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
add(v1): RequestStatesStats to RequestOutput (#24947)
HUIJONG JEONG committed about 1 year ago
[Perf] Optimize `reshape_and_cache` CUDA Kernel (#25955)
Jiangyun Zhu committed about 1 year ago
[Input] Remove unused `prompt` field (#26097)
Cyrus Leung committed about 1 year ago
[Misc] Remove typing.List (#26150)
Varun Sundar Rabindranath committed about 1 year ago
[BUG] Reorder model config creation (#26124)
ahao-anyscale committed about 1 year ago
`FusedMoE` support for the Transformers backend (#22650)
Harry Mellor committed about 1 year ago
[Model] Use `merge_by_field_config` for MM models (G) (#26117)
Cyrus Leung committed about 1 year ago
[Bugfix] Fix import `gemm_afp4wfp4` failure on AMD (#26068)
Zhewen Li committed about 1 year ago
[ROCm] [VL] [Bugfix] Fix vit flash attn dispatcher logic for ROCm (#26104)
TJian committed about 1 year ago
[Build/CI] Revert back to Ubuntu 20.04, install python 3.12 with uv (#26103)
Tyler Michael Smith committed about 1 year ago
[gpt-oss] disable tool server initialization if no tool in request (#25790)
Andrew Xia committed about 1 year ago
[Bug]: Limit num_reqs in dummy_run when max_num_seqs is small (#26144)
Benjamin Chislett committed about 1 year ago
[DeepSeek] Improve performance of DS MLA cache kernel (#26132)
Matthew Bonanni committed about 1 year ago
[CI/Build] Conditionally register cutlass_fp4_group_mm to fix building on Hopper (#26138)
Michael Goin committed about 1 year ago
[Attention] Move Backend enum into registry (#25893)
Matthew Bonanni committed about 1 year ago
[Bug][Benchmark] Fix duplicate req in oversampling (#26140)
Ekagra Ranjan committed about 1 year ago
[Log] Optimize DeepGEMM Missing Log (#26106)
Wentao Ye committed about 1 year ago
[Bugfix] Disable cascade attention with FlashInfer (#26130)
Michael Goin committed about 1 year ago
Fix MTP with deepep_low_latency (#25904)
Matthew Bonanni committed about 1 year ago
[Perf] Fix and reapply move apply w8a8 block fp8 linear to class (#25696)
ElizaWszola committed about 1 year ago