Example #1
0
    print(network_repr)
    writer.add_text('general', network_repr)

    if args.model == 'unconstrained':
        for module in network.modules():
            if hasattr(module, 'relax'):
                module.relax()
                print(f'relaxing {repr(module)}')

        network_repr = repr(network)
        print(network_repr)
        writer.add_text('general', network_repr)

        n_params = 0
        for param in network.parameters():
            n_params += param.numel()
        print(n_params, 'learnable parameters')

    if cuda:
        network = network.cuda()

    checkpoint = framework.load_checkpoint(args.model_path)
    start_epoch, best_score, model_dict, optim_dict = checkpoint

    network.load_state_dict(model_dict)

    framework.inspect(network,
                      val_loader,
                      args.artifacts,
                      early_stop=args.early_stop)
Example #2
0
    loss_fn.name = 'BCE'

    optim = torch.optim.Adam([
        {
            'params': network.l2_params(),
            'weight_decay': args.l2
        },
        {
            'params': network.nr_params(),
            'weight_decay': 0.
        },
    ],
                             lr=args.lr)

    if args.load:
        checkpoint = framework.load_checkpoint(args.load)
        start_epoch, best_score, model_dict, optim_dict = checkpoint

        network.load_state_dict(model_dict)
        optim.load_state_dict(optim_dict)
        fmt = 'Starting at epoch {}, best score {}. Loaded from {}'
        start_epoch += 1  # skip to the next after loaded
        msg = fmt.format(start_epoch, best_score, args.load)
    else:
        print('Set start epoch and best score to 0')
        best_score = 0.
        start_epoch = 0

    if args.model == 'unconstrained':
        for module in network.modules():
            if hasattr(module, 'relax'):