-
Notifications
You must be signed in to change notification settings - Fork 0
/
finetune.py
192 lines (167 loc) · 6.38 KB
/
finetune.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
# %%
import pickle
import torchvision
import torch
import torch.nn as nn
import torch.optim as optim
from torch.optim import lr_scheduler
import numpy as np
from torch.utils.data import SubsetRandomSampler
from torchvision import datasets, transforms
import time
import copy
import math
from torchvision.models.vgg import model_urls
# %% Normalise data
mean = torch.tensor([0.4631, 0.4483, 0.3237])
std = torch.tensor([0.2830, 0.2650, 0.2754])
transform_norm = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize(mean, std)
])
train_data = datasets.ImageFolder('/home/m/repo/pythonProject/zemiacko/dle/butterflies', transform_norm)
# %% Split dataset
VAL_SPLIT = 1/4
BATCH_SIZE = 32
train_size = len(train_data)
indices = list(range(train_size))
np.random.shuffle(indices)
split = int(np.floor((VAL_SPLIT) * train_size))
train_idx, valid_idx = indices[split:], indices[:split]
train_size = len(train_idx)
val_size = len(valid_idx)
train_set = torch.utils.data.Subset(train_data, train_idx)
val_set = torch.utils.data.Subset(train_data, valid_idx)
train_loader = torch.utils.data.DataLoader(train_set, batch_size=BATCH_SIZE, shuffle=False, num_workers=0)
val_loader = torch.utils.data.DataLoader(val_set, batch_size=BATCH_SIZE, shuffle=False, num_workers=0)
# %% Train
def train_model(model, loss_fce, optimizer, train_loader, val_loader, test_loader, device,num_epochs=20):
val_losses = np.zeros(shape=(num_epochs,), dtype=float)
val_accuracies = np.zeros(shape=(num_epochs,), dtype=float)
train_losses = np.zeros(shape=(num_epochs,), dtype=float)
train_accuracies = np.zeros(shape=(num_epochs,), dtype=float)
model = model.to(device)
best_model_wts = copy.deepcopy(model.state_dict())
best_acc = 0.0
best_loss = math.inf
for epoch in range(num_epochs):
print('Epoch {}/{}'.format(epoch, num_epochs - 1))
l = 0.0
c = 0.0
for x, target in train_loader:
x = x.to(device)
target = target.to(device)
# zero the parameter gradients
optimizer.zero_grad()
# forward
with torch.set_grad_enabled(True):
outputs = model(x)
energies, preds = torch.max(outputs, 1)
loss = loss_fce(outputs, target)
loss.backward()
optimizer.step()
# statistics
l += loss.item() * x.size(0)
c += torch.sum(preds == target.data).item()
optimizer.param_groups[0]['lr'] += (0.001)
epoch_loss = l / len(train_loader.dataset)
epoch_acc = c / len(train_loader.dataset)
print('Train Loss: {:.4f} Acc: {:.4f}'.format(epoch_loss, epoch_acc))
train_losses[epoch] = epoch_loss
train_accuracies[epoch] = epoch_acc
# validation
with torch.set_grad_enabled(False):
l = 0
c = 0
for x, target in val_loader:
x = x.to(device)
target = target.to(device)
# zero the parameter gradients
optimizer.zero_grad()
outputs = model(x)
energies, preds = torch.max(outputs, 1)
loss = loss_fce(outputs, target)
l += loss.item() * x.size(0)
c += torch.sum(preds == target.data).item()
epoch_loss = l / len(val_loader.dataset)
epoch_acc = c / len(val_loader.dataset)
print(f'Val Epoch: {epoch} mean loss: {epoch_loss}, mean acc: {epoch_acc}')
val_losses[epoch] = epoch_loss
val_accuracies[epoch] = epoch_acc
# deep copy the model
if epoch_acc > best_acc:
best_acc = epoch_acc
best_model_wts = copy.deepcopy(model.state_dict())
elif epoch_acc == best_acc and epoch_loss < best_loss:
best_loss = epoch_loss
best_model_wts = copy.deepcopy(model.state_dict())
print('Best val Acc: {:4f}'.format(best_acc))
results = {'train_acc:': train_accuracies, 'train_loss': train_losses, 'val_acc': val_accuracies,
'val_loss': val_losses}
# load best model weights
model.load_state_dict(best_model_wts)
model.eval()
# l = 0.0
# c = 0
# # Iterate over data.
# model.to(device)
# with torch.set_grad_enabled(False):
# for x, target in test_loader:
# x = x.to(device)
# target = target.to(device)
# # zero the parameter gradients
# optimizer.zero_grad()
# outputs = model(x)
# energies, preds = torch.max(outputs, 1)
# loss = loss_fce(outputs, target)
#
# l += loss.item() * x.size(0)
# c += torch.sum(preds == target.data).item()
#
# loss = l / len(test_loader.dataset)
# acc = c / len(test_loader.dataset)
#
# print('Model {} Loss: {:.4f} Acc: {:.4f}'.format(
# 'Test', loss, acc))
return model, results
# %% Test
test_data = datasets.ImageFolder('butterflies/train', transform_norm)
test_loader = torch.utils.data.DataLoader(test_data, batch_size=BATCH_SIZE, shuffle=False, num_workers=0)
# %% Load pretrained model
for k in model_urls.keys():
model_urls[k] = model_urls[k].replace('https://', 'http://')
model = torchvision.models.vgg11(pretrained=True)
model.eval()
#Unfreeze all parameters
for param in model.parameters():
param.requires_grad = True
def init_weights(m):
if type(m) == nn.Linear:
torch.nn.init.xavier_uniform(m.weight)
m.bias.data.fill_(0.01)
class_names = train_data.classes
n_features = model.classifier[-1].in_features
n_classes = len(class_names)
fc = nn.Linear(n_features, n_classes)
init_weights(fc)
model.classifier[-1] = fc
model.eval()
# %%
lr = 0.00001
momentum = 0.9
step_size = 5
gamma = 0.1
loss = nn.CrossEntropyLoss()
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
optimizer = optim.SGD(model.parameters(), lr=lr, momentum=momentum)
# Decay LR by a factor of 0.1 every 10 epochs
#scheduler = lr_scheduler.StepLR(optimizer, step_size=step_size, gamma=gamma)
# %%
best_model, results = train_model(model, loss, optimizer, train_loader, val_loader, test_loader, device)
best_model_save_name = f'Finetuned'
torch.save(best_model.state_dict(), best_model_save_name)
with open('history_finetune.pkl', 'wb') as f:
pickle.dump(results, f)
# %%
# Test model
# test(best_model, test_loader, loss)