BBDecoder/testing.py at main · sharjeel1999/BBDecoder · GitHub

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 os

import torch
import torch.nn as nn
import torch.nn.functional as F
import torchvision
from torchvision import transforms
from torchvision.models import resnet50, vit_b_16

from tqdm import tqdm

# Define the CNN architecture
class SimpleCNN(nn.Module):
    def __init__(self, num_classes=10):
        super(SimpleCNN, self).__init__()

        self.conv1 = nn.Conv2d(in_channels=3, out_channels=32, kernel_size=3, stride=1, padding=1)
        self.conv2 = nn.Conv2d(in_channels=32, out_channels=64, kernel_size=3, stride=1, padding=1)

        self.fc1 = nn.Linear(4096, 128)
        self.fc2 = nn.Linear(128, num_classes)

        self.pool = nn.MaxPool2d(kernel_size=2, stride=2)

    def forward(self, x):
        x = self.pool(F.relu(self.conv1(x)))
        x = self.pool(F.relu(self.conv2(x)))

        x = x.view(x.size(0), -1)
        # print('flat shape: ', x.shape)

        x = F.relu(self.fc1(x))
        x = self.fc2(x)
        return x

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

# model = SimpleCNN(num_classes=10)
model = resnet50().to(device)
# model = vit_b_16()

from BBDecoder import Master_analyzer
from torch.optim import Adam

optimizer = Adam(model.parameters(), lr=0.001)

save_path = 'save_folder'
wrapped_model = Master_analyzer(model, 'O:\\PCodes\\black_box\\', modular = False, depth = 3)


# print(wrapped_model)

# from torchview import draw_graph
# from BBDecoder.visualizer import draw_graph

# input_size = (1, 3, 32, 32)
# model_graph = draw_graph(wrapped_model, input_size = input_size, expand_nested=True, hide_module_functions=True)
# model_graph.visual_graph.render("model_visualization_with_index", format = "png")


# transform = transforms.Compose([
#     transforms.ToTensor(),
#     transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
# ])

# trainset = torchvision.datasets.CIFAR10(
#     root='DATASETS',
#     train=True,
#     download=True,
#     transform=transform
# )
# trainloader = torch.utils.data.DataLoader(trainset, batch_size=32, shuffle=True)

# testset = torchvision.datasets.CIFAR10(
#     root='DATASETS',
#     train=False,
#     download=True,
#     transform=transform
# )

# subset_indices = list(range(6))  # 96, Indices of the first 10 samples
# subset = torch.utils.data.Subset(testset, subset_indices)

# testloader = torch.utils.data.DataLoader(testset, batch_size=32, shuffle=False)

# classes = ('plane', 'car', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck')


# wrapped_model.initialize_analyser(layer_inds = [0, 1, 2, 3, 4],
#                                 grad_flag = True,
#                                 grad_hist_flag = True,
#                                 track_grads = True,
#                                 function_flag = False)


## Testing using a dummy dataset
# Epochs = 1
# for epoch in range(Epochs):

#     losses = []
#     for data in tqdm(testloader):
#         inputs, labels = data
#         inputs, labels = inputs.to(device), labels.to(device)

#         outputs = wrapped_model.forward_propagation(inputs)
#         loss = F.cross_entropy(outputs, labels)

#         optimizer.zero_grad()
#         wrapped_model.backward_propagation(loss, collect_grads = True, layers = [0, 1, 2, 3, 4, 5, 6])
#         optimizer.step()
#         losses.append(loss.item())

#     print('Epoch: ', epoch)
#     print('Average Loss: ', sum(losses)/len(losses))
#     save_folder = 'results'
#     if not os.path.exists(save_folder):
#         os.makedirs(save_folder)
#     wrapped_model.save_collected_grads(save_folder = save_folder, ep = epoch)


# wrapped_model.visualize_weight_hist('O:\\PCodes\\black_box\\save_folder\\Weights')
# # wrapped_model.threshold_pruning(0.01)
# weights_folder = os.path.join("results", "weights")
# if not os.path.exists(weights_folder):
#     os.makedirs(weights_folder)
# wrapped_model.visualize_weight_hist(path=weights_folder, layer_inds = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9])

# wrapped_model.save_tracked_data(save_folder = 'O:\\PCodes\\black_box\\save_folder')


# for batch_idx, (inputs, targets) in enumerate(testloader):
#     print(f"Batch {batch_idx}:")
#     print("Inputs:", inputs.shape)
#     print("Targets:", targets.shape)
#     wrapped_model.get_sim(torch.unsqueeze(inputs.cuda()[0, :, :, :], dim=0), layers = [0, 1, 2, 3, 4, 5, 6], dim = 1)
#     break