未验证 提交 9b9f3f09 编写于 作者: Y Yu Yang 提交者: GitHub

Merge pull request #9058 from reyoung/feature/parallel_do_bug

Fix models #725
...@@ -223,6 +223,8 @@ class Optimizer(object): ...@@ -223,6 +223,8 @@ class Optimizer(object):
params_grads = append_backward(loss, parameter_list, no_grad_set, params_grads = append_backward(loss, parameter_list, no_grad_set,
[error_clip_callback]) [error_clip_callback])
params_grads = sorted(params_grads, key=lambda x: x[0].name)
params_grads = append_gradient_clip_ops(params_grads) params_grads = append_gradient_clip_ops(params_grads)
# Add regularization if any # Add regularization if any
......
Markdown is supported
0% .
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