-
Notifications
You must be signed in to change notification settings - Fork 0
/
main.py
51 lines (42 loc) · 1.57 KB
/
main.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
import json
import random
import os
import numpy as np
import torch
import torch.nn as nn
from models.multimodal import MultiModalModel
from models.backbone import resnet18
from train import *
from utils import *
import os
import hydra
from torch.utils.tensorboard import SummaryWriter
def build_model(cfg):
if cfg.modality == 'None':
video_model = hydra.utils.instantiate(cfg.encoder_v).to(cfg.device)
audio_model = hydra.utils.instantiate(cfg.encoder_a).to(cfg.device)
model = MultiModalModel(video_model, audio_model,cfg.n_classes,cfg.fusion_method)
cfg.num_modal = 2
print(cfg.method, cfg.fusion_method)
return model
@hydra.main(config_path='cfgs', config_name='train', version_base=None)
def main(cfg):
print(cfg)
setup_seed(cfg.random_seed)
model = build_model(cfg)
if cfg.train:
if cfg.tensorboard:
tb_writer = SummaryWriter(log_dir=cfg.result_path)
else:
tb_writer = None
(train_loader, val_loader) = get_dataset(cfg)
train_logger, train_batch_logger, val_logger = get_logger(cfg)
model = nn.DataParallel(model, device_ids=cfg.gpu_device).cuda()
parameters = [p for p in model.parameters()]
optimizer = hydra.utils.instantiate(cfg.optimizer, params = parameters)
scheduler = hydra.utils.instantiate(cfg.scheduler, optimizer = optimizer)
train(train_loader,val_loader,model,train_logger,val_logger,train_batch_logger,tb_writer,cfg,\
optimizer, scheduler)
# method=CRMT_JT,CRMT_AT,CRMT_mix
if __name__ == '__main__':
main()