您好，训练的时候输出都是下面这样的每轮训练时间，请问能不能加上每轮loss的代码

beeper00 commented 3 years ago

I0702 21:27:52.672870 140382066132736 tpu_estimator.py:2159] global_step/sec: 2.20724 INFO:tensorflow:examples/sec: 70.6318 I0702 21:27:52.673159 140382066132736 tpu_estimator.py:2160] examples/sec: 70.6318 INFO:tensorflow:global_step/sec: 2.21268 I0702 21:27:53.124828 140382066132736 tpu_estimator.py:2159] global_step/sec: 2.21268 INFO:tensorflow:examples/sec: 70.8057 I0702 21:27:53.125130 140382066132736 tpu_estimator.py:2160] examples/sec: 70.8057 INFO:tensorflow:global_step/sec: 2.21184 I0702 21:27:53.576927 140382066132736 tpu_estimator.py:2159] global_step/sec: 2.21184 INFO:tensorflow:examples/sec: 70.7789 I0702 21:27:53.577238 140382066132736 tpu_estimator.py:2160] examples/sec: 70.7789 INFO:tensorflow:global_step/sec: 2.19004 I0702 21:27:54.033540 140382066132736 tpu_estimator.py:2159] global_step/sec: 2.19004 INFO:tensorflow:examples/sec: 70.0812 I0702 21:27:54.033850 140382066132736 tpu_estimator.py:2160] examples/sec: 70.0812 INFO:tensorflow:global_step/sec: 2.21437 I0702 21:27:54.485132 140382066132736 tpu_estimator.py:2159] global_step/sec: 2.21437 INFO:tensorflow:examples/sec: 70.8598 I0702 21:27:54.485425 140382066132736 tpu_estimator.py:2160] examples/sec: 70.8598

yuanxiaosc commented 3 years ago

@beeper00 你可以使用TensorBoard监控训练的全过程

newhg commented 2 years ago

请问加上loss了吗，求代码。tf小白不知如何添加

yuanxiaosc commented 2 years ago

每轮的loss统计本身就有，通过TensorBoard可以查看

newhg commented 2 years ago

通过加入logging_hook，模型训练时会输出loss。但tensorboard里面没有loss，只有时间。

------------------ 原始邮件 ------------------ 发件人: @.>; 发送时间: 2022年5月27日(星期五) 中午11:13 收件人: @.>; 抄送: "彳H @.>; @.>; 主题: Re: [yuanxiaosc/Entity-Relation-Extraction] 您好，训练的时候输出都是下面这样的每轮训练时间，请问能不能加上每轮loss的代码 (#69)

每轮的loss统计本身就有，通过TensorBoard可以查看

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

yuanxiaosc / Entity-Relation-Extraction

您好，训练的时候输出都是下面这样的每轮训练时间，请问能不能加上每轮loss的代码 #69