Fine Tuning

A process in machine learning in which a pre-trained AI model is adjusted to make it better at a specific task or to become an expert in a specific area.

Epochs

An Epoch is one full pass through the entire training dataset during the learning process of an AI model.

A good analogy is studying a textbook. If you read it once that is one epoch, twice that's two epochs. More epochs show the model is still learning and too many can indicate overfitting (memorizing instead of generalizing).

How It Works

The goal is to train the AI to handle specific tasks:

1. Start with a pre-trained model such as an LLM
2. Collect specialized data such as a small focussed dataset related to the required task
3. Train on new data to adjust the models internal weights
   • Epochs - the model reviews the dataset multiple times to learn patterns.
4. Optimization techniques
   • Pruning - removing extra nodes to make the model smaller and faster
   • Quantization - reducing precision to save memory and speed up processing
   • Validation - testing the model on unseen examples to ensure the model is trained effectively
5. Deploy the fine-tuned model
6. Monitor and improve