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 6, 23:00 UTC
main
[Test] `test_async_scheduling.py` improvements (#36340)
Nick Hill committed 8 months ago
[compile] Apply stored functorch config while finalizing loaded artifacts. (#36582)
Zhengxu Chen committed 8 months ago
[Model Runner V2] Use unpadded num_tokens for PW CUDA graph attn metadata (#36626)
Woosuk Kwon committed 8 months ago
[CI] Bump `mypy` version to 1.19.1 (#36104)
Harry Mellor committed 8 months ago
[ROCm][Perf] Enable `sparse_mla`'s cudagraph on ROCm platform (#35719)
Pleaplusone committed 8 months ago
[Bugfix][Sparse MLA] report indexer CG support properly (#36519)
Matthew Bonanni committed 8 months ago
Improvements to wvSplitKrc skinny GEMM solution (#34304)
Hashem Hashemi committed 8 months ago
AllenDou committed 8 months ago
Fix `hf_override_fn` when it modifies `model_type` (#35200)
Harry Mellor committed 8 months ago
Srinivasoo7 committed 8 months ago
[Bugfix] Avoid merging empty-only partitions into splitting-op subgraphs (#36595)
Jiangyun Zhu committed 8 months ago
fix: check HTTP status in batch read_file to prevent silent failures (#36397)
Alvin Tang committed 8 months ago
[Model] Consolidate score logic by introduce score_type (#36479)
wang.yuqi committed 8 months ago
[Misc] fix typo: homogenous-> homogeneous (2 lines change) (#36508)
SoluMilken committed 8 months ago
[Bugfix] Fix processor signature (#36630)
Raushan Turganbay committed 8 months ago
[Frontend][Core] Revert "Add shutdown timeout" (#34730 and #36270) (#36628)
Mark McLoughlin committed 8 months ago
Fix Qwen2.5-VL test for Transformers v5 (#36532)
Harry Mellor committed 8 months ago
[BUGFIX][Mamba][Qwen3.5] Zero freed SSM cache blocks on GPU (#35219)
Vadim Gimpelson committed 8 months ago
Chang Su committed 8 months ago
[Model Runner V2] Fix mm input embeddings lookup (#36588)
Nick Hill committed 8 months ago