Can the author provide the training code?

[Jingchensun]

Hi,

Thanks for Sharing the inference code. Can you share the training code as well? Also, how did you evaluate the different baseline on M2KR, could you also release these evaluation code?

[dcaffo98]

Dear @Jingchensun,

Thank you for your interest in our work. We are working hard for the release of ReT-2, so we do not plan to update this repository in the immediate future.
However, ReT was implemented entirely upon the Hugging Face's Transformers library, and the core training code is just about how to implement the fine-grained late-interaction contrastive loss, which you can find here. It should be easy to integrate it into a Transformers' Trainer.

Best regards,
David