Hugging Face Blog·· 7d agoAI score40
Sentence Transformers' Token Pooling Awareness Training: Enhancing Performance of Multi-vector Embedding Models
Token Pooling-Aware Training for Multi-Vector Embedding Models
AI digest
Sentence Transformers introduces Token Pooling for consciousness training, reducing the representation size by merging similar word vectors, while maintaining 97.8% of the information content, and reducing the document vectors to 1/4 of the original size, effectively improving the retrieval performance of multi-vector embedding models.
Source:Hugging Face Blog · huggingface.co