how onnx models stores weights and layer parameters

[kumar-utkarsh0317]

I am working on generating an ONNX translator for a machine learning framework, and for that, I need to have a clear understanding of how weights and layer parameters are stored in an ONNX model. After exploring several ONNX models, I realized that some of the ONNX models store their weights and layer parameters in separate initializers, while some of the ONNX models store their weights and layer parameters in their attributes class. Is there any way I can figure out how these parameters are stored in the model so that I can apply the appropriate algorithm to extract their parameters?