Closed ShungJhon closed 8 months ago
Is it normal when training the stage 1, the loss doesn't decrease
To generate models (e.g., Diffusion models), evaluate the results not with loss but with visualization.
Is it normal when training the stage 1, the loss doesn't decrease