AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

VERDICT: Training-Free Step-Wise Verification of Multimodal Reasoning via Disagreement-Aware Consensus

Rohit Sinha, Kunal Tilaganji, Tanuja Ganu, Nagarajan Natarajan, Amit Sharma, Vineeth N Balasubramanian

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Chain-of-thought multimodal reasoning · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use disagreement-aware consensus verification (VERDICT) over unverified base modelQwen2.5-VL-7B-Instruct · six benchmarks averageaccuracy (%): 66.31 -> 70.15Improved
use VERDICT consensus instead of Mean aggregation of the same three agentsVERDICT · six benchmarksaccuracy (%): +0.68 to +2.57Improved
use VERDICT consensus instead of Variance filteringVERDICT · six benchmarks averageaccuracy (%): 67.65 -> 70.15Improved
use frozen consensus verifier instead of trained domain-specific critic (Sherlock)Qwen2.5-VL-7B-Instruct · CV-Bench-3Daccuracy (%): drops 18.26 points on CV-Bench-3DHurt
add trained critic VisionSR1 to the base modelQwen2.5-VL-7B-Instruct · CV-Bench-3Daccuracy (%): drops 22.21 points on CV-Bench-3D and 18.22 points on BLINKHurt
add intelligent ranking (selection) to the filter (ablation)VERDICT · six benchmarksaccuracy (%): -1.17 to -2.17 when removedImproved
add rejection filtering to ranking (ablation)VERDICT · six benchmarksaccuracy (%): -0.26 to -1.08 when removedImproved
use coupled consensus scores instead of raw average scores in dual criterion (ablation)VERDICT · six benchmarksaccuracy (%): 2.59-2.95 pointsImproved
set Visual agent stubbornness highest rather than swapping V with Contextual (ablation)VERDICT · 3DSRBenchaccuracy (%): 59.02 -> 57.26Improved
keep Logical above Contextual stubbornness (undo L-C swap) (ablation)VERDICT · 3DSRBenchaccuracy (%): 59.02 -> 58.61Improved
relax confidence threshold tau to <=0.01 (disable filter) (ablation)VERDICT · 3DSRBenchaccuracy (%): 59.02 -> 68.71Improved
relax confidence threshold tau to <=0.01 (disable filter) (ablation)VERDICT · other five benchmarksaccuracy (%): +0.90 (AI2D) to +1.65 (MMStar) gain lostHurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy