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
- Init persistent cache for breaking modules in prefill breakable graph
- [Perf][DeepSeek-V4] Fuse inverse RoPE into wo_a FP8 quantization
- [HiCache] Isolate rank-sharded storage keys for MLA hybrid side pools
- [AMD][V4.1] Trim per-cycle bookkeeping launches in DSpark decode
- [Fix] Relay multimodal embeddings for GLM-5.3-Flash PP speculative prefill
- [Bug] #running-req / num_running_reqs grows monotonically under chunked prefill and over-reports load (hits the default config)
- fix(qwen_vl): neutralize the legacy "<image>" sentinel in chat message text
- [Perf] Avoid copying row-padded FlashInfer PAGED top-k logits
- [Bug] --fp4-gemm-backend marlin crashes with NVFP4 W4A4 checkpoints on SM100 (mm_fp4 has no 'marlin' backend)
- [Fix] [AMD] kda: Match the fused verify Q/K norm to its reference kernel
- Docs
- Python not yet supported