Fine Tuning

Definition

Fine-tuning involves taking an existing pre-trained model (such as BERT) and adjusting its parameters using a smaller dataset tailored for a specific application; this process enhances performance while reducing training time significantly compared to building models from scratch.

When Fine Tuning is used

BERT adaptations

Which positions need this?

Machine Learning Practitioners

Problem

Limited computational resources prevent extensive retraining on large datasets.

Example of how Fine Tuning is used in AI

A company fine-tunes a pre-trained language model on its domain-specific documents to improve its accuracy in understanding technical queries within its industry context.