An amortized estimator trained on attention features provides real-time faithful visual attributions for multimodal reasoning models, matching the faithfulness of exhaustive causal methods.
where” the model attends but “how much
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Real-Time Visual Attribution Streaming in Thinking Model
An amortized estimator trained on attention features provides real-time faithful visual attributions for multimodal reasoning models, matching the faithfulness of exhaustive causal methods.