Hey,
Great work! I was curious — how did you go about constructing the LLAVA Video 9k subset? Was it simply a random sample of 9k entries from LLAVA Video?
I ask because I noticed that some videos appear multiple times. After filtering for unique videos, I ended up with about 6,674. For the repeated ones, the associated questions seem to differ. So I was wondering if there was any particular selection criterion behind this subset or if it was just a random sampling.
Thanks,
Pulkit
Hey,
Great work! I was curious — how did you go about constructing the LLAVA Video 9k subset? Was it simply a random sample of 9k entries from LLAVA Video?
I ask because I noticed that some videos appear multiple times. After filtering for unique videos, I ended up with about 6,674. For the repeated ones, the associated questions seem to differ. So I was wondering if there was any particular selection criterion behind this subset or if it was just a random sampling.
Thanks,
Pulkit