-
Notifications
You must be signed in to change notification settings - Fork 4
/
plot.py
186 lines (169 loc) · 6.82 KB
/
plot.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
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
import argparse
import io
import os
import json
import argparse
import os
import os.path as osp
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
import numpy as np
DIV_LINE_WIDTH = 50
exp_idx = 0
units = dict()
def read_data(d, f, timesteps=1e6, action=4):
fn = osp.join(d, f)
lines = io.open(fn, 'r', newline='\n', encoding='utf8').readlines()
data = dict()
for line in lines:
di = json.loads(line)
for k, v in di.items():
if k not in data:
data[k] = [v]
else:
data[k].append(v)
if data['step'][-1] * action > timesteps:
break
return pd.DataFrame(data)
def get_datasets(logdir, condition=None, **kwargs):
global exp_idx
global units
datasets = []
for root, _, files in os.walk(logdir):
if 'eval.log' in files:
condition1 = condition
condition2 = condition1 + '-' + str(exp_idx)
exp_idx += 1
if condition1 not in units:
units[condition1] = 0
unit = units[condition1]
units[condition1] += 1
try:
exp_data = read_data(root, 'eval.log', timesteps=kwargs['timesteps'],
action=kwargs['action_repeat'])
except Exception as e:
print('Could not read from %s' % os.path.join(root, 'eval.log'))
print(e)
exit(1)
performance = kwargs.get('yaxis', 'eprewmean')
xaxis = kwargs.get('xaxis', 'step')
exp_data[xaxis] *= kwargs.get('action_repeat', 4)
exp_data.insert(len(exp_data.columns), 'Unit', unit)
exp_data.insert(len(exp_data.columns), 'Condition1', condition1)
exp_data.insert(len(exp_data.columns), 'Condition2', condition2)
exp_data.insert(len(exp_data.columns), 'Performance', exp_data[performance])
datasets.append(exp_data)
return datasets
def get_all_dataset(logdir, legend=None, select=None, exclude=None, **kwargs):
logdirs = []
log2class = dict()
clses = []
class2lengend = dict()
subdirs = os.listdir(logdir)
subdirs = sorted(subdirs)
for subdir in subdirs:
if select is None or all(x in subdir for x in select):
if exclude is None or all(x not in subdir for x in exclude):
cls = '-'.join(subdir.split('-')[:3] + subdir.split('-')[5:-1])
log2class[subdir] = cls
if cls not in clses:
clses.append(cls)
logdirs.append(subdir)
if legend is None or len(clses) != len(legend):
print('Subdir...\n' + '=' * DIV_LINE_WIDTH + '\n')
for subdir in clses:
print(subdir)
print('\n' + '=' * DIV_LINE_WIDTH)
if legend is not None:
print('Legend...\n' + '=' * DIV_LINE_WIDTH + '\n')
for l in legend:
print(l)
print('\n' + '=' * DIV_LINE_WIDTH)
else:
print('legend is None')
exit()
for cls, leg in zip(clses, legend):
class2lengend[cls] = leg
print('Plotting from...\n' + '=' * DIV_LINE_WIDTH + '\n')
print(logdir)
print('\n' + '=' * DIV_LINE_WIDTH)
data = []
for log in logdirs:
leg = class2lengend[log2class[log]]
print('\nloading data from {} / {}\n'.format(log, leg))
data.extend(get_datasets(osp.join(logdir, log), leg, **kwargs))
return data
def plot_data(data, xaxis='timesteps', value="Performance", condition="Condition1", smooth=1, **kwargs):
if smooth > 1:
y = np.ones(smooth)
for datum in data:
x = np.asarray(datum[value])
z = np.ones(len(x))
smoothed_x = np.convolve(x, y, 'same') / np.convolve(z, y, 'same')
datum[value] = smoothed_x
if isinstance(data, list):
data = pd.concat(data, ignore_index=True)
fig = plt.figure(kwargs['figname'], figsize=(8, 6))
sns.set(style="darkgrid", font_scale=2)
sns.tsplot(data=data, time=xaxis, value=value, unit="Unit", condition=condition, ci='sd',)
plt.xlabel(kwargs['xlabel'])
plt.ylabel(kwargs['ylabel'])
plt.legend(loc='best', fontsize=24 ).set_draggable(True)
xscale = np.max(np.asarray(data[xaxis])) > 5e3
if xscale:
# Just some formatting niceness: x-axis scale in scientific notation if max x is large
plt.ticklabel_format(style='sci', axis='x', scilimits=(0,0))
axis = plt.gca()
if not kwargs['disable_vline']:
axis.axvline(100000, color='c', lw=1)
axis.axvline(500000, color='r', lw=1)
plt.tight_layout(pad=0.5)
figdir = 'data/figs'
os.makedirs(figdir, exist_ok=True)
figname = osp.join(figdir, '{}.pdf'.format(kwargs['figname']))
fig.savefig(figname, format='pdf')
plt.show()
def make_plots(logdir, legend, xaxis, yaxis, xlabel, ylabel,
count, smooth, select, exclude, est, **kwargs):
data = get_all_dataset(logdir, legend, select, exclude,
xaxis=xaxis, yaxis=yaxis, **kwargs)
condition = 'Condition2' if count else 'Condition1'
estimator = getattr(np, est)
xlabel = xlabel if xlabel else xaxis
ylabel = ylabel if ylabel else yaxis
plot_data(data, xaxis=xaxis, value='Performance', condition=condition, smooth=smooth,
estimator=estimator, xlabel=xlabel, ylabel=ylabel, **kwargs)
def changeargs(args):
env_task = args.figname
args.action_repeat = 4
args.timesteps = 1e6
if env_task in ['cartpole-swingup_sparse',]:
args.timesteps = 3e6
if env_task in ['cartpole-swingup']:
args.action_repeat = 8
if env_task in ['hopper-hop', 'pendulum-swingup']:
args.timesteps = 4e6
if env_task in ['finger-spin', 'walker_walk']:
args.action_repeat = 2
def main():
parser = argparse.ArgumentParser()
parser.add_argument('logdir', type=str, )
parser.add_argument('--legend', type=str, nargs='*', default=None)
parser.add_argument('--xaxis', '-x', type=str, default='step', )
parser.add_argument('--yaxis', '-y', type=str, default='mean_episode_reward', )
parser.add_argument('--xlabel', type=str, default='Environment Steps')
parser.add_argument('--ylabel', type=str, default='Episode Score')
parser.add_argument('--figname', type=str, default='fig')
parser.add_argument('--count', action='store_true')
parser.add_argument('--smooth', '-s', type=int, default=1)
parser.add_argument('--select', nargs='*', default=None)
parser.add_argument('--exclude', nargs='*', default=None)
parser.add_argument('--est', default='mean')
parser.add_argument('--action_repeat', default=4, type=int)
parser.add_argument('--disable-vline', action='store_true', default=False)
args = parser.parse_args()
changeargs(args)
make_plots(**vars(args))
if __name__ == "__main__":
main()