Skip to content

P
Paddle

机器未来 / Paddle
与 Fork 源项目一致

Fork自 PaddlePaddle / Paddle

通知 1
Star 1
Fork 0
P
Paddle
提交 ed89b7b7 编写于 作者: T typhoonzero
浏览文件
操作

dist train use split_by_ref

上级 0c6eef3e
显示空白变更内容
内联 并排
Showing with 1 addition and 1 deletion
python/paddle/fluid/distribute_transpiler.py
浏览文件 @ ed89b7b7
...@@ -824,7 +824,7 @@ class DistributeTranspiler: ...@@ -824,7 +824,7 @@ class DistributeTranspiler:
for v in splited_vars: for v in splited_vars:
sections.append(v.shape[0]) sections.append(v.shape[0])
program.global_block().append_op( program.global_block().append_op(
type="split", type="split_byref",
inputs={"X": orig_var}, inputs={"X": orig_var},
outputs={"Out": splited_vars}, outputs={"Out": splited_vars},
attrs={"sections": sections} # assume split evenly attrs={"sections": sections} # assume split evenly
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或