We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
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
Enable pickle serialzation of Models with Merlin Schema Saved alongside.
Part of #669
Additional context
Noticed on this PR NVIDIA-Merlin/Merlin#770
We currently get a FileNotFoundError when tryting to pickle a model.
FileNotFoundError
This is because keras uses the tf.io in-memory filesystem which I think is a tensorflow io feature (I've not seen ram:// used outside of this context at least) when trying to pickle a model. And to use this RAM filesystem, you can't use regular python file io functionality it seems. We'll need to use tf.io functions to write things into this in-memory filesystem.
This looks like it's been replaced in TF 2.11 by a regular regular temporary directory on disk. However, since we currently aim to support versions back to TF 2.8, will need to handle both cases now.
The text was updated successfully, but these errors were encountered:
oliverholworthy
Successfully merging a pull request may close this issue.
Description
Enable pickle serialzation of Models with Merlin Schema Saved alongside.
Part of #669
Additional context
Noticed on this PR NVIDIA-Merlin/Merlin#770
We currently get a
FileNotFoundError
when tryting to pickle a model.This is because keras uses the tf.io in-memory filesystem which I think is a tensorflow io feature (I've not seen ram:// used outside of this context at least) when trying to pickle a model. And to use this RAM filesystem, you can't use regular python file io functionality it seems. We'll need to use tf.io functions to write things into this in-memory filesystem.
This looks like it's been replaced in TF 2.11 by a regular regular temporary directory on disk. However, since we currently aim to support versions back to TF 2.8, will need to handle both cases now.
The text was updated successfully, but these errors were encountered: