Closed hefeng1995 closed 3 years ago
The configurations to reproduce the results reported in the paper are available here.
What I want to ask is the different divisions of the same data set. For example, the MSRVTT data set, a variety of division have been experimented in the paper, but there is only one division in the open source code.
Indeed, in the paper we have reported results on both 1k-A and 1k-B MSRVTT splits. You can reproduce the results for the 1k-A split using this config. You can reproduce the results for the 1k-B split using this config.
Could you provide configs of different data sets, which will make it easier to reproduce the results.