How are the caption generated?
#2
by
guang-chen
- opened
I notice that the captions in this dataset are not ASR-generated as in the original HD-VILA-100M dataset. They look like texts generated by some image captioner or video captioner. What the specific model is used?