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.

Sign up for GitHub

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

Jump to bottom

Should we change the epsilon in batch_norm_layer to a variable instead of a fixed value 1e-5. #5548

Closed

peterzhang2029 opened this issue Nov 10, 2017 · 4 comments · Fixed by #5692

Contributor

peterzhang2029 commented Nov 10, 2017

In other DL platform(such as tensorflow, mxnet.), the default value of epsilon is 0.001

Member

jacquesqiao commented Nov 16, 2017 •

edited

Loading

what do you mean by variable?

Member

jacquesqiao commented Nov 16, 2017 •

edited

Loading

now the epsilon in batch_norm_layer is also an attribute that can be configured. https://github.com/PaddlePaddle/Paddle/blob/develop/python/paddle/v2/fluid/layers.py#L769

Contributor Author

peterzhang2029 commented Nov 16, 2017

I mean the previous v2 version, the implementation is:
https://github.com/PaddlePaddle/Paddle/blob/develop/python/paddle/trainer_config_helpers/layers.py#L3030

Member

jacquesqiao commented Nov 16, 2017

Got it! ~

peterzhang2029 mentioned this issue

Update the epsilon in batch_norm_layer to a variable in v2. #5692

Merged

lcy-seso closed this as completed in #5692

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment