Towards Reliable VLM Judges: State-Conditional Invariance and Presentation-Aware Diagnostics
· arxiv.org · popularity 6/100 · 0 similar stories today · Machine Learning Research (Artificial Intelligence)
Read the original at arxiv.org →
Summary
arXiv:2611.13596v1 Announce Type: new Abstract: Vision-language models (VLMs) are increasingly used as automated judges for multimodal and embodied tasks, yet aggregate accuracy or consistency alone does not reveal whether a judge changes its verdict for the right reason. We propose State-Conditional Invariance (SCI), a reliability principle requiring a judge to remain stable under semantics-preserving changes in presentation while responding to causal interventions that change the task outcome. We instantiate SCI in GridWM-Judge, a simulator-grounded diagnostic benchmark built from deterministic MiniGrid trajectories with aligned Full, evidence-removal, and counterfactual variants, plus presentation and representation probes. Across five diagnostic tasks, GridWM-Judge tests local perception, action-conditioned transitions, presentation stability, representation exchangeability, and Full-CF causal discrimination. Our diagnostic evaluations reveal heterogeneous failure modes, including sensitivity to evidence presentation and encoding, low activation on successful trajectories, and weak paired counterfactual discrimination. Perception-controlled analyses further indicate that visual grounding and objective recovery are substantial, model-dependent bottlenecks. These results motivate jointly reporting correctness, presentation stability, and causal sensitivity instead of relying on a single score.
Found in: ArXiv CS.AI. Rating basis: single outlet; outlet tier 2 (unlisted); 1 feeds; feed position 63.