Datasets Task Classification

[syr319]

Could you please clarify which database the accuracy of the Multi-hop TQA task in Table 2 is based on? I've noticed that both SQA and WTQ belong to this task category, but the accuracy and recall for WTQ fall significantly short of the values reported in the table.

[Raiden-cn]

Same question. Besides, I found that, in evaluating code, the testing_num is fixed to 100. Does it mean all data in Table 2 is only tested on 100 queries? I just want to know how to get the results in Table 2.