The time

[QWERZPY]

I wish to know the machine configuration and time required for the program. I've run the program on my machine and cost me about more than two hour on one step in eval.py

[gsp2014]

The running time for training NeuInfer on JF17K or WikiPeople is second-level, while it is hour-level for evaluation. Actually, it takes a lot of time to evaluate NeuInfer, since for each element of a test fact, we need to compute a score for each candidate (each element in the set of relations and attributes/entities and attribute values). Thus, eval.py and eval_bi-n.py are implemented in a multi-process manner to alleviate this problem.