pytorch-polynomial-lr-decay
pytorch-polynomial-lr-decay copied to clipboard
updata with warm_up
Update a new parameter to linearly warm up the learning rate. More, maybe remove redundant code?