CriticalPeriodMultiView/plot_multiple.py at main · mjkleinman/CriticalPeriodMultiView · 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
#!/usr/bin/env python
import sh
import os, random, time, sys
# import torch
from operator import itemgetter
import json
import itertools
from logger import Logger
import argparse
import numpy as np
import matplotlib as mpl
mpl.use('Agg')
import matplotlib.ticker as tck
import matplotlib.pyplot as plt
from pathlib import Path
import os
import seaborn as sns
from scipy.signal import savgol_filter

parser = argparse.ArgumentParser(description='PyTorch ImageNet Training')
parser.add_argument('experiment', nargs='+', type=str,
                    help='experiment to plot')
parser.add_argument('--format', '-f', default='pdf', type=str,
                    help='output a format')
parser.add_argument('--title', '-t', default=None, type=str,
                    help='title of the plot')
parser.add_argument('--min-y', '-y', default=None, type=float,
                    help='minimum value of y')
parser.add_argument('--max-x', '-x', default=None, type=float,
                    help='maximum value of y')
parser.add_argument('--save', '-s', action='store_true',
                    help='save the plot')
parser.add_argument('--hide-y', '-i', action='store_true',
                    help='hide y axis')
parser.add_argument('--save_dir', default=None, type=str)
parser.add_argument('--name', '-n', default='sresnet.pdf', type=str)
args = parser.parse_args()

plt.figure(figsize=(4,2.7))

for i, experiment in enumerate(args.experiment):
    with open('experiments/{}.json'.format(experiment)) as f:
        exlog = json.load(f)

    x, y = [], []
    for run_id in exlog['runs']:
        try:
            logger = Logger.load(run_id)
        except ValueError:
            continue
        start_epoch = logger['args'].start_epoch
        if start_epoch == 0:
            start_epoch = logger.get('train')[0]['epoch']
        test_errors = [v['error'] for v in logger.get('valid')]
        test_error = np.mean(test_errors[-5:])
        x += [start_epoch-1]
        y += [test_error]

    # if x[-1] < 200:
    #     plt.axes().xaxis.set_major_locator(tck.MultipleLocator(base=20))
    # else:
    # plt.axes().xaxis.set_major_locator(tck.MultipleLocator(base=80))

    y = 100 - np.array(y)
    print(y)
    plt.plot(x,y-y[0], marker='o',markersize=5, label='n={}'.format(i+1))


# if args.min_y is not None:
#     plt.ylim(args.min_y,None)
# else:
#     plt.ylim(min(y)-1.,None)
# # plt.ylim(98,None)
# if args.max_x is not None:
#     plt.xlim(0,args.max_x)
# else:
#     plt.xlim(0,max(x))

#
# vals = plt.axes().get_yticks()
# plt.axes().set_yticklabels(['{:.0f}%'.format(x) for x in vals])

plt.xlabel('Deficit removal (epoch)')
plt.ylabel('Relative test accuracy')
if args.title is not None:
    plt.title(args.title)
if args.hide_y:
    plt.gca().yaxis.label.set_visible(False)
    # plt.gca().yaxis.set_visible(False)

# plt.legend()
plt.legend(loc='center left', bbox_to_anchor=(1, 0.5), frameon=False)
plt.tight_layout()

output_dir = os.path.join('plots', args.save_dir) if args.save_dir is not None else 'plots'
print(output_dir)
Path(output_dir).mkdir(parents=True, exist_ok=True)
save_filename = args.name
savepath = os.path.join(output_dir, save_filename)
plt.savefig(savepath, format=args.format, dpi=None, bbox_inches='tight')