Datasets:

DFKI-SLT

/

few-nerd

like 18

Follow

Speech and Language Technology, DFKI 43

I think we had to disable the viewer for this dataset because it was downloading the data from an unreliable host (was too slow or failing all the time)

I agree.

Note that we also have an internal bug, because we should not have this message: error message could not be parsed: ValidationError: "cause_message" is not allowed to be empty.

Yes, unfortunately the data files are hosted by the authors of the FewNERD paper in the Tsinghua university cloud (https://cloud.tsinghua.edu.cn). When testing in the browser, the response time for all 3 download URLs is actually very good, but maybe not consistently?

Since the Few-NERD data is distributed under CC BY-SA 4.0 license, we could just directly host the data in the repository. This should make the download more reliable and faster. Take a look at my pull request #6.

Cool ! I'll re-enable the dataset viewer later today :)

Fixed!

perfect, thank you everyone!