-
Notifications
You must be signed in to change notification settings - Fork 3
/
train.py
161 lines (131 loc) · 3.94 KB
/
train.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
"""
Author: Jonathan Hampton
June 2020
"""
# Imports
import torch
import argparse
import pytorch_lightning as pl
from imitation_network import ImitationNetwork
from pytorch_lightning.callbacks import ModelCheckpoint
from pytorch_lightning.logging import TensorBoardLogger
def main():
""" Arguments """
parser = argparse.ArgumentParser(description="Train imitation network")
parser.add_argument(
'-lr', '--learning_rate',
type=float,
default=0.0002,
help="Learning rate for training network. Default = 0.0002")
parser.add_argument(
'-tb', '--train_batch_size',
type=int,
default=120,
help="Batch size for training data. Default = 120")
parser.add_argument(
'-vb', '--val_batch_size',
type=int,
default=120,
help="Batch size for validation data. Default = 120")
parser.add_argument(
'-t', '--train_data_dir',
type=str,
default="data-and-checkpoints/imitation_data/SeqTrain/",
help="path to training data. Default == data-and-checkpoints/imitation_data/SeqTrain/")
parser.add_argument(
'-v', '--val_data_dir',
type=str,
default="data-and-checkpoints/imitation_data/SeqVal/",
help="path to validation data. Default == data-and-checkpoints/imitation_data/SeqVal/")
parser.add_argument(
'-g', '--gpus',
default=1,
help="Number of GPUs. Default == 1.")
parser.add_argument(
'-n', '--run_name',
type=str,
default="",
help="Name of training run for logging.")
parser.add_argument(
'-e', '--max_epochs',
type=int,
default=1,
help="Maximum number of epochs: Default is 1.")
parser.add_argument(
'-ch', '--checkpoint_callback',
type=bool,
default=True,
help="Save checkpoints of network. Default = True")
parser.add_argument(
'-fc', '--from_checkpoint',
type=bool,
default=False,
help="begin training from checkpoint. Default = False")
parser.add_argument(
'-cp', '--checkpoint_path',
type=str,
default=None,
help="path to checkpoint to resume training with. Default = None")
parser.add_argument(
'-es', '--early_stop_callback',
type=bool,
default=True,
help="Enable early stopping. Default = True")
parser.add_argument(
'-pr', '--profiler',
type=bool,
default=False,
help="Enable profiler. Default = False")
parser.add_argument(
'-dc', '--data_cache_size',
type=int,
default=100,
help="Number of H5 files to be loaded at once in memory. Since there is training and validation datasets, size will be doubled. Default=100")
parser.add_argument(
'-l', '--loss_lambda',
type=float,
default=0.5,
help="Value of lambda in loss function.")
args = parser.parse_args()
hparams = argparse.Namespace(**{'learning_rate':args.learning_rate,
'train_batch_size': args.train_batch_size, 'val_batch_size': args.val_batch_size})
""" Setup Network """
checkpoint_callback = ModelCheckpoint(
filepath='data-and-checkpoints/model_checkpoints/{epoch}-{val_loss:.2f}',
save_last=True,
monitor='val_loss')
network = ImitationNetwork(
data_cache_size = args.data_cache_size,
lamb = args.loss_lambda,
hparams=hparams,
train_data_dir=args.train_data_dir,
val_data_dir=args.val_data_dir)
logger = TensorBoardLogger(
"training_logs",
name='Conditional Imitation Learning Network',
version=args.run_name)
if args.from_checkpoint:
if args.checkpoint_path == None:
Raise("Please specify path to checkpoint file (.ckpt)")
else:
trainer = pl.Trainer(
resume_from_checkpoint=args.checkpoint_path,
early_stop_callback=args.early_stop_callback,
max_epochs=args.max_epochs,
gpus=args.gpus,
logger=logger,
checkpoint_callback=checkpoint_callback,
profiler=args.profiler)
else:
trainer = pl.Trainer(
early_stop_callback=args.early_stop_callback,
max_epochs=args.max_epochs,
gpus=args.gpus,
logger=logger,
checkpoint_callback=checkpoint_callback,
profiler=args.profiler)
""" Train! :-) """
trainer.fit(network)
print("Training complete! Best checkpoint is", checkpoint_callback.best_model_path)
if __name__ == "__main__":
main()