File size: 1,710 Bytes
8b54513
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.

# This source code is licensed under the license found in the
# LICENSE file in the root directory of this source tree.

import math

def adjust_learning_rate(optimizer, it, args):
    """Decay the learning rate with half-cycle cosine after warmup"""
    if it < args.warmup_iters: # 1) linear warmup for warmup_iters steps
        lr = args.lr * it / args.warmup_iters
    elif it > args.lr_decay_iters: # 2) if it > lr_decay_iters, return min learning rate
        lr = args.min_lr
    else: # 3) in between, use cosine decay down to min learning rate
        decay_ratio = (it - args.warmup_iters) / (args.lr_decay_iters - args.warmup_iters)
        assert 0 <= decay_ratio <= 1
        coeff = 0.5 * (1.0 + math.cos(math.pi * decay_ratio))  # coeff ranges 0..1
        lr = args.min_lr + (args.lr - args.min_lr) * coeff

    for param_group in optimizer.param_groups:
        if "lr_scale" in param_group:
            param_group["lr"] = lr * param_group["lr_scale"]
        else:
            param_group["lr"] = lr
    return lr


def adjust_learning_rate_epoch(optimizer, epoch, args):
    """Decay the learning rate with half-cycle cosine after warmup"""
    if epoch < args.warmup_epochs:
        lr = args.lr * epoch / args.warmup_epochs
    else:
        lr = args.min_lr + (args.lr - args.min_lr) * 0.5 * \
            (1. + math.cos(math.pi * (epoch - args.warmup_epochs) / (args.epochs - args.warmup_epochs)))
    for param_group in optimizer.param_groups:
        if "lr_scale" in param_group:
            param_group["lr"] = lr * param_group["lr_scale"]
        else:
            param_group["lr"] = lr
    return lr