vllm/quantization at 970d6d0776076f17604077ba4d484cdadd604ceb - vllm - Luminance Code Repo

20231088/vllm

History

Tyler Michael Smith 970d6d0776

[Build][Kernel] Update CUTLASS to v3.6.0 (#11607 )

Signed-off-by: Tyler Michael Smith <tyler@neuralmagic.com>

2024-12-30 17:22:13 +08:00

..

[Kernel] fix types used in aqlm and ggml kernels to support dynamo (#7596 )

2024-08-16 14:00:11 -07:00

[CI/Build] Suppress divide-by-zero and missing return statement warnings (#7001 )

2024-08-05 16:00:01 -04:00

compressed_tensors

[BugFix] [Kernel] Fix GPU SEGV occurring in int8 kernels (#9391 )

2024-10-17 01:34:06 +00:00

[Kernel] Refactor Cutlass c3x (#10049 )

2024-12-19 07:00:18 +00:00

[torch.compile] Dynamic fp8 + rms_norm fusion (#10906 )

2024-12-13 03:19:23 +00:00

[torch.compile] Dynamic fp8 + rms_norm fusion (#10906 )

2024-12-13 03:19:23 +00:00

[AMD] Add support for GGUF quantization on ROCm (#10254 )

2024-11-22 21:14:49 -08:00

[Kernel][Misc] Use TORCH_LIBRARY instead of PYBIND11_MODULE for custom ops (#5047 )

2024-06-09 16:23:30 -04:00

[Model][Quantization] HQQ support through Marlin kernel expansion (#9766 )

2024-11-19 13:31:12 -08:00

[Build][Kernel] Update CUTLASS to v3.6.0 (#11607 )

2024-12-30 17:22:13 +08:00

[Kernel] Remove if-else with identical branches in marlin 2:4 (#10687 )

2024-11-26 22:55:32 -08:00

vectorization.cuh

[torch.compile] Dynamic fp8 + rms_norm fusion (#10906 )

2024-12-13 03:19:23 +00:00