Python pytorch_pretrained_bert.optimization.warmup_linear() Examples

The following are 9 code examples of pytorch_pretrained_bert.optimization.warmup_linear(). You can vote up the ones you like or vote down the ones you don't like, and go to the original project or source file by following the links above each example. You may also want to check out all available functions/classes of the module pytorch_pretrained_bert.optimization , or try the search function .
Example #1
Source File: bert_optim.py    From gobbli with Apache License 2.0 6 votes vote down vote up 
def __init__(self, params, lr, warmup=-1, t_total=-1, schedule='warmup_linear',
                 betas=(0.9, 0.999), eps=1e-6, weight_decay_rate=0.01,
                 max_grad_norm=1.0):
        if not lr >= 0.0:
            raise ValueError("Invalid learning rate: {} - should be >= 0.0".format(lr))
        if not 0.0 <= warmup < 1.0 and not warmup == -1:
            raise ValueError("Invalid warmup: {} - should be in [0.0, 1.0[ or -1".format(warmup))
        if not 0.0 <= eps:
            raise ValueError("Invalid epsilon value: {}".format(eps))
        if not 0.0 <= betas[0] < 1.0:
            raise ValueError("Invalid beta parameter at index 0: {}".format(betas[0]))
        if not 0.0 <= betas[1] < 1.0:
            raise ValueError("Invalid beta parameter at index 1: {}".format(betas[1]))
        defaults = dict(lr=lr, schedule=schedule, warmup=warmup, t_total=t_total,
                        betas=betas, eps=eps, weight_decay_rate=weight_decay_rate,
                        max_grad_norm=max_grad_norm)
        super(Adamax, self).__init__(params, defaults)
Example #2
Source File: bert_optim.py    From mt-dnn with MIT License 6 votes vote down vote up 
def __init__(self, params, lr, warmup=-1, t_total=-1, schedule='warmup_linear',
                 betas=(0.9, 0.999), eps=1e-6, weight_decay=0.01,
                 max_grad_norm=1.0):
        if not lr >= 0.0:
            raise ValueError("Invalid learning rate: {} - should be >= 0.0".format(lr))
        if not 0.0 <= warmup < 1.0 and not warmup == -1:
            raise ValueError("Invalid warmup: {} - should be in [0.0, 1.0[ or -1".format(warmup))
        if not 0.0 <= eps:
            raise ValueError("Invalid epsilon value: {}".format(eps))
        if not 0.0 <= betas[0] < 1.0:
            raise ValueError("Invalid beta parameter at index 0: {}".format(betas[0]))
        if not 0.0 <= betas[1] < 1.0:
            raise ValueError("Invalid beta parameter at index 1: {}".format(betas[1]))
        defaults = dict(lr=lr, schedule=schedule, warmup=warmup, t_total=t_total,
                        betas=betas, eps=eps, weight_decay=weight_decay,
                        max_grad_norm=max_grad_norm)
        super(Adamax, self).__init__(params, defaults)
Example #3
Source File: bert_optim.py    From mt-dnn with MIT License 6 votes vote down vote up 
def __init__(self, params, lr, warmup=-1, t_total=-1, schedule='warmup_linear',
                 betas=(0.9, 0.999), eps=1e-6, weight_decay=0.001,
                 max_grad_norm=1.0):
        if not lr >= 0.0:
            raise ValueError("Invalid learning rate: {} - should be >= 0.0".format(lr))
        if not 0.0 <= warmup < 1.0 and not warmup == -1:
            raise ValueError("Invalid warmup: {} - should be in [0.0, 1.0[ or -1".format(warmup))
        if not 0.0 <= eps:
            raise ValueError("Invalid epsilon value: {}".format(eps))
        if not 0.0 <= betas[0] < 1.0:
            raise ValueError("Invalid beta parameter at index 0: {}".format(betas[0]))
        if not 0.0 <= betas[1] < 1.0:
            raise ValueError("Invalid beta parameter at index 1: {}".format(betas[1]))
        defaults = dict(lr=lr, schedule=schedule, warmup=warmup, t_total=t_total,
                        betas=betas, eps=eps, weight_decay=weight_decay,
                        max_grad_norm=max_grad_norm)
        self.buffer = [[None, None, None] for ind in range(10)]
        super(RAdam, self).__init__(params, defaults)
Example #4
Source File: bert_optim.py    From gobbli with Apache License 2.0 5 votes vote down vote up 
def schedule_func(sch):
    try:
        f = eval(sch)
    except:
        f = warmup_linear
    return f
Example #5
Source File: NER_BERT_CRF.py    From NER-BERT-CRF with MIT License 5 votes vote down vote up 
def warmup_linear(x, warmup=0.002):
    if x < warmup:
        return x/warmup
    return 1.0 - x
Example #6
Source File: bert_optim.py    From MT-DNN with MIT License 5 votes vote down vote up 
def schedule_func(sch):
    try:
        f = eval(sch)
    except:
        f = warmup_linear
    return f
Example #7
Source File: bert_optim.py    From MT-DNN with MIT License 5 votes vote down vote up 
def __init__(
        self,
        params,
        lr,
        warmup=-1,
        t_total=-1,
        schedule="warmup_linear",
        betas=(0.9, 0.999),
        eps=1e-6,
        weight_decay=0.01,
        max_grad_norm=1.0,
    ):
        if not lr >= 0.0:
            raise ValueError("Invalid learning rate: {} - should be >= 0.0".format(lr))
        if not 0.0 <= warmup < 1.0 and not warmup == -1:
            raise ValueError("Invalid warmup: {} - should be in [0.0, 1.0[ or -1".format(warmup))
        if not 0.0 <= eps:
            raise ValueError("Invalid epsilon value: {}".format(eps))
        if not 0.0 <= betas[0] < 1.0:
            raise ValueError("Invalid beta parameter at index 0: {}".format(betas[0]))
        if not 0.0 <= betas[1] < 1.0:
            raise ValueError("Invalid beta parameter at index 1: {}".format(betas[1]))
        defaults = dict(
            lr=lr,
            schedule=schedule,
            warmup=warmup,
            t_total=t_total,
            betas=betas,
            eps=eps,
            weight_decay=weight_decay,
            max_grad_norm=max_grad_norm,
        )
        super(Adamax, self).__init__(params, defaults)
Example #8
Source File: bert_optim.py    From MT-DNN with MIT License 5 votes vote down vote up 
def __init__(
        self,
        params,
        lr,
        warmup=-1,
        t_total=-1,
        schedule="warmup_linear",
        betas=(0.9, 0.999),
        eps=1e-6,
        weight_decay=0.001,
        max_grad_norm=1.0,
    ):
        if not lr >= 0.0:
            raise ValueError("Invalid learning rate: {} - should be >= 0.0".format(lr))
        if not 0.0 <= warmup < 1.0 and not warmup == -1:
            raise ValueError("Invalid warmup: {} - should be in [0.0, 1.0[ or -1".format(warmup))
        if not 0.0 <= eps:
            raise ValueError("Invalid epsilon value: {}".format(eps))
        if not 0.0 <= betas[0] < 1.0:
            raise ValueError("Invalid beta parameter at index 0: {}".format(betas[0]))
        if not 0.0 <= betas[1] < 1.0:
            raise ValueError("Invalid beta parameter at index 1: {}".format(betas[1]))
        defaults = dict(
            lr=lr,
            schedule=schedule,
            warmup=warmup,
            t_total=t_total,
            betas=betas,
            eps=eps,
            weight_decay=weight_decay,
            max_grad_norm=max_grad_norm,
        )
        self.buffer = [[None, None, None] for ind in range(10)]
        super(RAdam, self).__init__(params, defaults)
Example #9
Source File: bert_optim.py    From mt-dnn with MIT License 5 votes vote down vote up 
def schedule_func(sch):
    try:
        f = eval(sch)
    except:
        f = warmup_linear
    return f