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 5, 23:30 UTC
main
Various Transformers v5 fixes (#38127)
Harry Mellor committed 8 months ago
[Cohere] Enable Cohere-Transcribe (#38120)
Ekagra Ranjan committed 8 months ago
[ROCm]: Update rope+kvcache fusion conditions and disable custom op by default (#36716)
Rohan Potdar committed 8 months ago
[MultiModal] add support for numpy array embeddings (#38119)
Guillaume Guy committed 8 months ago
[ROCm] Utilize persistent MLA kernel from AITER (#36574)
Sathish Sanjeevi committed 8 months ago
[ROCm] Fix MoE kernel test failures on gfx950 (#37833)
Andreas Karatzas committed 8 months ago
[Core][KV Connector] Remove use of num_cached_tokens in error handling (#38096)
Mark McLoughlin committed 8 months ago
[Frontend] Move APIServerProcessManager target server fn (#38115)
Nick Hill committed 8 months ago
[Misc] Reorganize inputs (#35182)
Cyrus Leung committed 8 months ago
[MoE Kernel] Flashinfer nvfp4 cutedsl moe kernel integration (#38050)
Yongye Zhu committed 8 months ago
[2/n] Migrate per_token_group_quant to torch stable ABI (#36058)
mikaylagawarecki committed 8 months ago
[CI/Docs] Improve aarch64/DGX Spark support for dev setup (#38057)
Ben Browning committed 8 months ago
[compile] Add some more startup tests for top models (#38046)
Richard Zou committed 8 months ago
[Refactor] Rename `WAITING_FOR_FSM` to `WAITING_FOR_STRUCTURED_OUTPUT_GRAMMAR` (#38048)
Wentao Ye committed 8 months ago
[Kernel] Optimize SM120 CUTLASS blockwise FP8 GEMM (#37970)
Necofish committed 8 months ago
[Bugfix] Preserve CUDA arch suffix (a/f) for SM12x — fixes NVFP4 NaN on desktop Blackwell (#37725)
RobTand committed 8 months ago
[Feature] EPLB Support for GPU Model Runner v2 (#37488)
Wentao Ye committed 8 months ago
[Feature] Support per-draft-model MoE backend via `--speculative-config` (#37880)
Andrii Skliar committed 8 months ago
[Docs] Add guide for editing agent instruction files (#37819)
Ben Browning committed 8 months ago