How do I fine-tune based on csv data in columns

Hello everyone. Could you help me? I have data in a csv file. There are multiple columns, such as quarter (of the year) that the result refers to, the group that the result refers to, the geographic location of the result, and of course, the percentage result.

There are about 400 rows of data. Can I fine-tune a GPT model with a sample of these data, and have it provide answers based on the entire dataset? Thanks :slight_smile:


could you find an answer to it? Same problem here …

Follow up: did you find the solution? same issues

Hi - the above is not a use case for fine-tuning.

You would use embeddings-based search to achieve this. Have a look at these resources for further details on embeddings:

And for reference when fine-tuning is suitable:

1 Like