aiops/ContraLSP/hmm/main.py [196:243]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
            optim="adam",
            lr=0.01,
        )
        explainer = ExtremalMask(classifier)
        _attr = explainer.attribute(
            x_test,
            additional_forward_args=(True,),
            trainer=trainer,
            mask_net=mask,
            batch_size=100,
        )
        attr["extremal_mask"] = _attr.to(device)

    if "gate_mask" in explainers:
        trainer = Trainer(
            max_epochs=500,
            accelerator=accelerator,
            devices=device_id,
            log_every_n_steps=2,
            deterministic=deterministic,
            logger=TensorBoardLogger(
                save_dir=".",
                version=random.getrandbits(128),
            ),
        )
        mask = GateMaskNet(
            forward_func=classifier,
            model=nn.Sequential(
                RNN(
                    input_size=x_test.shape[-1],
                    rnn="gru",
                    hidden_size=x_test.shape[-1],
                    bidirectional=True,
                ),
                MLP([2 * x_test.shape[-1], x_test.shape[-1]]),
            ),
            lambda_1=lambda_1,
            lambda_2=lambda_2,
            optim="adam",
            lr=0.01,
        )
        explainer = GateMask(classifier)
        _attr = explainer.attribute(
            x_test,
            additional_forward_args=(True,),
            trainer=trainer,
            mask_net=mask,
            batch_size=x_test.shape[0],
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



aiops/ContraLSP/switchstate/main.py [191:238]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
            optim="adam",
            lr=0.01,
        )
        explainer = ExtremalMask(classifier)
        _attr = explainer.attribute(
            x_test,
            additional_forward_args=(True,),
            trainer=trainer,
            mask_net=mask,
            batch_size=100,
        )
        attr["extremal_mask"] = _attr.to(device)

    if "gate_mask" in explainers:
        trainer = Trainer(
            max_epochs=500,
            accelerator=accelerator,
            devices=device_id,
            log_every_n_steps=2,
            deterministic=deterministic,
            logger=TensorBoardLogger(
                save_dir=".",
                version=random.getrandbits(128),
            ),
        )
        mask = GateMaskNet(
            forward_func=classifier,
            model=nn.Sequential(
                RNN(
                    input_size=x_test.shape[-1],
                    rnn="gru",
                    hidden_size=x_test.shape[-1],
                    bidirectional=True,
                ),
                MLP([2 * x_test.shape[-1], x_test.shape[-1]]),
            ),
            lambda_1=lambda_1,
            lambda_2=lambda_2,
            optim="adam",
            lr=0.01,
        )
        explainer = GateMask(classifier)
        _attr = explainer.attribute(
            x_test,
            additional_forward_args=(True,),
            trainer=trainer,
            mask_net=mask,
            batch_size=x_test.shape[0],
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -