candidate_training/utils.py at master · MLI-lab/candidate_training · 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
import numpy as np

def readProgressLog(out_file, two_tests=True):

    prog_log = {keyword:list() for keyword in ['train', 'test']}
    if two_tests:
        prog_log.update({'subset':list()})

    with open(out_file, 'r') as fn:
        new_lines = fn.readlines()

    cntr = 3
    sub_cnt = None
    if two_tests:
        sub_cnt = 2

    for new_line in new_lines:

        if cntr == 0:
            prog_log['train'].append(float(new_line.split(',')[0]))
            cntr += 1

        elif cntr == 1:
            prog_log['test'].append(float(new_line.split(',')[0]))
            cntr += 1

        elif cntr == sub_cnt:
            prog_log['subset'].append(float(new_line.split(',')[0]))
            cntr += 1

        if new_line[:5] == 'Epoch':
            cntr = 0

    return prog_log

def get_max_index(acc_dict, dataset='test'):
    acc_arr = np.array(acc_dict[dataset])
    acc_max = acc_arr.max()
    return acc_max, np.where(acc_arr == acc_max)

def get_min_index(acc_dict, dataset='test'):
    acc_arr = np.array(acc_dict[dataset])[40:]
    acc_min = acc_arr.min()
    return acc_min, np.where(acc_arr == acc_min)

def readRedundancyProgress(out_file):

    redun_dict = dict()
    cntr = 3
#     set_trace()

    with open(out_file, 'r') as fn:
        new_lines = fn.readlines()

    for new_line in new_lines:
        if new_line[:9] == 'cifar_len':
            cifar_len = new_line.split(', ')[0].split(' = ')[1]
            times_cifar = new_line.split(', ')[1].split(' = ')[1]
            tiny_frac = new_line.split(', ')[2].split(' = ')[1].strip()
            if not cifar_len in redun_dict.keys():
                redun_dict[cifar_len] = dict()
            if not times_cifar in redun_dict[cifar_len].keys():
                redun_dict[cifar_len][times_cifar] = dict()
            if not tiny_frac in redun_dict[cifar_len][times_cifar].keys():
                redun_dict[cifar_len][times_cifar][tiny_frac] = dict()
                redun_dict[cifar_len][times_cifar][tiny_frac]['train'] = list()
                redun_dict[cifar_len][times_cifar][tiny_frac]['test'] = list()

        if cntr == 0:
            redun_dict[cifar_len][times_cifar][tiny_frac]['train'].append(float(new_line.split(',')[0]))
            cntr += 1

        elif cntr == 1:
            redun_dict[cifar_len][times_cifar][tiny_frac]['test'].append(float(new_line.split(',')[0]))
            cntr += 1

        if new_line[:5] == 'Epoch':
            cntr = 0

    return redun_dict

def readGradResults(out_file, two_tests=False):

    prog_log = {keyword:list() for keyword in ['train', 'test', 'loss']}
    if two_tests:
        prog_log.update({'cifar_test':list()})

    with open(out_file, 'r') as fn:
        new_lines = fn.readlines()

    cntr = 4
    cifar_cnt = 5
    loss_cnt = 2
    if two_tests:
        cifar_cnt = 2
        loss_cnt = 3
    epoch_arr = []

    for new_line in new_lines:

        if cntr == 0:
            prog_log['train'].append(float(new_line.split(',')[0]))
            cntr += 1

        elif cntr == 1:
            prog_log['test'].append(float(new_line.split(',')[0]))
            cntr += 1

        elif cntr == cifar_cnt:
            prog_log['cifar_test'].append(float(new_line.split(',')[0]))
            cntr += 1

        elif cntr == loss_cnt:
            prog_log['loss'].append(float(new_line.split(',')[0]))
            cntr += 1

        if new_line[:5] == 'Epoch':
            cntr = 0
            epoch_arr.append(new_line.split(' ')[-1])

    return prog_log, epoch_arr