MaxSim
A scoring function in ColBERT-style late interaction retrieval that computes the relevance score between a query and a document by summing the maximum similarity over their vectors.
Timeline (last 90 days)
ViDoRe adopts a ColBERT-style late interaction mechanism: documents are split into patches, each patch generates a dense vector, queries are likewise encoded as multi-vectors, and scoring is done via MaxSim by summing the per-vector maximum similarities.
ColBERT's core innovation is the MaxSim operation: for each token vector in a query, it finds the most similar token vector among all token vectors in the document, and sums these maximum similarity scores to obtain the final matching score.
All Facts (2)
ViDoRe adopts a ColBERT-style late interaction mechanism: documents are split into patches, each patch generates a dense vector, queries are likewise encoded as multi-vectors, and scoring is done via MaxSim by summing the per-vector maximum similarities.
50%UnverifiedColBERT's core innovation is the MaxSim operation: for each token vector in a query, it finds the most similar token vector among all token vectors in the document, and sums these maximum similarity scores to obtain the final matching score.
50%