update fp16 gray_list for tensor parallel (#33660)

1681a2dd · WangXi · GitHub · f91dfe15 · 1681a2dd
显示空白变更内容
内联并排

Showing with 2 addition and 0 deletion

python/paddle/fluid/contrib/mixed_precision/fp16_lists.py python/paddle/fluid/contrib/mixed_precision/fp16_lists.py +2 -0

未找到文件。
--- a/python/paddle/fluid/contrib/mixed_precision/fp16_lists.py
+++ b/python/paddle/fluid/contrib/mixed_precision/fp16_lists.py
@@ -146,6 +146,8 @@ gray_list = {
    'cast',
    'fused_bn_add_activation',
    'c_identity',
+    'c_concat',
+    'c_allreduce_sum',
 }

 # The set of ops that don't support fp16 calculation