Optimized deepseek-v3/r1 model performance on mxfp4 run (#10008)

Co-authored-by: wunhuang <wunhuang@amd.com>
Co-authored-by: HAI <hixiao@gmail.com>
Co-authored-by: Hubert Lu <55214931+hubertlu-tw@users.noreply.github.com>
This commit is contained in:
kk
2025-09-04 15:11:22 -07:00
committed by GitHub
co-authored by wunhuang HAI Hubert Lu
parent 93088b6975
commit e96973742c
8 changed files with 486 additions and 64 deletions
@@ -0,0 +1,13 @@
from aiter.ops.triton.batched_gemm_afp4wfp4_pre_quant import (
batched_gemm_afp4wfp4_pre_quant,
)
from aiter.ops.triton.fused_mxfp4_quant import (
fused_flatten_mxfp4_quant,
fused_rms_mxfp4_quant,
)
__all__ = [
"fused_rms_mxfp4_quant",
"fused_flatten_mxfp4_quant",
"batched_gemm_afp4wfp4_pre_quant",
]