vllm
https://github.com/vllm-project/vllm
Python
A high-throughput and memory-efficient inference and serving engine for LLMs
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 supported59 Subscribers
Add a CodeTriage badge to vllm
Help out
- Issues
- [Perf][Spec Decode] Rejection sample can skip loads in all-greedy case
- [Bugfix][Perf] Exclude draft-only buckets from min_num_layers in KV cache grouping
- [Bugfix] Omit inactive CP checkpointing kwargs in FlashInfer GDN prefill
- [Perf] Pre-warm float32 greedy-with-penalties sampler and GDN RMSNorm buckets
- [Bugfix][KV Connector][Mooncake] Fix transfer address overflow on Intel XPU
- [Bugfix][Frontend] Stop API workers and coordinator on parent loss
- [ROCm][Perf] Capture the routed MoE chain before the shared-expert aux branch for small decode batches
- [Bugfix][Multimodal] Skip the encoder CUDA graph for precomputed embeddings
- [RFC]: Lifecycle hooks for endpoint plugins (readiness and graceful drain), with an out-of-tree Riva-compatible speech API
- [Core][MRv2] Build ModelState through an overridable runner method
- Docs
- Python not yet supported