the required version of onnx==1.16 fails to provide value information for an edge after quantizeLinear after shape inference from the call model = onnx.shape_inference.infer_shapes(model) within the function check_qtype_by_node_type. This eventually leads to an assert error since this information is needed to check expected types. This is resolved in onnx==1.17. However, the current version of onnxruntime has limited support for onnx==1.17 and its default opset.
the required version of onnx==1.16 fails to provide value information for an edge after quantizeLinear after shape inference from the call model = onnx.shape_inference.infer_shapes(model) within the function check_qtype_by_node_type. This eventually leads to an assert error since this information is needed to check expected types. This is resolved in onnx==1.17. However, the current version of onnxruntime has limited support for onnx==1.17 and its default opset.