-
Notifications
You must be signed in to change notification settings - Fork 0
/
metrics.py
executable file
·245 lines (216 loc) · 8.98 KB
/
metrics.py
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
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
# Copyright (c) Aishwarya Kamath & Nicolas Carion. Licensed under the Apache License 2.0. All Rights Reserved
"""
Various utilities related to track and report metrics
"""
import datetime
import time
from collections import defaultdict, deque
import torch
from bisect import bisect_right
from torch.utils.tensorboard import SummaryWriter
class SmoothedValue:
"""Track a series of values and provide access to smoothed values over a
window or the global series average.
"""
def __init__(self, window_size=20, fmt=None):
if fmt is None:
fmt = "avg: {avg:.4f} med: {median:.4f} ({global_avg:.4f})"
self.deque = deque(maxlen=window_size)
self.total = 0.0
self.count = 0
self.fmt = fmt
def update(self, value, num=1):
self.deque.append(value)
self.count += num
self.total += value * num
@property
def median(self):
d = torch.tensor(list(self.deque))
return d.median().item()
@property
def avg(self):
d = torch.tensor(list(self.deque), dtype=torch.float32)
return d.mean().item()
@property
def global_avg(self):
return self.total / self.count
@property
def max(self):
return max(self.deque)
@property
def value(self):
return self.deque[-1]
def __str__(self):
return self.fmt.format(
median=self.median, avg=self.avg, global_avg=self.global_avg, max=self.max, value=self.value
)
class MetricLogger(object):
def __init__(self, delimiter="\t", path=None):
self.meters = defaultdict(SmoothedValue)
self.delimiter = delimiter
self.writer = None
if path:
self.writer = SummaryWriter(path)
def updateTensorboard(self, epoch, mode='training'):
for name, meter in self.meters.items():
if 'lr' not in name and 'unscaled' not in name:
self.writer.add_scalar(mode + '/' + name + ' avg', meter.avg, epoch)
self.writer.add_scalar(mode + '/' + name + ' global avg', meter.global_avg, epoch)
self.writer.add_scalar(mode + '/' + name + ' med', meter.median, epoch)
elif 'lr' in name:
self.writer.add_scalar(mode + '/' + name, meter.value, epoch)
def closeWriter(self):
self.writer.close()
def update(self, **kwargs):
for k, v in kwargs.items():
if isinstance(v, torch.Tensor):
v = v.item()
assert isinstance(v, (float, int))
self.meters[k].update(v)
def __getattr__(self, attr):
if attr in self.meters:
return self.meters[attr]
if attr in self.__dict__:
return self.__dict__[attr]
raise AttributeError("'{}' object has no attribute '{}'".format(type(self).__name__, attr))
def __str__(self):
loss_str = []
for name, meter in self.meters.items():
loss_str.append("{}: {}".format(name, str(meter)))
return self.delimiter.join(loss_str)
def add_meter(self, name, meter):
self.meters[name] = meter
def log_every(self, iterable, print_freq, header=None):
i = 0
if not header:
header = ""
start_time = time.time()
end = time.time()
iter_time = SmoothedValue(fmt="{avg:.4f}")
data_time = SmoothedValue(fmt="{avg:.4f}")
space_fmt = ":" + str(len(str(len(iterable)))) + "d"
if torch.cuda.is_available():
log_msg = self.delimiter.join(
[
header,
"[{0" + space_fmt + "}/{1}]",
"eta: {eta}",
"{meters}",
"time: {time}",
"data: {data}",
"max mem: {memory:.0f}",
]
)
else:
log_msg = self.delimiter.join(
[header, "[{0" + space_fmt + "}/{1}]", "eta: {eta}", "{meters}", "time: {time}", "data: {data}"]
)
MB = 1024.0 * 1024.0
for obj in iterable:
data_time.update(time.time() - end)
yield obj
iter_time.update(time.time() - end)
if i % print_freq == 0 or i == len(iterable) - 1:
eta_seconds = iter_time.global_avg * (len(iterable) - i)
eta_string = str(datetime.timedelta(seconds=int(eta_seconds)))
if torch.cuda.is_available():
print(
log_msg.format(
i,
len(iterable),
eta=eta_string,
meters=str(self),
time=str(iter_time),
data=str(data_time),
memory=torch.cuda.max_memory_allocated() / MB,
)
)
else:
print(
log_msg.format(
i, len(iterable), eta=eta_string, meters=str(self), time=str(iter_time), data=str(data_time)
)
)
i += 1
end = time.time()
total_time = time.time() - start_time
total_time_str = str(datetime.timedelta(seconds=int(total_time)))
print("{} Total time: {} ({:.4f} s / it)".format(header, total_time_str, total_time / len(iterable)))
@torch.no_grad()
def accuracy(output, target, topk=(1,)):
"""Computes the precision@k for the specified values of k"""
if target.numel() == 0:
return [torch.zeros([], device=output.device)]
maxk = max(topk)
batch_size = target.size(0)
_, pred = output.topk(maxk, 1, True, True)
pred = pred.t()
correct = pred.eq(target.view(1, -1).expand_as(pred))
res = []
for k in topk:
correct_k = correct[:k].view(-1).float().sum(0)
res.append(correct_k.mul_(100.0 / batch_size))
return res
def adjust_learning_rate(
optimizer,
epoch: int,
curr_step: int,
num_training_steps: int,
args,
):
"""Adjust the lr according to the schedule.
Args:
Optimizer: torch optimizer to update.
epoch(int): number of the current epoch.
curr_step(int): number of optimization step taken so far.
num_training_step(int): total number of optimization steps.
args: additional training dependent args:
- lr_drop(int): number of epochs before dropping the learning rate.
- fraction_warmup_steps(float) fraction of steps over which the lr will be increased to its peak.
- lr(float): base learning rate
- lr_backbone(float): learning rate of the backbone
- text_encoder_backbone(float): learning rate of the text encoder
- schedule(str): the requested learning rate schedule:
"step": all lrs divided by 10 after lr_drop epochs
"multistep": divided by 2 after lr_drop epochs, then by 2 after every 50 epochs
"linear_with_warmup": same as "step" for backbone + transformer, but for the text encoder, linearly
increase for a fraction of the training, then linearly decrease back to 0.
"all_linear_with_warmup": same as "linear_with_warmup" for all learning rates involved.
"""
num_warmup_steps: int = round(args.fraction_warmup_steps * num_training_steps)
if args.schedule == "step":
gamma = 0.1 ** (epoch // args.lr_drop)
text_encoder_gamma = gamma
elif args.schedule == "multistep":
milestones = list(range(args.lr_drop, args.epochs, 50))
gamma = 0.5 ** bisect_right(milestones, epoch)
text_encoder_gamma = gamma
elif args.schedule == "linear_with_warmup":
gamma = 0.1 ** (epoch // args.lr_drop)
if curr_step < num_warmup_steps:
text_encoder_gamma = float(curr_step) / float(max(1, num_warmup_steps))
else:
text_encoder_gamma = max(
0.0,
float(num_training_steps - curr_step) / float(max(1, num_training_steps - num_warmup_steps)),
)
elif args.schedule == "all_linear_with_warmup":
if curr_step < num_warmup_steps:
text_encoder_gamma = float(curr_step) / float(max(1, num_warmup_steps))
else:
text_encoder_gamma = max(
0.0,
float(num_training_steps - curr_step) / float(max(1, num_training_steps - num_warmup_steps)),
)
gamma = text_encoder_gamma
elif args.schedule == "lambda":
x = max(0, epoch-args.warmup_steps)
gamma = max(0.75 ** (0.05*x), args.min_lr)
text_encoder_gamma = gamma
else:
raise NotImplementedError
base_lrs = [args.lr, args.lr_backbone, args.text_encoder_lr]
gammas = [gamma, gamma, text_encoder_gamma]
assert len(optimizer.param_groups) == len(base_lrs)
for param_group, lr, gamma_group in zip(optimizer.param_groups, base_lrs, gammas):
param_group["lr"] = lr * gamma_group