Swinir training configuration issue？[Bug] about mmagic HOT 1 CLOSED

Hello, I encountered the same problem as you. This issue is due to a bug in mmagic where the subclass "BaseSampleWiseMetric" calls the evaluate method from its superclass "BaseMetric". To resolve this issue, modify the evaluate method in the "BaseSampleWiseMetric" class located at 'mmagic/mmagic/evaluation/metrics/base_sample_wise_metric.py' as follows:

def evaluate(self, size: Optional[int] = None) -> dict:
    if size is None:
        size = self.size
    return super().evaluate(size)

This change should fix the problem.

from mmagic.