This message was deleted.
# neural-forecast
s
This message was deleted.
k
Most models have early stopping by default, you can control its definition through ‘early_stop_patience_steps’, ‘valid_loss’, ‘valid_check_steps’. Here is the documentation of the NHITS model. https://nixtla.github.io/neuralforecast/models.nhits.html#nhits Additionally if you are interested in the source code check this example of the inner workings. https://github.com/Nixtla/neuralforecast/blob/main/neuralforecast/common/_base_windows.py#L126
👍 1