The loss curve cannot be fitted.

vijaylee / Continual_Anomaly_Detection

Official code for ACMMM 2022 paper: Towards Continual Adaptation in Industrial Anomaly Detection

13 stars 2 forks source link

The loss curve cannot be fitted. #2

Closed mengxujing closed 6 months ago

mengxujing commented 10 months ago

I trained the entire network, and found that only on the first task did the loss curve converge with a flatter loss landscape, but with incremental steps, the loss curves for the new tasks were not fitted.