-
Notifications
You must be signed in to change notification settings - Fork 19
[Accuracy diff No. 110, 111] Fix accuracy diff for paddle.nn.functional.binary_cross_entropy, paddle.nn.functional.binary_cross_entropy_with_logits API #267
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
如果修改可以全部 pass 的话,请将所有会报 [not compare] 错误的配置,从:tester/api_config/5_accuracy/accuracy_7.txt 移动至 tester/api_config/5_accuracy/accuracy_gpu_error_grads_diff.txt,便于后续复测以及修复~
好的 |
@NKNaN 以及 tester/api_config/5_accuracy/accuracy_gpu_error.txt 中的所有 [not compare] 出错配置,也复制一份到 tester/api_config/5_accuracy/accuracy_gpu_error_grads_diff.txt 中,保留原有的便于复测 ,辛苦同学 |
不好意思同学,是我描述有点问题,最终应该这样处理:
|
我 reset 一下 |
好滴,看上去误删了很多,这个应该是 git 索引的问题,文件偏大就会有这个问题😂 |
重新提了一个 pr |
this pr can be closed safely |
目前 paddle.nn.functional.binary_cross_entropy, paddle.nn.functional.binary_cross_entropy_with_logits 对应的反向算子只计算 x_grad 不计算 label_grad,而 torch 计算 x_grad 和 target_grad (label_grad):


第一次回测结果:

paddle.nn.functional.binary_cross_entropy_with_logits 能全部通过
paddle.nn.functional.binary_cross_entropy 有两个 case 未通过:
对这两个 case 二次回测结果:
