Humans + AI > Humans + AI: Resources > Fine-tuning AI models
Fine-tuning AI models
GPT (Generative Pre-Trained Transformer) Models are, by definition, pre-trained on a body of data. Using these broad models as a base, they can be “fine-tuned” for a specific task using a relevant dataset. Much of the power of generative AI is in being able to fine tune Large Language Models to your specific context and requirements. The following articles provide guidance on how to fine-tune generative AI models, as much as possible for non-technical readers.
Fine tuning
Author: OpenAI, an American artificial intelligence research laboratory
Read article
Description: Learn how to customize a generative AI model for your application, by OpenAI. Fine-tuning is currently only available for the following base models: davinci, curie, babbage, and ada.
Fine-tuning pre-trained models for generative AI applications
Author: Akash Takyar, CEO LeewayHertz
Read article
Description: Fine-tuning pre-trained models as a reliable technique for creating custom high-performing generative AI applications for business-specific use cases.
Unleashing the Power of GPT-3: Fine-Tuning for Superhero Descriptions
Author: Olivier Caelen, Machine Learning Researcher at Worldline
Read article
Description: Step by step guide to generate synthetic data to refine the GPT-3 model and how to do fine-tuning. They show a use case of creating a superhero, but the same method can be used for any use case.
How to fine-tune a GPT-3 model
Author: Kristian, author at All About AI
Read article
Description: A a step-by-step guide on how to fine-tune a GPT-3 large language model.
Recent Advances in Language Model Fine-tuning
Author: Sebastian Ruder, research scientist at Google
Read article
Description: A general overview of recent methods to fine-tune large pre-trained language models.
