[MIEB] Update VLM2Vec model

[XMHZZ2018]

Thanks for the great work on MTEB and MIEB, as well as for including our VLM2Vec!

We have released some new version of VLM2Vec. Could you please let me know how to update it here? I’d be happy to work on this if you could provide me some guidance.

It seems I need to update this script. Please let me know if my understanding is correct: https://github.com/embeddings-benchmark/mteb/blob/mieb/mteb/models/vlm2vec_models.py

Also, it seems the current instruction is "<|image_1|> Represent the given image.", I'm wondering whether MIEB would support task-specific instruction. Thanks!

[KennethEnevoldsen]

Thanks for the interest @XMHZZ2018, we are working on reformulating the MIEB interface (#1551). This reformulation should allow for prompts, multiple images pr. document etc.

cc @gowitheflow-1998