Dylanself123

Why lr *100 when load pretrained model?

2

Thank you for your work. I wonder why lr*100 and i have found the rate of convergence could get higher with original lr. Thanks a lot. { 'params': parameters['base_parameters'], 'lr':...