Tag
This paper introduces MoganColBERT-TR, a late-interaction multi-vector retrieval model for Turkish, which achieves competitive zero-shot performance on Turkish BEIR datasets through distillation training from previous encoders.