Hugging Face Blog·· 2026-08-18
Multi-Vector (Late Interaction) Embedding Models with Sentence Transformers
Multi-Vector (Late Interaction) Embedding Models with Sentence Transformers
AI summary
Sentence Transformers v6.0 adds a fourth model type, MultiVectorEncoder, for ColBERT-style late-interaction retrieval. It directly loads PyLate, Stanford-NLP ColBERT checkpoints and visual document retrieval models from colpali-engine.
Selection record
Threshold 60Official, first-handFirst 62Second 62
AdmittedSum of both 124 ≥ twice the threshold 120
- Source tier
- Official, first-hand; this tier's threshold is 60
- Pre-filter
- passed：介绍多向量嵌入模型及检索用法
- Why it was chosen
- Complete multi-vector retrieval examples and measured indexing costs support decisions about adding late interaction to an existing retrieval stack.
A model scores each item twice, independently, against one written standard, out of 100. An item is admitted only when the two scores add up to twice the threshold. The threshold is set per source tier.
Source: Hugging Face Blog · huggingface.co