I have a question regarding model evaluation. I do not see any validation loss or validation metrics in the training script. What is the recommended way to determine whether the model is overfitting or underfitting during FLUX.2 Inpainting + ControlNet training? Is there a standard validation setup, evaluation metric, or early-stopping strategy that you recommend?
I have a question regarding model evaluation. I do not see any validation loss or validation metrics in the training script. What is the recommended way to determine whether the model is overfitting or underfitting during FLUX.2 Inpainting + ControlNet training? Is there a standard validation setup, evaluation metric, or early-stopping strategy that you recommend?