--- title: "Training and Finetuning Multi-Vector Embedding Models with Sentence Transformers" url: https://daily.dev/posts/training-and-finetuning-multi-vector-embedding-models-with-sentence-transformers-8yshfdoae source_url: https://huggingface.co/blog/train-multi-vector-encoder type: article source: "Hugging Face" published: 2026-08-26T13:51:30.579Z updated: 2026-08-26T14:22:44.435Z tags: ["python", "embeddings"] reading_time: 25 upvotes: 0 comments: 0 language: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Training and Finetuning Multi-Vector Embedding Models with Sentence Transformers **[Hugging Face](https://daily.dev/sources/huggingface)** · 25 min read · 0 upvotes · 0 comments ## Summary Sentence Transformers v6.0 introduces MultiVectorEncoder, a new model type for ColBERT-style late-interaction retrieval, along with a full training pipeline. The post walks through every training component (model, dataset, loss, training arguments, evaluator, trainer) and documents a real domain finetune on 4.4M medical question-passage pairs, producing a model that outperforms over 50 general-purpose dense, sparse, lexical, and multi-vector retrievers on a medical benchmark. Key findings include that unsupervised pre-finetuning checkpoints adapt better than fully-finished ones, that lifting document length caps to match long passages is critical, and that quantized multi-vector indexes can be smaller than dense embeddings while scoring higher. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is the MultiVectorEncoder in Sentence Transformers and what version introduced it? MultiVectorEncoder is a new model type added in Sentence Transformers v6.0 for training ColBERT-style late-interaction retrieval models, which keep one vector per token instead of compressing a document into a single embedding, scored using the MaxSim operator. It comes with a complete training pipeline covering models, datasets, losses, training arguments, evaluators, and a trainer class. _See how daily.dev tracks library releases like this for teams evaluating retrieval architectures._ ### Should I start finetuning a ColBERT-style multi-vector model from an unsupervised checkpoint or a fully finetuned one? Start from an unsupervised, pre-supervised checkpoint rather than a fully finished one. In tests across two model families, unsupervised checkpoints like lightonai/mLateOn-unsupervised adapted far better to a new domain and overtook their finished siblings despite starting lower, while fully finished checkpoints barely moved or regressed at every learning rate tried. _Developers choosing a starting checkpoint for domain retrieval can compare approaches on daily.dev._ ### How much does document length truncation hurt retrieval accuracy for long passages in ColBERT-style models? Truncating long documents to the common 180-512 token caps used by classic ColBERT and dense checkpoints can cost up to 0.24 NDCG@10 on passages averaging 941 tokens, more than any difference between model architectures. Lifting the document length cap improved every multi-vector model tested by 0.08 to 0.24 NDCG@10, and even a dense model gained 0.03 from the same treatment. _Track retrieval tuning trade-offs like document length caps alongside other embedding research on daily.dev._ ## Similar posts on daily.dev - [Training and Finetuning Multimodal Embedding & Reranker Models with Sentence Transformers](https://daily.dev/posts/training-and-finetuning-multimodal-embedding-reranker-models-with-sentence-transformers-zdk7kc4g6) · Hugging Face · 0 upvotes · 0 comments --- Tags: [#python](https://daily.dev/tags/python), [#embeddings](https://daily.dev/tags/embeddings) [View this post on daily.dev](https://daily.dev/posts/training-and-finetuning-multi-vector-embedding-models-with-sentence-transformers-8yshfdoae)