Broadcast the master weight along with param for distributed training. (#52638)

* Broadcast the master weight along with param for distributed training. * Fix codestyle.

Broadcast the master weight along with param for distributed training. (#52638)
* Broadcast the master weight along with param for distributed training. * Fix codestyle.
d12588d2 · Yiqun Liu · GitHub · ba9a22db · d12588d2
展开全部隐藏空白更改
内联并排

Showing with 814 addition and 484 deletion

python/paddle/distributed/fleet/meta_optimizers/sharding_optimizer.py ...e/distributed/fleet/meta_optimizers/sharding_optimizer.py +814 -484

未找到文件。
--- a/python/paddle/distributed/fleet/meta_optimizers/sharding_optimizer.py
+++ b/python/paddle/distributed/fleet/meta_optimizers/sharding_optimizer.py