I'm working with transformer based model, but unable to formulate this learning rate scheduler with PyTorch
lrate = d_model ^ 0.5 * min( step_num ^ 0.5, step_num * warmup_steps ^ -1.5)
Thanks
I'm working with transformer based model, but unable to formulate this learning rate scheduler with PyTorch
lrate = d_model ^ 0.5 * min( step_num ^ 0.5, step_num * warmup_steps ^ -1.5)
Thanks
Copyright © 2021 Jogjafile Inc.