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 supported62 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Bugfix][Quantization] Count only checkpoint elements for padded online-quant MoE weights
- [Bugfix] Skip request body JSON serialization when its log level is disabled
- [Bugfix][Model Loader] Evict loaded shards when checkpoint exceeds RAM
- [Attention] Support NVFP4 KV cache at head_dim 512 (Gemma 4) on SM100
- [Bug]: prompt_logprobs=0 bypasses fast-prefill guard, causing V1 crash or V2 silent errors
- [Bug]: FP8 MoE per-tensor + CUDA graphs causes cross-request contamination via padding row amax
- [Kernel] Use 16-row tiles for SM89 int8 CUTLASS GEMM at M in (16, 64] when they don't add waves
- [Bug]: speculative_config.method="ngram_gpu" with compilation_config.mode="NONE" kills engine init — ValueError: No compilation mode is set.
- [Bugfix] Exclude cudagraph padding rows from the per-tensor FP8 MoE amax
- [Bugfix][Multimodal] Auto derive max_num_batched_tokens from max_model_len and max_num_seqs
- Docs
- Python not yet supported