Can anyone give a simple example to load and run inference on GPT-2 model in JAVA?

I have exported the GPT-2 model in the ONNX format but I am not able to find any examples to perform the inference in the Java. Please help and share some code.

Thank you 🙏🏻

I was able to somehow get the following code running but I am sure there can be more like setting temperature, automatically tokenizing the string, sampling, topK etc... I am not sure how.

import ai.onnxruntime.*;
import java.nio.LongBuffer;
import java.util.Map;

public class onnx_importer {
    public static void main(String args[]) throws OrtException {
        // Load the model and create InferenceSession
        System.out.println("This is the model loading");
        String modelPath = "decoder_model.onnx";
        OrtEnvironment env = OrtEnvironment.getEnvironment();
        OrtSession session = env.createSession(modelPath);

        // The actual tokens (manually because I don't know where to find the tokenizer!)
        long[] inputData = new long[]{32, 1, 2, 35};  // Your input data in the appropriate format
        long[] inputShape = new long[]{1, 4};  // Shape of the input data

        // Attention mask needed (all ones - same shape as the input ids)
        long[] inputData_a = new long[]{1, 1, 1, 1};  // Your input data in the appropriate format
        long[] inputShape_a = new long[]{1, 4};  // Shape of the input data

        OnnxTensor inputTensor = OnnxTensor.createTensor(env, LongBuffer.wrap(inputData), inputShape);
        OnnxTensor attTensor = OnnxTensor.createTensor(env, LongBuffer.wrap(inputData_a), inputShape_a);

        // This is the forward pass so that we can get the logits
        var inputs = Map.of("input_ids", inputTensor, "attention_mask", attTensor);
        var results = session.run(inputs);

        // This is the output (one of the outputs)
        OnnxValue a = results.get(0);
        System.out.println(a.getValue());
        System.out.println(a.getInfo());
        System.out.println(a.getInfo());

    }
}

Start from the Java code in the issue and review the repository's existing inference examples for the appropriate entry point. Done would require a clear, runnable GPT-2 ONNX Runtime Java example that explains tokenization and the requested generation options, but the issue does not name a target file or test.