Hi, how to understand the LF-hdvila-8m? #38
Description
Is the line in 'lfvila8m_clipid.jsonl' a video clips-sentence pair? And I see an variational number of video-clips per row. So how the video-clips of 'lfvila8m_clipid.jsonl' is divided from the original ‘hdvila_clip_text_100m.jsonl’？ In addition to the selection of videos with more than 4 clips mentioned in the paper, are there any details?
