Proposal: Release of Persian MTEB v2 - Should It Replace v1?

[mehran-sarmadi]

We're planning to release version 2 of the Persian MTEB soon, which includes the following changes:

Removal of low-quality data and examples with unreasonable results

Addition of new high-quality data

Use of a hard-negative version of large retrieval datasets, so that testing is both faster and more accessible for others, without requiring heavy hardware

Now the question is:
Should this replace version 1, or, like the English version, should version 1 go into legacy and this new one take its place as the main benchmark?

[Samoed]

Should this replace version 1, or, like the English version, should version 1 go into legacy and this new one take its place as the main benchmark?

Yes, it should be like English