A collaborative parametric knowledge calibration framework for retrieval-augmented KB-VQA enables bidirectional knowledge sharing between retriever and generator, yielding a 4.7% accuracy gain and 7.5% boost to base MLLMs via late interaction and reflective answering.
Ross, and Alireza Fathi
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Enabling Collaborative Parametric Knowledge Calibration for Retrieval-Augmented Vision Question Answering
A collaborative parametric knowledge calibration framework for retrieval-augmented KB-VQA enables bidirectional knowledge sharing between retriever and generator, yielding a 4.7% accuracy gain and 7.5% boost to base MLLMs via late interaction and reflective answering.