Fine-tune ColBERT using teacher scores

Thanks for making this framework!

I'd like to know if it is possible to fine-tune e.g. the answerai colbert v1 small model using a teacher model and its scores' instead of positive and negative documents?

The issue names no files, tests, or entry points, so first inspect the training-related parts of the repository and determine whether teacher scores are supported. Done would require a documented, tested decision about fine-tuning with teacher scores instead of positive and negative documents.

Written by the indexing model from the issue text.