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] tokenizer: Fix abort race before scheduler dispatch
- [Feat] dsv4.1: add vit rope kernel opt for dsv4.1.
- fix: AudioFlamingo3 fallback via Transformers backend + generic ASR adapter default
- [Perf] attention: Skip the triton MLA K rescale when k_scale is 1.0
- [Perf] [AMD] attention: Use paged Gluon MLA decode in the aiter backend on gfx1250
- [Perf] [AMD] mla: Fuse RoPE, concat and KV-cache write for MLA decode on gfx1250
- [Feat] sgl-router: Add a Tair KVCache backend to the KV Indexer
- [Fix] model-loader: clone runai-streamed tensors when caching fused a_proj weights
- [Fix] HiCache: Include indexer pools in hybrid host memory budgets
- [Fix] [AMD] KDA: Guard NVIDIA-only decode parity on HIP
- Docs
- Python not yet supported