-
Notifications
You must be signed in to change notification settings - Fork 4
/
options.py
198 lines (168 loc) · 8.74 KB
/
options.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
187
188
189
190
191
192
193
194
195
196
197
198
# MIT License
#
# Copyright (c) 2020 Mehran Maghoumi
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in all
# copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.
# ----------------------------------------------------------------------------------------------------------------------
import argparse
import numpy as np
import os
import platform
import time
import torch
import uuid
from dataloader.dataset import Dataset
# ----------------------------------------------------------------------------------------------------------------------
class Options:
"""This class defines options passed via command-line arguments."""
def __init__(self):
"""
The constructor
"""
self.parser = argparse.ArgumentParser(description='DeepNAG Training')
self.parser.add_argument('--datapath', default='data',
help='Path where the datasets are located')
self.parser.add_argument('--dataset-name', choices=Dataset.dataset_names, default='dollar-gds',
help=F"Dataset to train on: {' | '.join(Dataset.dataset_names)}")
self.parser.add_argument('--seed', default=1234, type=int,
help='Random number generator seed, pass nothing or -1 to use a random seed')
# Training arguments
self.parser.add_argument('--model', type=str, default='DeepNAG', choices=['DeepNAG', 'DeepGAN'],
help="The network model to train: ('DeepNAG' | 'DeepGAN')")
self.parser.add_argument('--latent-dim', type=int, default=32,
help='Latent space dimension')
self.parser.add_argument('--resample-n', type=int, default=64,
help='The number of equidistant points to spatially resample every sample to')
self.parser.add_argument('--lr', type=float, default=1e-4,
help="Adam's learning rate")
self.parser.add_argument('--beta0', type=float, default=0.5,
help="Adam's beta0 value")
self.parser.add_argument('--beta1', type=float, default=0.9,
help="Adam's beta1 value")
self.parser.add_argument('--batch-size', type=int, default=64,
help='Batch size')
self.parser.add_argument('--epochs', type=int, default=None,
help='Number of epochs to run the training for. Default optimals will be used if not provided.')
# DeepGAN-specific parameters
self.parser.add_argument('--deepgan-critic-iters', type=int, default=5,
help="Number of steps to train DeepGAN's critic per every training step of the generator.")
self.parser.add_argument('--deepgan-lambda', type=float, default=10,
help="WGAN-GP's loss regularizer.")
# Logging
self.parser.add_argument('--use-tensorboard', type=int, default=0, choices=[0, 1],
help='Determines whether to use tensorboard for logging')
self.parser.add_argument('--vis-frequency', type=int, default=100,
help='Determines after how many epochs to visualize the results (zero will disable visualization)')
self.parser.add_argument('--checkpoint-frequency', type=int, default=10000,
help='Determines after how many epochs to save a checkpoint of the trained model (zero will disable frequent checkpointing)')
self.parser.add_argument('--experiment-name', type=str, default=None,
help="Optional name for the experiment. This name will be used in the log directory's name")
# Trained model evaluation
self.parser.add_argument('--evaluate', type=str, default=None,
help='Path to a saved checkpoint to evaluate')
self.datapath = None
self.dataset_name = None
self.seed = None
self.model = None
self.latent_dim = None
self.resample_n = None
self.lr = None
self.beta0 = None
self.beta1 = None
self.batch_size = None
self.epochs = None
self.deepgan_critic_iters = None
self.deepgan_lambda = None
self.use_tensorboard = None
self.vis_frequency = None
self.checkpoint_frequency = None
self.experiment_name = None
self.evaluate = None
# Run-specific parameters
self.run_name = None
self.run_dir = None
self.run_tb_dir = None # The tensorboard directory of this run
self.run_checkpoint_dir = None # The checkpoint directory of this run
self.run_log_file = None # The stdout dump file of this run
self.run_err_file = None # The stderr dump file of this run
self.use_cuda = torch.cuda.is_available()
def parse(self):
"""
Parse the command-line arguments
"""
args = self.parser.parse_args()
self.datapath = args.datapath
self.dataset_name = args.dataset_name
self.seed = args.seed
self.model = args.model
self.latent_dim = args.latent_dim
self.resample_n = args.resample_n
self.lr = args.lr
self.beta0 = args.beta0
self.beta1 = args.beta1
self.batch_size = args.batch_size
self.epochs = args.epochs
self.deepgan_critic_iters = args.deepgan_critic_iters
self.deepgan_lambda = args.deepgan_lambda
self.use_tensorboard = args.use_tensorboard
self.vis_frequency = args.vis_frequency
self.checkpoint_frequency = args.checkpoint_frequency
if args.experiment_name is not None:
self.experiment_name = args.experiment_name.replace(' ', '-')
self.evaluate = args.evaluate
# Take action with some of the parameters
self.seed = hash(uuid.uuid4()) if (self.seed is None or self.seed == -1) else self.seed
torch.manual_seed(self.seed)
np.random.seed(self.seed)
# If the number of epochs is not set, use some optimal defaults (only if training).
if self.epochs is None and self.evaluate is None:
self.epochs = 25000 if self.model == 'DeepNAG' else 150000
print(F'The number of training epochs was automatically set to {self.epochs}')
if self.deepgan_critic_iters < 1:
raise Exception(F"The value '{self.deepgan_critic_iters}' is invalid for --deepgan-critic-iters")
# Run-specific names and unique strings
self.make_run_dir()
def make_run_dir(self):
"""
Creates the run directories to dump the logs and checkpoints.
"""
timestamp = time.strftime("%Y%m%d-%H%M%S")
# Build the run system's name
system_name = str(platform.node())
self.run_name = F"{self.dataset_name}-{self.model}-{timestamp}-{system_name}"
# Append the optional experiment name
if self.experiment_name is not None:
self.run_name += F"-{self.experiment_name}"
# Make all the run directories
if not os.path.isdir('logs'):
os.mkdir('logs')
self.run_dir = os.path.join('logs', self.run_name)
os.mkdir(self.run_dir)
self.run_tb_dir = os.path.join(self.run_dir, 'tensorboard')
os.mkdir(self.run_tb_dir)
self.run_checkpoint_dir = os.path.join(self.run_dir, 'checkpoints')
os.mkdir(self.run_checkpoint_dir)
os.mkdir(os.path.join(self.run_dir, 'logs'))
self.run_log_file = os.path.join(self.run_dir, 'logs', 'log.txt')
self.run_err_file = os.path.join(self.run_dir, 'logs', 'err.txt')
def get_path_from_root(self, path):
"""
:return: the specified path, prepended by the data path root.
"""
return os.path.join(self.datapath, path)