fix] spec: Greedy requests keep the argmax prefill draft under rejection sampling
- Docs
- Python not yet supported