Кривая потерь — график, показывающий, как потери модели меняются по итерациям или эпохам обучения. Она используется для оценки сходимости, диагностики недообучения или переобучения и сравнения обучающих, валидационных или тестовых потерь во времени.
A plot of loss as a function of the number of training iterations. The following plot shows a typical loss curve: !A Cartesian graph of loss versus training iterations, showing a rapid drop in loss for the initial iterations, followed by a gradual drop, and then a flat slope during the final iterations. Loss curves can help you determine when your model is converging or overfitting. Loss curves can plot all of the following types of loss: - training loss - validation loss - test loss See also generalization curve. See Overfitting: Interpreting loss curves in Machine Learning Crash Course for more information.