Details of Quantitative comparision in Table 1

[sojinleeme]

I want to know about details of evaluation procedure using MSR-VTT dataset.
How large was the time interval divided so that each video clip was considered two frames of video?

There are many details missing from the paper, and I can't reproduce the main results of Table 1.

I look forward to your quick reply :)