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
- [Fix] PD: Retain Mooncake buffers until transfer drain
- [Feat] attention: Support sparse query positions in FlashInfer prefill
- [Fix] mooncake: Support Ascend Direct endpoint naming
- [Fix] kt: Pass the layer's SwiGLU clamp and SwiGLU-OAI to the CPU experts
- [Feat] dsv41: FP8 MegaAttention prefill workspace for DeepSeek-V4.1-Flash
- [Feat] [Apple] mps: Warm up Unlimited-OCR v2 at startup and size its KV pool from the window budget
- [Perf] [Apple] mps: Encode and prefill the next Unlimited-OCR v2 window under the current decode
- [Feat] nemotron_labs_diffusion: Add AR and block-diffusion support
- [Perf] layernorm_sp: Add --layernorm-sp-min-tokens to run small extend passes replicated
- [Fix] disaggregation: Bind TransferKVChunk to room lifecycle generation
- Docs
- Python not yet supported