-
Notifications
You must be signed in to change notification settings - Fork 74k
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
Saver errors in 0.7.0 #1135
Comments
Thanks for reporting this bug! You can safely ignore the warnings, which are related to the new The |
1. The cifar10_train.py example model was emitting warnings, because of non-Variable objects in the `tf.moving_average_variables()` collection. This change fixes that by only adding `Variable`-typed objects to that collection in `moving_averages.py` (which better agrees with the definition in `tf.GraphKeys.MOVING_AVERAGES_VARIABLES`). 2. Saver.save() now calls `tf.gfile.MakeDirs(os.path.dirname(save_path))`, which fails if `save_path` does not contain a directory component. This change fixes the implementation of `tf.gfile.MakeDirs('')` to be a no-op (which better matches the internal library that it is shadowing). Fixes #1123. Fixes #1135. Change: 114895020
If someone can validate that the above commit fixes the problem with Saver, that would be appreciated! |
(I was able to run cifar10_train.py with no problems after the above commit). |
It is working fine now, thanks! |
work around TensorFlow bug #1135 <tensorflow/tensorflow#1135> which makes it necessary for a train_dir string to be suffixed with a "/"
In 0.7, there are different errors about Saver, first Warnings in serialization, such as moving averages, or dictionaries:
and then, it throws an OS Error:
But this is working well in 0.6.0, so Saver might have some issues in 0.7.0
The text was updated successfully, but these errors were encountered: