-
Notifications
You must be signed in to change notification settings - Fork 0
/
simclr_train.py
189 lines (168 loc) · 7.9 KB
/
simclr_train.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
#配置train_loader、test_loader
import torch
import sys
import numpy as np
import os
import yaml
import matplotlib.pyplot as plt
import pandas as pd
from torch.utils.data import DataLoader
import torchvision.transforms as transforms
from torchvision import datasets
import torchvision
from IPython.core.display import JSON
import os
import gdown
import torch
import torch.nn as nn
from pif.influence_functions_new import get_gradient,tracin_get
import time
import tqdm
from torch.autograd import grad
#加载backbone的下载
def get_file_id_by_model(folder_name):
file_id = {'resnet18_100-epochs_stl10': '14_nH2FkyKbt61cieQDiSbBVNP8-gtwgF',
'resnet18_100-epochs_cifar10': '1lc2aoVtrAetGn0PnTkOyFzPCIucOJq7C',
'resnet50_50-epochs_stl10': '1ByTKAUsdm_X7tLcii6oAEl5qFRqRMZSu'}
return file_id.get(folder_name, "Model not found.")
def accuracy(output, target, topk=(1,)):
"""Computes the accuracy over the k top predictions for the specified values of k"""
with torch.no_grad():
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].reshape(-1).float().sum(0, keepdim=True)
res.append(correct_k.mul_(100.0 / batch_size))
return res
def get_cifar10_data_loaders(download, shuffle=False, batch_size=128):
train_dataset = datasets.CIFAR10('./data', train=True, download=download,
transform=transforms.ToTensor())
train_loader = DataLoader(train_dataset, batch_size=batch_size,
num_workers=0, drop_last=False, shuffle=shuffle)
#train_loader加载了50000张图片
test_dataset = datasets.CIFAR10('./data', train=False, download=download,
transform=transforms.ToTensor())
test_loader = DataLoader(test_dataset, batch_size=2*batch_size,
num_workers=0, drop_last=False, shuffle=shuffle)
return train_loader, test_loader
folder_name = 'resnet18_100-epochs_cifar10'
file_id = get_file_id_by_model(folder_name)
print(folder_name, file_id)
print('start download:')
'''
os.system('gdown https://drive.google.com/uc?id={}'.format(file_id))
gdown.download('https://drive.google.com/uc?id={}'.format(file_id),'D:/tracin-main/resnet18_100-epochs_cifar10')
print('download finished.start unzip')
os.system('unzip {}'.format(folder_name))
'''
device = 'cuda' if torch.cuda.is_available() else 'cpu'
print("Using device:", device)
#train_loader, test_loader = get_cifar10_data_loaders(download=True,shuffle=False,batch_size=256)
model = torchvision.models.resnet18(pretrained=False, num_classes=10).to(device)
# define loss function
optimizer = torch.optim.Adam(model.parameters(), lr=3e-4, weight_decay=0.0008)
criterion = torch.nn.CrossEntropyLoss().to(device)
epochs=2
for epoch in range(epochs):
score_list = []
for j in [10,20,30,40,50,60,70,80,90,100,110,120,130,140,150,160,170,180,190]:
#加载相应simclr模型
checkpoint = torch.load('D:/tracin-main/simclr_tracin_pth/tracin_simclr_{}_{}.pth.tar'.format(epoch,j), map_location=device)
state_dict = checkpoint['state_dict']
for k in list(state_dict.keys()):
if k.startswith('backbone.'):
if k.startswith('backbone') and not k.startswith('backbone.fc'):
# remove prefix
state_dict[k[len("backbone."):]] = state_dict[k]
del state_dict[k]
log = model.load_state_dict(state_dict, strict=False)
assert log.missing_keys == ['fc.weight', 'fc.bias']
for name, param in model.named_parameters():
if name not in ['fc.weight', 'fc.bias']:
param.requires_grad = False
parameters = list(filter(lambda p: p.requires_grad, model.parameters()))
assert len(parameters) == 2 # fc.weight, fc.bias
train_loader, test_loader = get_cifar10_data_loaders(download=True,shuffle=False,batch_size=2)
#print(parameters)
print('*****************************************************************************************')
print('this is tracin_batch=128 for simclr_pretrain_batch=256 epoch{} step{}'.format(epoch,j))
print('*****************************************************************************************')
#根据simclr的特点，按照batch来tracin
count_test=0
score_all=0
score_all_mean=0
#df=pd.DataFrame()
score_list=[]
top1_train_accuracy = 0
for counter, (x_batch, y_batch) in enumerate(test_loader):#counter是什么？是一个batch里有多少数据吗？
#print('these model only require parameters')
x_batch = x_batch.to(device)
y_batch = y_batch.to(device)
print('true{}'.format(list(y_batch)))
test_list=[]
for i in y_batch:
test_list.append(int(i))
test_label=test_list
logits = model(x_batch)
#print('true{}'.format(y_batch))
loss = criterion(logits, y_batch)
print('loss is:',loss)
test_loss=(float(loss))
grad_z_test = grad(loss, parameters)
grad_z_test = get_gradient(grad_z_test, model)
#optimizer.zero_grad()
#loss.backward()
#optimizer.step()
top1 = accuracy(logits, y_batch, topk=(1,))
top1_train_accuracy += top1[0]
#print(top1[0])
count_test = count_test+1
count_train=0
score_on_all=0
score_test_batch=[]
batch_label=[]
train_loss=[]
batch_score=[]
for counter, (x_batch, y_batch) in enumerate(train_loader):
#print('true{}'.format(list(y_batch)))
x_batch = x_batch.to(device)
y_batch = y_batch.to(device)
train_list=[]
for i in y_batch:
train_list.append(int(i))
#print('true{}'.format(list(y_batch)))
#print(int(y_batch))
batch_label.append(train_list)
logits = model(x_batch)
loss = criterion(logits, y_batch)
#print('train_loss is:',float(loss))
train_loss.append(float(loss))
grad_z_train = grad(loss, parameters)
grad_z_train = get_gradient(grad_z_train, model)
count_train=count_train+1
score = tracin_get(grad_z_test, grad_z_train)
#print('------tracin_score is------',float(score))
batch_score.append(float(score))
score_test_batch.append(float(score))
score_on_all+=score
#print('test_batch{} get score:{} on train_batch{}'.format(count_test,score,count_train))
score_list.append(score_test_batch)
score_mean=score_on_all/count_train
score_all+=score_on_all
score_all_mean+=score_mean
#print('test_batch{} get all_score:{} and mean_score:{} on train_data on simclr_pretrain_epoch{}_step{}'.format(count_test,score_on_all,score_mean,epoch,j))
df=pd.DataFrame()
#all_label.append(batch_label)
#all_train_loss.append(train_loss)
df['y_batch_label']=batch_label
df['train_loss']=train_loss
df['batch_score']=batch_score
df.to_csv("D:/tracin-main/2batch/tracin_test_label{}-test_loss{}_epoch{}_step{}.csv".format(len(test_label),test_loss,epoch,j), index=False)
print(count_test)
score_tmin=score_all_mean/count_test
print('mean score is {} on simclr_pretrain_epoch{}_step{}'.format(score_tmin,epoch,j))
print('all score is {} on simclr_pretrain_epoch{}_step{}'.format(score_all,epoch,j))