TensorMONK/ImageNet.py at master · elucideye/TensorMONK · 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
""" tensorMONK's :: ImageNet                                                 """

from __future__ import print_function, division
import sys
import timeit
import argparse
import numpy as np
import torch
from torch.autograd import Variable
from core import *
import torch.optim as neural_optimizer
# ============================================================================ #


def train():
    args = parse_args()
    tensor_size = (1, 3, 224, 224)
    file_name = "./models/" + args.Architecture.lower()
    embedding_net, embedding_net_kwargs = NeuralArchitectures.Models("resnext50")

    train_loader, n_labels = NeuralEssentials.FolderITTR(args.trainDataPath, args.BSZ, tensor_size, args.cpus,
                                                         functions=[], random_flip=True)
    test_loader, n_labels = NeuralEssentials.FolderITTR(args.testDataPath, args.BSZ, tensor_size, args.cpus,
                                                        functions=[], random_flip=False)

    Model = NeuralEssentials.MakeCNN(file_name, tensor_size, n_labels,
                                     embedding_net=embedding_net,
                                     embedding_net_kwargs=embedding_net_kwargs,
                                     loss_net=NeuralLayers.CategoricalLoss,
                                     loss_net_kwargs={"type": args.loss_type, "distance": args.loss_distance},
                                     default_gpu=args.default_gpu, gpus=args.gpus,
                                     ignore_trained=args.ignore_trained)
    params = list(Model.netEmbedding.parameters()) + list(Model.netLoss.parameters())
    if args.optimizer.lower() == "adam":
        optimizer = neural_optimizer.Adam(params)
    elif args.optimizer.lower() == "sgd":
        optimizer = neural_optimizer.SGD(params, lr=args.learningRate)
    else:
        raise NotImplementedError

    # Usual training
    for _ in range(args.Epochs):
        timer = timeit.default_timer()
        Model.netEmbedding.train()
        Model.netLoss.train()
        max_i = 1000
        for i, (tensor, targets) in enumerate(train_loader):
            Model.meterIterations += 1

            # forward pass and parameter update
            Model.netEmbedding.zero_grad()
            Model.netLoss.zero_grad()
            features = Model.netEmbedding(Variable(tensor))
            loss, (top1, top5) = Model.netLoss((features, Variable(targets)))
            # loss = margin_loss / features.size(0)
            loss.backward()
            optimizer.step()

            # updating all meters
            Model.meterTop1.append(float(top1.cpu().data.numpy() if torch.__version__.startswith("0.4")
                                         else top1.cpu().data.numpy()[0]))
            Model.meterTop5.append(float(top5.cpu().data.numpy() if torch.__version__.startswith("0.4")
                                         else top5.cpu().data.numpy()[0]))
            Model.meterLoss.append(float(loss.cpu().data.numpy() if torch.__version__.startswith("0.4")
                                         else loss.cpu().data.numpy()[0]))

            Model.meterSpeed.append(int(float(args.BSZ)/(timeit.default_timer()-timer)))
            timer = timeit.default_timer()

            print("... {:6d} :: Cost {:2.3f} :: Top1/Top5 - {:3.2f}/{:3.2f} :: {:4d} I/S         ".format(
                  Model.meterIterations, Model.meterLoss[-1], Model.meterTop1[-1],
                  Model.meterTop5[-1], Model.meterSpeed[-1]), end="\r")
            sys.stdout.flush()
            max_i = i

        # save every epoch and print the average of epoch
        print("... {:6d} :: Cost {:1.3f} :: Top1/Top5 - {:3.2f}/{:3.2f} :: {:4d} I/S     ".format(Model.meterIterations,
              np.mean(Model.meterLoss[-max_i:]), np.mean(Model.meterTop1[-max_i:]),
              np.mean(Model.meterTop5[-max_i:]), int(np.mean(Model.meterSpeed[-max_i:]))))
        NeuralEssentials.SaveModel(Model)

        test_top1, test_top5 = [], []
        Model.netEmbedding.eval()
        Model.netLoss.eval()
        for i, (tensor, targets) in enumerate(test_loader):
            Model.netEmbedding.zero_grad()
            Model.netLoss.zero_grad()
            features = Model.netEmbedding(Variable(tensor))
            loss, (top1, top5) = Model.netLoss((features, Variable(targets)))

            test_top1.append(float(top1.cpu().data.numpy() if torch.__version__.startswith("0.4")
                             else top1.cpu().data.numpy()[0]))
            test_top5.append(float(top5.cpu().data.numpy() if torch.__version__.startswith("0.4")
                             else top5.cpu().data.numpy()[0]))
        print("... Test accuracy - {:3.2f}/{:3.2f} ".format(np.mean(test_top1), np.mean(test_top5)))
        Model.netEmbedding.train()
        Model.netLoss.train()

    print("\nDone with training")
    return Model
# ============================================================================ #


def parse_args():
    parser = argparse.ArgumentParser(description="ImageNet using TensorMONK!!!")
    parser.add_argument("-A", "--Architecture", type=str, default="residual50",
                        choices=["residual18", "residual34",
                                 "residual50", "residual101", "residual152",
                                 "resnext50", "resnext101", "resnext152",
                                 "seresidual50", "seresidual101", "seresidual152",
                                 "inceptionv4", "mobilev1", "mobilev2",
                                 "shuffle1", "shuffle2", "shuffle3", "shuffle4", "shuffle8"])

    parser.add_argument("-B", "--BSZ", type=int,  default=32)
    parser.add_argument("-E", "--Epochs", type=int,  default=6)

    parser.add_argument("--optimizer", type=str, default="sgd", choices=["adam", "sgd"])
    parser.add_argument("--learningRate", type=float, default=0.06)

    parser.add_argument("--loss_type", type=str, default="entr", choices=["entr", "smax", "tentr", "tsmax", "lmcl"])
    parser.add_argument("--loss_distance", type=str, default="dot", choices=["cosine", "dot"])

    parser.add_argument("--default_gpu", type=int,  default=0)
    parser.add_argument("--gpus", type=int,  default=1)
    parser.add_argument("--cpus", type=int,  default=6)

    parser.add_argument("--trainDataPath", type=str,  default="./data/ImageNet/train")
    parser.add_argument("--testDataPath", type=str,  default="./data/ImageNet/validation")
    parser.add_argument("-I", "--ignore_trained", action="store_true")

    return parser.parse_args()


if __name__ == '__main__':
    Model = train()