File size: 9,827 Bytes
07f408f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
import math
import torch
from torch.optim.optimizer import Optimizer


class AdaBelief(Optimizer):
    r"""Implements AdaBelief algorithm. Modified from Adam in PyTorch

    Arguments:
        params (iterable): iterable of parameters to optimize or dicts defining
            parameter groups
        lr (float, optional): learning rate (default: 1e-3)
        betas (Tuple[float, float], optional): coefficients used for computing
            running averages of gradient and its square (default: (0.9, 0.999))
        eps (float, optional): term added to the denominator to improve
            numerical stability (default: 1e-16)
        weight_decay (float, optional): weight decay (L2 penalty) (default: 0)
        amsgrad (boolean, optional): whether to use the AMSGrad variant of this
            algorithm from the paper `On the Convergence of Adam and Beyond`_
            (default: False)
        decoupled_decay (boolean, optional): (default: True) If set as True, then
            the optimizer uses decoupled weight decay as in AdamW
        fixed_decay (boolean, optional): (default: False) This is used when weight_decouple
            is set as True.
            When fixed_decay == True, the weight decay is performed as
            $W_{new} = W_{old} - W_{old} \times decay$.
            When fixed_decay == False, the weight decay is performed as
            $W_{new} = W_{old} - W_{old} \times decay \times lr$. Note that in this case, the
            weight decay ratio decreases with learning rate (lr).
        rectify (boolean, optional): (default: True) If set as True, then perform the rectified
            update similar to RAdam
        degenerated_to_sgd (boolean, optional) (default:True) If set as True, then perform SGD update
            when variance of gradient is high
    reference: AdaBelief Optimizer, adapting stepsizes by the belief in observed gradients, NeurIPS 2020

    For a complete table of recommended hyperparameters, see https://github.com/juntang-zhuang/Adabelief-Optimizer'
    For example train/args for EfficientNet see these gists
      - link to train_scipt: https://gist.github.com/juntang-zhuang/0a501dd51c02278d952cf159bc233037
      - link to args.yaml: https://gist.github.com/juntang-zhuang/517ce3c27022b908bb93f78e4f786dc3
    """

    def __init__(
            self, params, lr=1e-3, betas=(0.9, 0.999), eps=1e-16, weight_decay=0, amsgrad=False,
            decoupled_decay=True, fixed_decay=False, rectify=True, degenerated_to_sgd=True):

        if not 0.0 <= lr:
            raise ValueError("Invalid learning rate: {}".format(lr))
        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]))

        if isinstance(params, (list, tuple)) and len(params) > 0 and isinstance(params[0], dict):
            for param in params:
                if 'betas' in param and (param['betas'][0] != betas[0] or param['betas'][1] != betas[1]):
                    param['buffer'] = [[None, None, None] for _ in range(10)]

        defaults = dict(
            lr=lr, betas=betas, eps=eps, weight_decay=weight_decay, amsgrad=amsgrad,
            degenerated_to_sgd=degenerated_to_sgd, decoupled_decay=decoupled_decay, rectify=rectify,
            fixed_decay=fixed_decay, buffer=[[None, None, None] for _ in range(10)])
        super(AdaBelief, self).__init__(params, defaults)

    def __setstate__(self, state):
        super(AdaBelief, self).__setstate__(state)
        for group in self.param_groups:
            group.setdefault('amsgrad', False)

    @torch.no_grad()
    def reset(self):
        for group in self.param_groups:
            for p in group['params']:
                state = self.state[p]
                amsgrad = group['amsgrad']

                # State initialization
                state['step'] = 0
                # Exponential moving average of gradient values
                state['exp_avg'] = torch.zeros_like(p)

                # Exponential moving average of squared gradient values
                state['exp_avg_var'] = torch.zeros_like(p)
                if amsgrad:
                    # Maintains max of all exp. moving avg. of sq. grad. values
                    state['max_exp_avg_var'] = torch.zeros_like(p)

    @torch.no_grad()
    def step(self, closure=None):
        """Performs a single optimization step.
        Arguments:
            closure (callable, optional): A closure that reevaluates the model
                and returns the loss.
        """
        loss = None
        if closure is not None:
            with torch.enable_grad():
                loss = closure()

        for group in self.param_groups:
            for p in group['params']:
                if p.grad is None:
                    continue
                grad = p.grad
                if grad.dtype in {torch.float16, torch.bfloat16}:
                    grad = grad.float()
                if grad.is_sparse:
                    raise RuntimeError(
                        'AdaBelief does not support sparse gradients, please consider SparseAdam instead')

                p_fp32 = p
                if p.dtype in {torch.float16, torch.bfloat16}:
                    p_fp32 = p_fp32.float()

                amsgrad = group['amsgrad']
                beta1, beta2 = group['betas']
                state = self.state[p]
                # State initialization
                if len(state) == 0:
                    state['step'] = 0
                    # Exponential moving average of gradient values
                    state['exp_avg'] = torch.zeros_like(p_fp32)
                    # Exponential moving average of squared gradient values
                    state['exp_avg_var'] = torch.zeros_like(p_fp32)
                    if amsgrad:
                        # Maintains max of all exp. moving avg. of sq. grad. values
                        state['max_exp_avg_var'] = torch.zeros_like(p_fp32)
                
                # perform weight decay, check if decoupled weight decay
                if group['decoupled_decay']:
                    if not group['fixed_decay']:
                        p_fp32.mul_(1.0 - group['lr'] * group['weight_decay'])
                    else:
                        p_fp32.mul_(1.0 - group['weight_decay'])
                else:
                    if group['weight_decay'] != 0:
                        grad.add_(p_fp32, alpha=group['weight_decay'])

                # get current state variable
                exp_avg, exp_avg_var = state['exp_avg'], state['exp_avg_var']

                state['step'] += 1
                bias_correction1 = 1 - beta1 ** state['step']
                bias_correction2 = 1 - beta2 ** state['step']

                # Update first and second moment running average
                exp_avg.mul_(beta1).add_(grad, alpha=1 - beta1)
                grad_residual = grad - exp_avg
                exp_avg_var.mul_(beta2).addcmul_(grad_residual, grad_residual, value=1 - beta2)

                if amsgrad:
                    max_exp_avg_var = state['max_exp_avg_var']
                    # Maintains the maximum of all 2nd moment running avg. till now
                    torch.max(max_exp_avg_var, exp_avg_var.add_(group['eps']), out=max_exp_avg_var)

                    # Use the max. for normalizing running avg. of gradient
                    denom = (max_exp_avg_var.sqrt() / math.sqrt(bias_correction2)).add_(group['eps'])
                else:
                    denom = (exp_avg_var.add_(group['eps']).sqrt() / math.sqrt(bias_correction2)).add_(group['eps'])
                
                # update
                if not group['rectify']:
                    # Default update
                    step_size = group['lr'] / bias_correction1
                    p_fp32.addcdiv_(exp_avg, denom, value=-step_size)
                else:
                    # Rectified update, forked from RAdam
                    buffered = group['buffer'][int(state['step'] % 10)]
                    if state['step'] == buffered[0]:
                        num_sma, step_size = buffered[1], buffered[2]
                    else:
                        buffered[0] = state['step']
                        beta2_t = beta2 ** state['step']
                        num_sma_max = 2 / (1 - beta2) - 1
                        num_sma = num_sma_max - 2 * state['step'] * beta2_t / (1 - beta2_t)
                        buffered[1] = num_sma

                        # more conservative since it's an approximated value
                        if num_sma >= 5:
                            step_size = math.sqrt(
                                (1 - beta2_t) *
                                (num_sma - 4) / (num_sma_max - 4) *
                                (num_sma - 2) / num_sma *
                                num_sma_max / (num_sma_max - 2)) / (1 - beta1 ** state['step'])
                        elif group['degenerated_to_sgd']:
                            step_size = 1.0 / (1 - beta1 ** state['step'])
                        else:
                            step_size = -1
                        buffered[2] = step_size

                    if num_sma >= 5:
                        denom = exp_avg_var.sqrt().add_(group['eps'])
                        p_fp32.addcdiv_(exp_avg, denom, value=-step_size * group['lr'])
                    elif step_size > 0:
                        p_fp32.add_(exp_avg, alpha=-step_size * group['lr'])
                
                if p.dtype in {torch.float16, torch.bfloat16}:
                    p.copy_(p_fp32)

        return loss