Spaces:

SeaEval

/

SeaEval_Leaderboard

like 4

Running

Not really prepared. One possibility is to convert 'all_results.json' as CSV.

Not really prepared. One possibility is to convert 'all_results.json' as CSV.

Thanks for your reply. I checked the format of 'all_results.json' but found it is not directly the results showing up in the leaderboards. I have no idea calculating the metrics on my own based on the raw evaluations. Is that possible to export them via the app.py?

I would suggest deriving from the JSON file as it contains all the information.

The zero shot results are the Median value of five runs.

I do not think it is easily convertible from app.py as they are not in any CSV format but more like a Dict.

I would suggest deriving from the JSON file as it contains all the information.

The zero shot results are the Median value of five runs.

I do not think it is easily convertible from app.py as they are not in any CSV format but more like a Dict.

Is there any possibility to convert them into dataframe? then I can transform them further into CSV?

Yes. You can do it.

Should be easily convertable.