set_random_seed(_env['seed'])
project_name = _root.split("/")[-1]
run_name = (f"{_model['name']}_{_model['size']}-"
f"lr_{_training['lr']}-bsz_{_training['batch_size']}-"
f"seed_{_env['seed']}")
now = datetime.now().strftime('%Y-%m-%d_%Hh%Mm%Ss')
tokenizer = get_tokenizer(_model['name'], _model['size'])
train_dataset = CustomDataset(_root, 'train', tokenizer, _training["max_len"])
dev_dataset = CustomDataset(_root, 'dev', tokenizer, _training["max_len"])
Model = get_model_class(_model['name'])
Opt = get_optim_class(_model['opt'])
Loss_fn = get_loss_fn_class(_model['loss'])
model = Model(n_outputs=train_dataset.n_outputs, size=_model['size'],
pretrained_model_path=str2bool(_model['pretrained_model_path']))
metric_dic = {
"acc": Accuracy(),
"precision": Precision()
}
callbacks = [
ModelCheckpoint(f"{_save_model_root}/{run_name}.pth", monitor='dev_loss', mode="min")
]
trainer = Trainer(model=model, loss_fn_class=Loss_fn, optimizer_class=Opt, metrics=metric_dic)
trainer.fit(train_dataset, dev_dataset, lr=_training['lr'], epochs=_training['epochs'],
batch_size=_training['batch_size'], callbacks=callbacks)
""" load model """
model = prepare_model(args)
model.cuda()
""" define loss """
criterion = nn.CrossEntropyLoss()
""" setup metrics """
metric = MeanIOUScore(9)
""" setup optimizer """
optimizer = torch.optim.Adam(model.parameters(), lr=args.lr, weight_decay=args.weight_decay)
""" setup tensorboard """
writer = SummaryWriter(os.path.join(args.save_dir, "train_info"))
""" setup trainer """
trainer = Trainer(
model,
optimizer,
criterion,
args.accumulate_gradient,
train_loader,
val_loader,
writer,
metric,
args.save_dir,
)
trainer.fit(args.epochs)