Hi there! I just try to train maskfeat pt model with modified k400_MVITv2_S_16x4_MaskFeat_PT.yaml config, and it seems work well. But it confuse me that the 'lr' and 'top1_loss' is always be 0, is this normal?
This is the std out log, I find the lr in 'SOLVER' is set to 0.0001, but training log show it becomes to 0
Hi there! I just try to train maskfeat pt model with modified k400_MVITv2_S_16x4_MaskFeat_PT.yaml config, and it seems work well. But it confuse me that the 'lr' and 'top1_loss' is always be 0, is this normal?
This is the std out log, I find the lr in 'SOLVER' is set to 0.0001, but training log show it becomes to 0
And my yaml file is like: