src/optim.py [118:127]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
    ):
        super().__init__(
            params, lr=warmup_init_lr, betas=betas, eps=eps, weight_decay=weight_decay,
        )

        # linearly warmup for the first warmup_updates
        self.warmup_updates = warmup_updates
        self.warmup_init_lr = warmup_init_lr
        warmup_end_lr = lr
        self.lr_step = (warmup_end_lr - warmup_init_lr) / warmup_updates
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



src/optim.py [180:189]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
    ):
        super().__init__(
            params, lr=warmup_init_lr, betas=betas, eps=eps, weight_decay=weight_decay,
        )

        # linearly warmup for the first warmup_updates
        self.warmup_updates = warmup_updates
        self.warmup_init_lr = warmup_init_lr
        warmup_end_lr = lr
        self.lr_step = (warmup_end_lr - warmup_init_lr) / warmup_updates
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -