Where do you store the corresponding character names in the m3-bench-robot dataset?

Hi, thank you for sharing your dataset.

I try to visualize the m3-bench-robot memory graph, and I can load the memory graphs for each video, and it does have character_mappings (which stores the mapping of faces/voice to characters). Where do you store each character's name?

Do I have to manually find them in the subtitles provided in the dataset?

Please advise.

Inspect the m3-bench-robot dataset files and the per-video memory graphs, focusing on character_mappings and the provided subtitles. Determine whether character names are stored separately or only recoverable from the subtitles, then document the location and expected lookup process for users visualizing the graphs.

Written by the indexing model from the issue text.