提交 160aa641 编写于 作者: T typhoonzero

follow comments

上级 ded49533
......@@ -235,6 +235,7 @@ mini-batch forward的output的值。获得的值类型均为 :code:`numpy.ndarra
"average absolute diff scores: %.6f") %
(event.pass_id, event.batch_id, event.cost, diff))
注意:此方法不能获取 :code:`paddle.layer.recurrent_group` 里step的内容,但可以获取 :code:`paddle.layer.recurrent_group` 的输出。
7. 如何在训练过程中获得参数的权重和梯度
-----------------------------------------------
......
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