Traditional vector search methods typically employ sentence embeddings to locate similar content. However, generating sentence embeddings through pooling token embeddings can potentially sacrifice fine-grained details present at the token level. ColB...