修复W4A8 PP8崩溃问题

4020670f · zhuwenwen · 3320343d · 4020670f
Commit 4020670f authored Sep 23, 2025 by zhuwenwen
Show whitespace changes
Inline Side-by-side

Showing with 2 additions and 3 deletions

vllm/model_executor/layers/quantization/utils/w4a8_utils.py vllm/model_executor/layers/quantization/utils/w4a8_utils.py +2 -3

No files found.
--- a/vllm/model_executor/layers/quantization/utils/w4a8_utils.py
+++ b/vllm/model_executor/layers/quantization/utils/w4a8_utils.py
-
 import torch
 import numpy as np

 try:
    from lightop import awq_marlin_repack_w4a8
-    use_lightop = True
+    use_lightop = False
 except Exception:
    use_lightop = False