Spaces:
Running
Access to detailed data?
Hi, love this project! Is there any way to access the raw comparison data? A dataframe with model_a, model_b, and winner would already be great, in the ideal case also some information about the prompt. It would be a fantastic resource for research.
Yes, this would be great to have!
Thanks for your interest! Yes, we will release pairwise data next week. Stay tuned!
@sbdzdz
@vishaal27
We have released a version of voting data with model names and the winner model. You can find it here: https://huggingface.co/datasets/WildVision/wildvision-arena-data
For more detailed conversation data, we need to ensure the safety of the text and image prompts before releasing.
Hi, thanks for the amazing work! Any updates for the source of the voting data?