Add an example of importing a CSV to the tuning cookbook

[pcoet]

Description of the feature request:

Add an example of importing a CSV using pandas.read_csv to the tuning cookbook: https://github.com/google-gemini/cookbook/blob/main/quickstarts/Tuning.ipynb

AI Studio accepts csv files as datasets for tuning jobs. The tuning cookbook currently demonstrates creating a tuned model using an list of dictionaries. This is a request for an additional sample demonstrating importing a training dataset from a local csv file using pandas.read_csv, to mirror the functionality of importing a csv file through AI Studio.

What problem are you trying to solve with this feature?

Help users understand how to tune a model using data from a CSV.

Any other information you'd like to share?

No response

[pcoet]

This could also be a new notebook based on the existing tuning notebook.

[markmcd]

I don't think this necessarily needs a new guide, at least not for Python. genai.create_tuned_model() can take a CSV file handle or string filename as the training_data arg.

So this

genai.create_tuned_model(
    training_data=[
        { ... }, { ... }, etc
    ], ...
)

becomes

genai.create_tuned_model(
    training_data='my_file.csv', ...
)

It supports JSON files too.