loss_epoch[idx] += print_loss
loss_updates[idx] += 1
for component_index in range(num_patches):
if opt.validate:
validation_loss = validate(opt, models[component_index], test_loader) #test_loader corresponds to validation set here
logs.append_val_loss(validation_loss)
logs.append_train_loss([x / loss_updates[idx] for idx, x in enumerate(loss_epoch)])
logs.create_log(models[component_index], epoch=epoch, optimizer=optimizers[component_index], component_idx=component_index)
if __name__ == "__main__":
opt = arg_parser.parse_args()
arg_parser.create_log_path(opt)
opt.training_dataset = "unlabeled"
# random seeds
torch.manual_seed(opt.seed)
torch.cuda.manual_seed(opt.seed)
np.random.seed(opt.seed)
if opt.device.type != "cpu":
torch.backends.cudnn.benchmark = True
num_patches = opt.grid_dims * opt.grid_dims
# load model
models, optimizers = [None for i in range(num_patches)], [None for i in range(num_patches)]
for i in range(num_patches):
"Step [{}/{}], Time (s): {:.1f}, Acc1: {:.4f}, Acc5: {:.4f}, Loss: {:.4f}"
.format(step, total_step,
time.time() - starttime, acc1, acc5, sample_loss))
starttime = time.time()
print("Testing Accuracy: ", epoch_acc1 / total_step)
return epoch_acc1 / total_step, epoch_acc5 / total_step, loss_epoch / total_step
if __name__ == "__main__":
opt = arg_parser.parse_args()
add_path_var = "linear_model"
arg_parser.create_log_path(opt, add_path_var=add_path_var)
opt.training_dataset = "train"
# random seeds
torch.manual_seed(opt.seed)
torch.cuda.manual_seed(opt.seed)
np.random.seed(opt.seed)
# load pretrained model
context_model, _ = load_vision_model.load_model_and_optimizer(
opt, reload_model=True, calc_loss=False)
context_model.module.switch_calc_loss(False)
## model_type=2 is supervised model which trains entire architecture; otherwise just extract features
if opt.model_type != 2:
context_model.eval()