Evaluation Procedure for Reporting Performance

[willyfh]

In your paper, you conducted experiments for 5 epochs. In reference to this issue (#36), it is mentioned that you reported performance based on the best scores on the validation set. Could you elaborate further on which metric was used for this evaluation? Was it the recall@1 of the text-to-video metric? Additionally, in which epoch did you achieve the best score? In some cases, I noticed that the performance was already optimal after only the first epoch.