Created by: juncaipeng

[Core ] Update quantization: save scales in op attrs by <inputname_index, scale_value>