Steps

Dataset Selection

We begin with the layoric/labeled-multiple-choice-explained dataset, which includes reasoning provided by GPT-3.5-turbo. reasoning explanations serve as a starting point but may differ from Mistral's reasoning style.

1. 00-poe-generate-mistral-reasoning.ipynb: To align with Mistral, we need to create a refined dataset: derek-thomas/labeled-multiple-choice-explained-mistral-reasoning.
2. 01-poe-dataset-creation.ipynb: Then we need to create our prompt experiments.
3. 02-autotrain.ipynb: We generate autotrain jobs on spaces to train our models.
4. 03-poe-token-count-exploration.ipynb: We do some quick analysis so we can optimize our TGI settings.
5. 04-poe-eval.ipynb: We finally evaluate our trained models.

The flowchart is Clickable