epoch, h[_]['ally_train'][-1], h[_]['ally_valid'][-1], correct / total)) checkpoint_location = \ 'checkpoints/{}/{}_training_history_{}.pkl'.format( expt, model, time_stamp) sep() logging.info('Saving: {}'.format(checkpoint_location)) pkl.dump(h, open(checkpoint_location, 'wb')) if __name__ == "__main__": expt = 'mimic' model = 'ind_gan' marker = 'H' pr_time, fl_time = time_stp() logger(expt, model, fl_time, marker) log_time('Start', pr_time) args = comparison_argparse() main(model=model, time_stamp=fl_time, device=args['device'], ally_classes=args['n_ally'], advr_1_classes=args['n_advr_1'], advr_2_classes=args['n_advr_2'], encoding_dim=args['dim'], hidden_dim=args['hidden_dim'], leaky=args['leaky'], test_size=args['test_size'],
logging.info('Saving: {}'.format(model_ckpt)) torch.save(ally.state_dict(), model_ckpt) for idx, advr in enumerate(advrs): model_ckpt = 'ckpts/{}/models/{}_advr_{}.stop'.format( expt, template, idx) logging.info('Saving: {}'.format(model_ckpt)) torch.save(advr.state_dict(), model_ckpt) if __name__ == '__main__': expt = 'mnist' model = 'encd_pretrain' marker = 'A' pr_time, fl_time = time_stp() logger(expt, model, fl_time, marker) log_time('Start', pr_time) args = eigan_argparse() main( model=model, device=args['device'], ally_classes=args['n_ally'], advr_classes=args['n_advr'], batch_size=args['batch_size'], n_epochs=args['n_epochs'], lr_encd=args['lr_encd'], lr_ally=args['lr_ally'], lr_advr=args['lr_advr'], expt=args['expt'],