Saving a model's structure to reuse it in the future

[Andrew Rodriguez]

Hi guys,

I’m making some tests with a simple dataset that I’m using to fit a LLM model (I named it "Fitting Data"). 

Now, I want to save the model's structure to a folder so I can reuse it in the future with a new dataset that has a similar structure in terms of input variables to "Fitting Data," but where I don't know the target values.

Is there a way to do this? I hope I've explained everything clearly.

[Alessandro Piazza]

Hi Andrew! Yes, you can do it!

Let’s take a look at this flow, where I divided “Fitting Data” in one line and “Future Data” in the other:

In the upper line of the flow, we performed these steps:

• Fit a LLM model to the “Fitting Data”.
• Converted the ruleset generated by the LLM into a dataset (if you use a type of model that doesn’t export rules, you need a Model to Dataset task).
• Exported the datamanager into a folder (you can choose the format and the folder that you prefer).

In the lower line of the flow, we performed these steps:

• We re-imported the file containing the rules in data structure.
• We re-converted the rules contained in this file from data structure to rules structure, through a Dataset to Ruleset task.
• With a Select Flows task, we kept the rules from the Dataset to Ruleset task and the data from the “Future Data” dataset.
• Through an Apply Model task, we applied the rules to the data .

And we are done!