-
Notifications
You must be signed in to change notification settings - Fork 0
/
train_baseline.py
223 lines (172 loc) · 6.65 KB
/
train_baseline.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
import sys
from pathlib import Path
FILE = Path(__file__).resolve()
ROOT = FILE.parents[0] # root directory
if str(ROOT) not in sys.path:
sys.path.append(str(ROOT)) # add ROOT to PATH
import torch
import models
import loaders
import argparse
import os
import time
from torch import optim
from configs.config_util import get_cfg
from utils.time_util import print_time, get_current_time
from sklearn.metrics import classification_report
from misc import draw_acc_and_loss
from misc import print_yml_cfg
from utils.args_util import print_args
from utils.general import update_best_model, init_seeds
from misc import get_loss_fn, get_scheduler
import warnings # ignore warnings
warnings.filterwarnings("ignore")
parser = argparse.ArgumentParser()
parser.add_argument('--data_name', default='cifar-10-lt-ir100')
parser.add_argument('--model_name', default='resnet32')
parser.add_argument('--lr', type=float, default='1e-2')
parser.add_argument('--data_loader_type', type=int, default='0')
parser.add_argument('--epochs', type=int, default='200')
parser.add_argument('--lr_scheduler', type=str, default='cosine')
parser.add_argument('--loss_type', type=str, default='bsl')
parser.add_argument('--fl_gamma', type=float, default=2.0)
parser.add_argument('--gpu_id', type=str, default='0')
def main():
args = parser.parse_args()
print_args(args)
init_seeds()
# get cfg
data_name = args.data_name
model_name = args.model_name
cfg_filename = "one_stage.yml"
cfg = get_cfg(cfg_filename)[data_name]
print_yml_cfg(cfg)
# result path
result_path = os.path.join("./work_dir/baseline",
model_name, data_name,
f"lr{args.lr}", f"{args.lr_scheduler}_lr_scheduler",
f"{args.loss_type}_loss",
get_current_time())
if not os.path.exists(result_path):
os.makedirs(result_path)
print(f"\n[INFO] result will save in:\n{result_path}\n")
# add some cfg
cfg["best_model_path"] = None
cfg["result_path"] = result_path
cfg["best_acc"] = 0
cfg["g_train_loss"] = []
cfg["g_train_acc"] = []
cfg["g_test_loss"] = []
cfg["g_test_acc"] = []
lr = float(args.lr)
momentum = cfg["optimizer"]["momentum"]
weight_decay = float(cfg["optimizer"]["weight_decay"])
epochs = args.epochs
print(f"\n[INFO] total epoch: {epochs}")
# device
os.environ["CUDA_VISIBLE_DEVICES"] = args.gpu_id
device = torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')
print('-' * 42, '\n[Info] use device:', device)
# data loader
if args.data_loader_type == 0:
data_loaders, _ = loaders.load_data(data_name=data_name)
# model
model = models.load_model(
model_name=model_name,
in_channels=cfg["model"]["in_channels"],
num_classes=cfg["model"]["num_classes"]
)
model.to(device)
# loss
loss_fn = get_loss_fn(args, cfg, device)
# optimizer
optimizer = optim.SGD(
params=model.parameters(),
lr=lr,
momentum=momentum,
weight_decay=weight_decay
)
# lr scheduler
scheduler = get_scheduler(args.lr_scheduler, optimizer, epochs)
begin_time = time.time()
for epoch in range(epochs):
epoch_begin_time = time.time()
cur_lr = float(optimizer.state_dict()['param_groups'][0]['lr'])
print(f"\nEpoch {epoch+1}")
print("[INFO] lr is:", cur_lr)
print("-" * 42)
train(data_loaders["train"], model, loss_fn, optimizer, device, cfg)
test(data_loaders["val"], model, loss_fn, optimizer, epoch, device, args, cfg)
scheduler.step()
draw_acc_and_loss(cfg["g_train_loss"], cfg["g_test_loss"], cfg["g_train_acc"], cfg["g_test_acc"], result_path)
print_time(time.time()-epoch_begin_time, epoch=True)
print("Done!")
print_time(time.time()-begin_time)
def train(dataloader, model, loss_fn, optimizer, device, cfg):
train_loss, correct = 0, 0
y_pred_list = []
y_train_list = []
size = len(dataloader.dataset)
num_batches = len(dataloader)
model.train()
for batch, (X, y) in enumerate(dataloader):
y_train_list.extend(y.numpy())
X, y = X.to(device), y.to(device)
with torch.set_grad_enabled(True):
# Compute prediction error
pred = model(X)
loss = loss_fn(pred, y)
train_loss += loss.item()
y_pred_list.extend(pred.argmax(1).cpu().numpy())
correct += (pred.argmax(1) == y).type(torch.float).sum().item()
# Backpropagation
optimizer.zero_grad()
loss.backward()
optimizer.step()
if batch % 10 == 0:
loss, current = loss.item(), batch * len(X)
print(f"train | loss: {loss:>7f} [{current:>5d}/{size:>5d}]", flush=True)
train_loss /= num_batches
correct /= size
cfg["g_train_loss"].append(train_loss)
cfg["g_train_acc"].append(correct)
print("-" * 42)
print(classification_report(y_train_list, y_pred_list, digits=4))
def test(dataloader, model, loss_fn, optimizer, epoch, device, args, cfg):
y_pred_list = []
y_train_list = []
size = len(dataloader.dataset)
num_batches = len(dataloader)
model.eval()
test_loss, correct = 0, 0
for batch, (X, y) in enumerate(dataloader):
y_train_list.extend(y.numpy())
X, y = X.to(device), y.to(device)
with torch.set_grad_enabled(True):
pred = model(X)
loss = loss_fn(pred, y)
test_loss += loss.item()
y_pred_list.extend(pred.argmax(1).cpu().numpy())
correct += (pred.argmax(1) == y).type(torch.float).sum().item()
if batch % 10 == 0:
loss, current = loss.item(), batch * len(X)
print(f"val | loss: {loss:>7f} [{current:>5d}/{size:>5d}]", flush=True)
test_loss /= num_batches
cfg["g_test_loss"].append(test_loss)
correct /= size
cfg["g_test_acc"].append(correct)
if correct > cfg["best_acc"]:
cfg["best_acc"] = correct
print(f"\n[FEAT] Epoch {epoch+1}, update best acc:", correct)
model_name=f"best-model-acc{correct:.4f}.pth"
model_state = {
'epoch': epoch,
'model': model.state_dict(),
'optimizer': optimizer.state_dict(),
'acc': correct,
}
update_best_model(cfg, model_state, model_name)
print(f"\nTest Error: Accuracy: {(100*correct):>0.2f}%, Avg loss: {test_loss:>8f} \n")
print(classification_report(y_train_list, y_pred_list, digits=4))
if __name__ == '__main__':
main()