EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[Sync] Update mxfp4.py (20250827) (#9724)

Browse Source 
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Shiyang Chen <shiyang@x.ai>
This commit is contained in:
Lianmin Zheng
2025-08-27 17:00:09 -07:00
committed by GitHub
parent 28684f909d
commit 4aeba40d7b
1 changed files with 9 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
python/sglang/srt/layers/quantization/mxfp4.py
View File

@@ -66,10 +66,15 @@ _is_hip = is_hip()
if _is_hip:
# import aiter
from aiter import ActivationType, QuantType, dtypes
from aiter.fused_moe import fused_moe
from aiter.ops.triton.quant import dynamic_mxfp4_quant
from aiter.utility.fp4_utils import e8m0_shuffle
try:
from aiter import ActivationType, QuantType, dtypes
from aiter.fused_moe import fused_moe
from aiter.ops.triton.quant import dynamic_mxfp4_quant
from aiter.utility.fp4_utils import e8m0_shuffle
except ImportError as err:
ActivationType = QuantType = dtypes = fused_moe = dynamic_mxfp4_quant = (
e8m0_shuffle
) = err
def _swizzle_mxfp4(quant_tensor, scale, num_warps):