A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
main
[V1][Mamba] - Enable V1 by default for Mamba Models (#23650)
Asaf Joseph Gardin committed about 1 year ago
[Multimodal] Generate mm_hash based on request metadata when caching is turned off (#23690)
Roger Wang committed about 1 year ago
Disable `torch.compile` for dynamic rope models in Transformers backend (#23738)
Harry Mellor committed about 1 year ago
ci: Add arm64 docker build to release pipeline (#23210)
Eli Uriegas committed about 1 year ago
[Docs] Fix warnings in `mkdocs build` (continued) (#23743)
Hyogeun Oh (오효근) committed about 1 year ago
[V1] [Hybrid] Enable compile and piecewise CUDA graph for MiniMax-Text models (#22589)
Thomas Parnell committed about 1 year ago
[Model] Merge `SupportsMultiModalWithRawInput` with `SupportsMultiModal` (#23749)
Cyrus Leung committed about 1 year ago
[BugFix] Fix topk_softmax assert (#19764)
Luka Govedič committed about 1 year ago
[Docs] Remove in-tree Gaudi install instructions (#23628)
Harry Mellor committed about 1 year ago
[LogitsProcs] Deduplicate built-in LP implementation logic (#23362)
Nick Hill committed about 1 year ago
[Doc]: upgrade version of crate-ci tool for improved typo detection (#23755)
Didier Durand committed about 1 year ago
[Model] Enable native HF format InternVL support (#23742)
Isotr0py committed about 1 year ago
[Feature] Add Hopper DeepGEMM E8M0 for DeepSeekV3.1 scale_fmt (#23666)
Wentao Ye committed about 1 year ago
Only run `get_attr_docs` if generating help text (#23723)
Harry Mellor committed about 1 year ago
[Model] Interface to enable batch-level DP support (#23733)
Cyrus Leung committed about 1 year ago
Fix pre-commit on main (#23747)
Harry Mellor committed about 1 year ago
Add vLLM Korea Meetup in the README.md and meetups.md (#23746)
rebel-hongseok committed about 1 year ago
[Model] Explicit `default_pooling_type` interface (#23736)
Cyrus Leung committed about 1 year ago
[V1] [Hybrid] Disable prefix caching by default for hybrid or mamba-based models (#23716)
Thomas Parnell committed about 1 year ago
[Misc] Remove unnecessary `_send_reconfig_message()` in `core_client.py` (#23127)
Nick Hill committed about 1 year ago