如何在Java中加载一个带有'predict‘Sgnature Def的Tensorflow SavedModel？

import org.tensorflow.SavedModelBundle;
import org.tensorflow.Session;
import org.tensorflow.Tensor;

public class TensorFlowExample {
    public static void main(String[] args) {
        // 加载SavedModel
        SavedModelBundle savedModel = SavedModelBundle.load("path/to/saved_model", "serve");

        // 创建Session
        Session session = savedModel.session();

        // 获取Signature Def
        MetaGraphDef metaGraphDef = savedModel.metaGraphDef();

        // 获取Signature Def的输入和输出
        SignatureDef signatureDef = metaGraphDef.getSignatureDefOrThrow("predict");
        TensorInfo inputTensorInfo = signatureDef.getInputsOrThrow("input");
        TensorInfo outputTensorInfo = signatureDef.getOutputsOrThrow("output");

        // 创建输入Tensor
        float[] inputData = {1.0f, 2.0f, 3.0f};
        Tensor<Float> inputTensor = Tensor.create(inputData, Float.class);

        // 运行模型
        Tensor<?> outputTensor = session.runner()
                .feed(inputTensorInfo.getName(), inputTensor)
                .fetch(outputTensorInfo.getName())
                .run()
                .get(0);

        // 获取输出Tensor的值
        float[] outputData = new float[3];
        outputTensor.copyTo(outputData);

        // 打印输出结果
        for (float value : outputData) {
            System.out.println(value);
        }

        // 关闭Session和SavedModel
        session.close();
        savedModel.close();
    }
}