Overfitting
When a model memorises the training examples instead of learning the pattern.
A model that scores brilliantly on its training data but poorly on new data has overfitted. It memorised the specific examples, including their noise, instead of learning the general rule. The tell is a widening gap between training performance and validation performance. For language models the same problem shows up as memorising and regurgitating specific documents, which is both a privacy risk and a legal one.
Related terms
Training
The expensive process that sets a model's parameters by exposing it to data.
Dataset
A collection of examples used to train, evaluate, or ground a system.
Evaluation
Measuring whether the system does the job, on a fixed set of cases, repeatedly.
Fine-tuning
Further training of an existing model on your own data to shape its behaviour.
Want this working inside your team?
It starts with one conversation about your workflow.