CLIRFeb 18

ColBERT-Zero: To Pre-train Or Not To Pre-train ColBERT models

arXiv:2602.16609v12 citationsh-index: 6
Originality Highly original
AI Analysis

This addresses the efficiency and performance trade-offs in training multi-vector retrieval models for information retrieval tasks, offering a more cost-effective approach.

The paper tackles the problem of whether to pre-train multi-vector models like ColBERT, showing that large-scale multi-vector pre-training yields stronger models, with ColBERT-Zero outperforming existing models and setting new state-of-the-art results for its size.

Current state-of-the-art multi-vector models are obtained through a small Knowledge Distillation (KD) training step on top of strong single-vector models, leveraging the large-scale pre-training of these models. In this paper, we study the pre-training of multi-vector models and show that large-scale multi-vector pre-training yields much stronger multi-vector models. Notably, a fully ColBERT-pre-trained model, ColBERT-Zero, trained only on public data, outperforms GTE-ModernColBERT as well as its base model, GTE-ModernBERT, which leverages closed and much stronger data, setting new state-of-the-art for model this size. We also find that, although performing only a small KD step is not enough to achieve results close to full pre-training, adding a supervised step beforehand allows to achieve much closer performance while skipping the most costly unsupervised phase. Finally, we find that aligning the fine-tuning and pre-training setups is crucial when repurposing existing models. To enable exploration of our results, we release various checkpoints as well as code used to train them.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes