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 supported51 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- MRV2 `StagedWriteTensor.apply_write` computes row offsets in int32 and walks off the end of the buffer once `row_idx * stride(0) > 2**31`
- [Kernel] Fuse AWQ dequantization with GEMM in batch-invariant mode on SM89
- [Bug][ROCm][Distributed] Custom all-reduce start barrier is fully relaxed — peer-data reads unordered against remote (XGMI) writes → torn reads
- [Frontend] Migrate VLLM_ENABLE_RESPONSES_API_STORE to CLI flag
- [Feature]: Integrate CFT counted-write MoE all-to-all kernel
- Engine-initiated aborts end chat completion streams as if they completed: HTTP 200, terminal finish_reason "abort" (not an OpenAI enum value), then an unconditional data: [DONE]
- [Rust Frontend] Add MiniCPM5 tool parser support
- [Bug]: Encountered `openai_harmony.HarmonyError` when using GPT-OSS-120B.
- [Model] Add Xing4_0 support
- [ROCm][AMD] Qwen3.8-2.4T-A95B gfx950 / MI355X Performance Optimization
- Docs
- Python not yet supported