We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
There was an error while loading. Please reload this page.
1 parent c5c4aa2 commit afc13b6Copy full SHA for afc13b6
python/paddle/distributed/collective.py
@@ -1013,6 +1013,7 @@ def _parallel_linear(x,
1013
main_block = paddle.static.default_main_program().current_block()
1014
startup_block._find_var_recursive(linear.weight.name).is_distributed = True
1015
main_block._find_var_recursive(linear.weight.name).is_distributed = True
1016
+
1017
# set is_distributed for splited bias
1018
# if a linear layer is splited by row, each rank would hold a complete bias and they should be the same in each rank.
1019
# if a linear layer is splited by col, the bias would also be split into each rank as its weight
0 commit comments