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 supported60 Subscribers
Add a CodeTriage badge to vllm
Help out
- Issues
- [Model Loader] Find safetensors shards listed in a subdirectory
- [Tests][MoE] Test bias and router weight are applied before the cast
- [Bugfix][Rust Frontend] Fix cancellation during request submission
- [ROCm][Bugfix] W4A16 dense prefill cache: count cached allocator blocks as free in the budget check
- [Perf][LoRA][MoE] Use problem-sized shrink/expand configs for MoE-LoRA batches of up to 256 tokens
- [Kernel][LoRA][MoE] Keep the MoE-LoRA shrink/expand intermediate in fp32 and round the output once
- [Perf][LoRA][MoE] Remove the small-batch fused MoE-LoRA kernel
- [ROCm] Skip RTLD_GLOBAL libtorch_cpu.so promotion when HSA_TOOLS_LIB is set
- [Quantization][MoE] Keep CSF-compressed NVFP4 expert scales resident on B12X
- [Perf][KV Offload] Speed up CPU-only Tiering startup
- Docs
- Python not yet supported