Official repository for Facet-Probe, an evaluation framework for auditing order sensitivity in multimodal large language models. You can find the corresponding pre-print here: https://arxiv.org/abs/2606.26079

Executable audits for latent-state claims in AI alignment and high-stakes evaluation.