-
Notifications
You must be signed in to change notification settings - Fork 0
/
main.py
136 lines (108 loc) · 4.47 KB
/
main.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
import sys
import argparse
from MAS_utils import *
from main_utils import *
import copy
import torch.utils.data
from random import shuffle
import numpy as np
import torchvision.transforms as transforms
import torchvision.models as models
import torchvision.datasets as datasets
import torch.nn.functional as F
import torch.optim as optim
from torch.autograd import Variable
import torch.nn as nn
import torch
torch.backends.cudnn.benchmark = True
parser = argparse.ArgumentParser(description="Testing")
parser.add_argument("--batch_size", default=16, type=int,
help="The batch size you want to use")
parser.add_argument("--epochs", default=50, type=int,
help="no of epochs you want to run for")
parser.add_argument("--num_freezed", default=2, type=int,
help="No of Layers you want to freeze while extracting features")
parser.add_argument("--lr", default=.008, type=float,
help='The initial Learning Rate you want to set')
parser.add_argument("--lamda", default=.01, type=float,
help="Regularisation Parameter")
parser.add_argument("--use_gpu", default=torch.cuda.is_available(),
type=bool, help="if to use gpu while training or not")
args = parser.parse_args()
batch_size = args.batch_size
epochs = args.epochs
num_frozen = args.num_freezed
lr = args.lr
scheduler_lambda = args.lamda
use_gpu = args.use_gpu
path = os.getcwd()
train_dataloaders = []
test_dataloaders = []
train_datasets = []
test_datasets = []
num_classes = []
path = os.getcwd()
datapath = os.path.join(path, "Data")
transforms = {
'train_transforms': transforms.Compose([
transforms.RandomResizedCrop(224),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
]),
'test_transforms': transforms.Compose([
transforms.Resize(256),
transforms.CenterCrop(224),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
])
}
# split the data into the number tasksyou want to before this step
# for our case these are 4
# 4 folders in Data folder with no as their task number
for ndir in sorted(os.listdir(datapath)):
# objects for the train and test image folder
train_image_folder = datasets.ImageFolder(os.path.join(
datapath, ndir, "train"), transform=transforms['train_transforms'])
test_image_folder = datasets.ImageFolder(os.path.join(
datapath, ndir, "test"), transform=transforms['test_transforms'])
# Loading the data into batches
train_dataloader = torch.utils.data.DataLoader(
train_image_folder, batch_size=batch_size, shuffle=True, num_workers=3)
test_dataloader = torch.utils.data.DataLoader(
test_image_folder, batch_size=batch_size, shuffle=True, num_workers=1)
train_size = len(train_image_folder)
test_size = len(test_image_folder)
num_classes.append(len(train_image_folder.classes))
# appending the datloader that we have created just now for future use
train_dataloaders.append(train_dataloader)
test_dataloaders.append(test_dataloader)
train_datasets.append(train_size)
test_datasets.append(test_size)
no_tasks = len(train_dataloaders)
print(no_tasks)
# initialising the model
model = SharedModel(models.alexnet(pretrained=True))
# model training on the given no of tasks
for task in range(1, no_tasks+1):
print("Training starting on the task #{}".format(task))
trdataloader = train_dataloaders[task-1]
tedataloader = test_dataloaders[task-1]
train_size = train_datasets[task-1]
test_size = test_datasets[task-1]
no_of_classes = num_classes[task-1]
# refer to the model utils
model = model_initialiser(no_of_classes, use_gpu)
# We have the initialised model and reg params inside the model
MAS(model, task, epochs, no_of_classes, lr, scheduler_lambda,
num_frozen, use_gpu, trdataloader, tedataloader, train_size, test_size)
print("The training process on all the tasks are completed")
print("Testing Phase")
# test the sets on the test sets of the tasks
for task in range(1, no_tasks + 1):
print("Testiong the model on task{}".format(task))
dataloader = test_dataloaders[task-1]
dset_size = train_datasets[task-1]
no_of_classes = num_classes[task-1]
forgetting = compute_forgetting(task, dataloader, dset_size, use_gpu)
print("The forgetting undergone on task {} is {}".format(task, forgetting))