abzoo/llama3-finetuned-Pima-Indians-Diabetes

Model

I used the finetuning notebook offerd by unsloth.

Complete notebook can be found in notebooks/Create_synthetic_dataset or on GoogleColab

Data

The dataset was created using an existing dataset Pima Indians Diabetes Database

The steps to create and produce the data was as follows:

1. Turn every column of the dataset into into text description of the patient using prepared text templates.
2. Prompt gpt4-o asking for a complete analysis, detailed illustration and diagnostics.
3. Use gpt4-o output to train a smaller model like llama 8b