Closed zkaiWu closed 2 years ago
If I remember correctly, the none
relationship may be needed in single predicate prediction but not needed in multiple predicate estimation.
The data loader should handle both bases automatically. Further changes may not be needed.
According to the code, I think the relationships.txt shouldn't contain 'none'. Is that right? But the relationships.txt I download using preparation.sh contain the 'none' relation. Should I delete none in relationships.txt after run gen_data_gt.py