Datasets:

Modalities:
Text
Formats:
json
Languages:
English
Size:
< 1K
ArXiv:
Libraries:
Datasets
Dask
License:

The image corresponding to the annotation file

#3
by SiyuanH - opened

Hi, thanks for your sharing.

In the JSON file, the "raw_html" is provided, however, when I try to render it, it already loses much spatial and visual information. Would there be some screenshots for the annotated websites?

I noticed that you mentioned that there would be some files recording the annotation traces and the original website is recorded, can those videos also be shared?

OSU NLP Group org

Thanks for your interest in our work.
You can find the raw dump with full traces and snapshots in our GitHub repo: https://github.com/OSU-NLP-Group/Mind2Web#raw-dump-with-full-traces-and-snapshots
We will also release a version of the dataset where each action is paired with a webpage screenshot soon.

Cool!

We will also release a version of the dataset where each action is paired with a webpage screenshot soon.

I am looking forward to this collection! Please do keep us informed and updated!

SiyuanH changed discussion status to closed

Hi @boyuanzheng010

Is there any update on the new version of the dataset, where each action is paired with a webpage screenshot soon.

SiyuanH changed discussion status to open
OSU NLP Group org

Hi @SiyuanH , this version is about to finish. We will release this along with our follow-up work. We will update in this channel once it is released.
Thanks for your patience.

Hi @boyuanzheng010

Is there any update for the multimodal datasets? Can we access to them now?

Bests

Sign up or log in to comment