let session = null;

async function loadModel() {
    try {
        session = await ort.InferenceSession.create("iris_model.onnx");

        document.getElementById("result").innerText =
            "Model loaded successfully!";

    } catch (error) {
        document.getElementById("result").innerText =
            "Error loading model: " + error;
    }
}

async function predict() {

    if (!session) {
        alert("Model is still loading. Please wait.");
        return;
    }

    const sepalLength = parseFloat(
        document.getElementById("sepalLength").value
    );

    const sepalWidth = parseFloat(
        document.getElementById("sepalWidth").value
    );

    const petalLength = parseFloat(
        document.getElementById("petalLength").value
    );

    const petalWidth = parseFloat(
        document.getElementById("petalWidth").value
    );

    const input = new Float32Array([
        sepalLength,
        sepalWidth,
        petalLength,
        petalWidth
    ]);

    const tensor = new ort.Tensor(
        "float32",
        input,
        [1, 4]
    );

    const results = await session.run({
        float_input: tensor
    });

    const prediction = results.label.data[0];

    const flowers = [
        "Setosa",
        "Versicolor",
        "Virginica"
    ];

    document.getElementById("result").innerText =
        "Prediction: " + flowers[prediction];
}

loadModel();