API Design about chroma

[jeffchuber]

@atroyn does this seem right to you?

[atroyn]

I think the input functions need to be three separate ones, plus a 'bare' one.

• Input the model's training set.
• Input any extra datapoints, for example from an incident. Each one should be a separate set.
• Input data to be prioritized for labeling.

This makes it clearer to the user what they should be doing at each step, even if underneath it all calls the same 'naked' function.

[jeffchuber]

@atroyn I agree with this. We should plan to have thin semantic friendly wrapper's around the log api, and perhaps others as well.

[jmontrose]

FYI, I'm drafting this as an OpenAPI/swagger def with docs.

[jeffchuber]

Closing this, we will continue to evolve the API in code and in the notion