Test set languages

[Maud Andruszak]

Hi,

I would like to know how the test will be done considering the languages. Like, what happens if I decide to cover only part of the languages ? And are we going to know the language of each row in the test set ?

Best regards,

Maud Andruszak

[cagri coltekin]

Hi,

We have just released the test set, and updated the submission
information. To answer your questions quickly:

- The format of the test set is very similar to the training set.
There is a separate file for each parliament/task However, as
in the training data, we do not mark the language of the speech
for multilingual parliaments.

- You can cover as many (or little) as the parliament/task pairs.
The prediction files are also expected as separate files for
each parliament (you need to submit a zip archive when
submitting runs). Please see the evaluator [1] for further
description of the submission format, and a toy example.

Best,
Cagri

[1] https://github.com/coltekin/ideology-power-st-baseline/tree/main/evaluation

> --
> You received this message because you are subscribed to the Google Groups "Ideology and Power in Parliamentary Speeches" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to ideology-and-power-in-parlia...@googlegroups.com.
> To view this discussion on the web, visit https://groups.google.com/d/msgid/ideology-and-power-in-parliamentary-speeches/bb40c9aa-d5c6-402a-92ed-e5cca7645396n%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.