/
train.py
38 lines (34 loc) · 1.38 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
from torch.utils.data import DataLoader
import numpy as np
from configs import config_dict
from datasets import dataset_dict
import torch
import time
import argparse
import os
from CaRTS import build_model
def parse_args():
parser = argparse.ArgumentParser()
parser.add_argument("--config", type=str, help="Name of the configuration file.")
parser.add_argument("--model_path", type=str, default=None, help="Path to the model checkpoint file")
args = parser.parse_args()
return args
if __name__ == "__main__":
args = parse_args()
cfg = config_dict[args.config]
use_gpu = torch.cuda.is_available()
if use_gpu:
print("use_gpu")
device = torch.device("cuda")
else:
device = torch.device("cpu")
train_dataset = dataset_dict[cfg.train_dataset['name']](**(cfg.train_dataset['args']))
train_dataloader = DataLoader(train_dataset, batch_size=4, shuffle=True)
validation_dataset = dataset_dict[cfg.validation_dataset['name']](**(cfg.validation_dataset['args']))
validation_dataloader = DataLoader(validation_dataset, batch_size=1, shuffle=False)
model = build_model(cfg.model, device)
if args.model_path is None:
loss_plot = model.train_epochs(train_dataloader, validation_dataloader)
else:
model.load_parameters(args.model_path)
loss_plot = model.train_epochs(train_dataloader, validation_dataloader)