Post
463
I'm really out of ideas, so I wanted to ask you. What kind of dataset would you would like to see? What would be useful for your research?
Join the conversation
Join the community of Machine Learners and AI enthusiasts.
Sign Up
I'm not a data analyst, let alone a researcher, but I have the data I want.
- Danbooru tag / plain English corresponding table
- E621 tag / plain English corresponding table
This would probably make a more lightweight and precise tagger for creating image-generating AI models that interpret natural language in English. Whether I myself would build it or not is beside the point.
With the advent of FLUX, the use of natural language has become practically recommended, so I think there is a demand for some people.
Danbooru dataset
https://huggingface.co/datasets/isek-ai/danbooru-wiki-2024
Danbooru / JA
https://huggingface.co/datasets/p1atdev/danbooru-ja-tag-pair-20240715
E621 dataset
?
It really depends on the field of research, but for many, a well-structured, real-world dataset related to economic trends, consumer behavior, or social patterns could be super useful. Personally, I’d be interested in a dataset focused on financial health and credit usage trends, especially post-pandemic. If you’re working with any financial data, American First National Bank’s customer service https://www.pissedconsumer.com/company/american-first-national-bank/customer-service.html might be a good resource—they could provide insight into the types of financial data people are seeking or even connect you with relevant reports.
In this post