OpenAI Fine-Tuning (0/4)

Hi and welcome to this course on fine-tuning ChatGPT with the OpenAI API. In this course we’ll not just look at how to deal with the fine-tuning API endpoints itself, but also how to run the fine-tuned model, and most importantly, the data preparation and validation steps so that you will be able to do this yourself using any data you want to when you are done.

For our fine-tuning example, we have a really fun one here. We are going to clone Chris! Yes, you read that correctly😉. We will use the Finxter emails Chris always sends us to train our own ChrisGPT! ChrisGPT is going to learn how to write emails in the distinctive Finxter style, on any topic we choose!

  • In part 1 we’ll start by looking at what fine-tuning is and why we would want to do it, and look at the type of data we will need in order to fine-tune our own ChatGPT models, getting started on our data preparation.
  • Part 2 will have us really dive into the data preparation, also using ChatGPT to make our work a bit easier, creating a perfectly formatted JSON dataset for training ChrisGPT.
  • In part 3 we’ll look at flattening our data into JSONL format, and then data validation to make sure we get it right the first time. We’ll also calculate the cost so that we know exactly what we’re getting into.
  • Part 4 is all about the fine-tuning API and how to use it. We’ll train ChrisGPT and then have a lot of fun having our own Robo🤖Chris🧑🏼 write our emails about anything we want!

I hope you’re as excited as I am to get started. Let’s dive in!