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
- Add anti-starvation aging to the lpm schedule policy (--lpm-aging-tokens-per-pass)
- [Fix] Align KV cache eviction priority direction with scheduler when --schedule-low-priority-values-first is enabled
- [Fix] Validate user-supplied input_ids against the vocabulary
- [CI] Route b200 diffusion tests through pip FA4 to match consistency references
- [Fix] Call post_load_weights on direct/custom weight-update paths
- [Feature] KVarN: KV-cache compression for hybrid linear-attention models
- [FlashInfer] Plan wrappers from layer-local attention metadata
- [Bugfix] Bound the grammar-compile thread pool and expose SGLANG_GRAMMAR_COMPILE_MAX_WORKERS
- [FlashInfer] Add SM12x b12x NVFP4 MoE support (Laguna S 2.1 + DFlash working in NVFP4 on SM121)
- [DSv4][Bugfix] Fix DSpark verifier state rewrite window
- Docs
- Python not yet supported