karylab_agents/vllm
1
0
Fork 0
forked from Karylab-cklius/vllm
Code Pull Requests 2 Actions 1 Packages Activity

[Bugfix] compressed-tensors: allow int8 grouped WNA16 MoE on Marlin (#47154)

Browse Source 
Signed-off-by: Joe Rowell <joerowell4@gmail.com>
Co-authored-by: Robert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>
This commit is contained in:
Joe Rowell
2026-06-30 12:50:46 -07:00
committed by GitHub
co-authored by Robert Shaw
parent 345b28ff2f
commit c8d2f3cb14
1 changed files with 0 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe/compressed_tensors_moe_wna16_marlin.py
-1
View File
@@ -80,7 +80,6 @@ class CompressedTensorsWNA16MarlinMoEMethod(CompressedTensorsMoEMethod):
else:
scale = kInt4StaticGroupScale
elif self.num_bits == 8:
assert self.group_size == -1
scale = kInt8StaticGroupScale
else:
raise ValueError(