learning curve

A learning curve is a diagnostic visualization that plots a model's performance metric (such as error rate, accuracy, or loss) against a measure of training exposure — typically training set size, number of epochs, or training time. The curve reveals how quickly a model learns: steep initial improvement followed by a plateau indicates the model is approaching its capacity; a flat curve suggests insufficient capacity or poor learning signal. Learning curves diagnose bias-variance trade-offs — high bias (underfitting) shows training and validation curves converging at poor performance, while high variance (overfitting) shows a large gap between them. Used to decide whether more data, more features, or more model capacity will help. [formal: tabula progressionis doctrinae | substrate: mind | horizon: hours | explicit: yes | epoch: 0.01]