Saving models using h5 files for safety #16875
Labels
Comments
|
Hi @naklecha, thanks for your interest in scikit-learn. HDF5 format is not used in scikit-learn and its specific support is unlikely to be added to the library. |
|
Onnx or pmml are more relevant technologies than h5, as noted at
https://scikit-learn.org/stable/modules/model_persistence.html
@cmarmo, users are often unfamiliar with these model export tools. I
suspect we should have better documentation guiding the workflow from
experimentation to production.
|
|
@jnothman, I will have a look to the current documentation and propose something. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Is there a way I can save Sklearn models into h5 files as they are safer than pickle files or joblib files. The issue with pickle files is that a malicious program can easily be introduced while loading pickle files whereas h5 files only contain data and will be very easy to load models even from untrusted sources.
The text was updated successfully, but these errors were encountered: