def main():
args = parse_args()
model_script = load_module(args.model_path)
cfg = init_experiment(args)
model, model_cfg = model_script.init_model(cfg)
# prepare dataset
dataset = YKDataset(window_size=cfg.window_size, dic_path=cfg.DIC_PATH)
dataset_size = len(dataset)
indices = list(range(dataset_size))
kf = KFold(n_splits=cfg.n_folds, shuffle=True)
for train_indices, val_indices in kf.split(indices):
print('t', train_indices)
print('v', val_indices)
train_sampler = SubsetRandomSampler(train_indices)
val_sampler = SubsetRandomSampler(val_indices)
trainer = Trainer(model, cfg, model_cfg, dataset, train_sampler,
val_sampler, cfg.optimizer)
logger.info(f'Total Epochs: {cfg.n_epochs}')
for epoch in range(cfg.n_epochs):
trainer.training(epoch)
trainer.validation(epoch)
cfg = init_experiment(args)
def main():
args = parse_args()
model_script = load_module(args.model_path)
cfg = init_experiment(args)
model, model_cfg = model_script.init_model(cfg)
# prepare dataset
dataset = YKDataset(window_size=cfg.window_size, dic_path=cfg.DIC_PATH)
dataset_size = len(dataset)
indices = list(range(dataset_size))
split = int(np.floor(VAL_SPLIT * dataset_size))
if SHUFFLE_DATASET:
np.random.seed(RANDOM_SEED)
np.random.shuffle(indices)
train_indices, val_indices = indices[split:], indices[:split]
train_sampler = SubsetRandomSampler(train_indices)
val_sampler = SubsetRandomSampler(val_indices)
trainer = Trainer(model, cfg, model_cfg, dataset, train_sampler,
val_sampler, cfg.optimizer)
logger.info(f'Total Epochs: {cfg.n_epochs}')
for epoch in range(cfg.n_epochs):
trainer.training(epoch)
trainer.validation(epoch)