Review Core Classes

[dlersch]

While a full example workflow is being developed, we should review the core classes and agree if it offers all the functionality we may or may not need. We will not come to a final final decision, but there should be an agreement among us. For this issue, please:

• Review the core classes
• Note down any concerns questions you may have
• Suggest, if possible, alternatives

Any discussion related to the core classes should be captured in this issue.

[ahmedmohammed107]

1. While I was writing split_dataframe_v0 data prep, keras_mlp_v0, and keras_trainer_v0, I noticed that there are some methods that I defined the exact same way for all three modules such as get_info, save_config, and load_config. So why not define a default behavior for these methods in the core instead of simply raising NotImplementedError in the core itself ? Of course we can override these methods if we want.

2. Can we define in the model core save_config(), save_model(), and save() which calls both save_config() and save_model() similar to what I did in "https://github.com/JeffersonLab/jlab_datascience_core/blob/main/jlab_datascience_toolkit/models/keras_mlp_v0.py" ? Likewise, for load_config(), load_model(), and load()

[dlersch]

@ahmedmohammed107, thank you for bringing this up. lets discuss this in the group