Dataset are scraped from Instagram Public Data. Lots of thanks to creators in Hong Kong on Instagram.
Weighting from 0.6 ~ 0.8. This can be used with other LoRa like KoreanDollLike.
Please use it reasonably.
