Fine-Tuning Using The Bakery: Step-by-Step Guide

This guide will walk you through the process of fine-tuning a model using datasets in The Bakery.

Step 1: Creating Your Raw Fine-Tuning Dataset

Start by navigating to the top of your screen and selecting Create Dataset.

Next, choose a name for your dataset, assign a category and tags, and select Raw Dataset as the dataset type.

A raw dataset contains unprocessed data that will be used to train and fine-tune your model. Unlike a vector dataset, which is optimized for querying or searching, raw datasets are essential for fine-tuning because they contain detailed training data without any preprocessing.

Once the dataset is created, it will be labeled with the special identifier RAW.

Now, navigate to the newly created dataset and upload the file for the fine-tuning process. In this example, we are using a file in Parquet format.

Step 2: Fine-Tuning Your Model

While in the Files section, locate the Fine Tune option on the right-hand side of your screen.

You can now begin the fine-tuning process by filling in the required details. For this tutorial, we’ll be using a base model from the marketplace. For more information on purchasing and listing models or datasets on The Bakery, refer to the documentation.

Create a name for the fine-tuned model.