Is dtype of calibration_data the same as dtype of onnx?

[canyueduxuan]

Make sure you already checked the examples and documentation before submitting an issue.

How would you like to use ModelOpt

• My question is, I want to quantize my ONNX model (float32) to int8. Should the dtype (data type) of the calibration data be FLOAT32 or INT8?

import modelopt.onnx.quantization as moq

calibration_data = np.load(calibration_data_path)

moq.quantize(
    onnx_path=onnx_path,
    calibration_data=calibration_data,
    output_path="quant.onnx",
    quantize_mode="int8",
)