[Bugfix] compressed-tensors: restore int8 grouped WNA16 MoE support (#52002)

Signed-off-by: Yohann Prigent <prigent.yohann@gmail.com> Co-authored-by: Misha Goin <mgoin64@gmail.com>
2026-08-19 07:55:59 -07:00
parent 2d7f42b4f3
commit be06873198
1 changed files with 0 additions and 1 deletions
@@ -82,7 +82,6 @@ class CompressedTensorsWNA16MoEMethod(CompressedTensorsMoEMethod):
            else:
                scale = kInt4StaticGroupScale
        elif self.num_bits == 8:
-            assert self.group_size == -1
            scale = kInt8StaticGroupScale
        else:
            raise ValueError(