Add BEIR-PL datasets to MTEB

[kwojtasi]

Add tasks and updated README for datasets in BEIR-PL (BEIR benchmark in Polish language).

[Muennighoff]

Amazing work!

I would standardize the names to all be e.g. DBPediaPL.
Currently, it seems there's HotpotQAPL, but also DBPedia-pl & ArguAna-PL.

Do you want to merge this first & then lateron add the missing datasets (CQA, Touche etc.) in a separate PR?

[kwojtasi]

I have updated names to ArguAna-PL, it seems to be the most readable option, but if you want other names standardization I can change it.

We can merge it and I will add another PR for all CQADupstack and Touche datasets.

[Muennighoff]

Looks good to me!
Can you run all tasks for at least 1 model and provide the result files here? I will then update the leaderboard to have a tab for BEIR-PL

[kwojtasi]

Hi, I have evaluated "distiluse-base-multilingual-cased-v2" from SentenceTransformers. Attaching results.
results.zip

Also did some minor changes that were required to run the evaluation.

[Muennighoff]

Suggested change

	from .BeIRTask import *
	from .CrosslingualTask import *
	from .MultilingualTask import *
	from .BeIRPLTask import *
	from .BeIRPLTask import *
	from .BeIRTask import *
	from .CrosslingualTask import *
	from .MultilingualTask import *

[Muennighoff]

Looks great, amazing work!
I've added the results in a new leaderboard tab for Polish under Retrieval: https://huggingface.co/spaces/mteb/leaderboard

Feel free to add more models by either sending the result files or adding the results to the model card of the models.

Will merge this if fine with you!