Clarification on the R@1_sum Metric in Table 1

[hwan-sig]

I have a question regarding the R@1_sum metric reported in Table 1 of the paper. At first glance, it seems that R@1_sum might represent the sum of i2t_R@1 and t2i_R@1, but based on the reported numbers, that doesn't appear to be the case.

For example, for CreamFL:

i2t_R@1 = 49.66

t2i_R@1 = 38.94
→ Their sum is 88.60, but R@1_sum is reported as 132.88.

Could you clarify how the R@1_sum is actually computed? Is it a typo, or does it represent a different aggregation?

Thanks in advance for the clarification!