Do Multimodal Models Decide Consistently on Equivalent Optical Spectra?
cs.LG, cs.CL, physics.optics
Submitted: 2026-10-30
Updated: 2026-10-30
Comments: 20 pages, 4 figures, 11 tables. Code and data: https://github.com/zwang-devv/spectra-decision-audit
Code: https://github.com/zwang-devv/spectra-decision-audit
License: http://creativecommons.org/licenses/by/4.0/
The gist: A spectrum can be given to a model as a table in nanometers or as a plot in photon energy, and the decision should not depend on which one it sees.
Terminology
Abstract
A spectrum can be given to a model as a table in nanometers or as a plot in photon energy, and the decision should not depend on which one it sees. When the data needed for an answer are missing, the model should say so. We test representation invariance and abstention on measured optical spectra of silicon metasurfaces (MOCLIP) and gold nanoparticles, with labels computed from the data. Each spectrum is shown as four tables (nm, μ m, scientific notation, eV) and two plots (nm, eV), and partial spectra remove data inside or outside the queried range. We evaluate the decision models d3-nano and Jev 1.13, three GPT-6 Luna configurations and the Qwen3.5-2B base model. Changing only the representation moves accuracy by up to 41 percentage points within one model, mostly on photon-energy tables and plots. With the same evidence, asking in photon energy instead of wavelength helps some model-view pairs by 16-50 points and hurts others by up to 25. When the relevant extremum is removed, unadapted models rarely answer "undetermined". LoRA fine-tuning of d3-nano improves comparisons on the trained table formats, but much of its gain on unanswerable questions comes from a shortcut: it abstains whenever the input is truncated. Training with answerable truncations avoids the shortcut in all six seeds, with a 90-100 point gap between abstention after inside and outside cuts on held-out structures; with the same amount of extra original-task data, five of six seeds keep the shortcut. On gold spectra the two training sets perform similarly. Multimodal language models used for scientific decisions should be tested on equivalent representations and on missing evidence, not only on average accuracy. We release the tasks, per-response results and analysis code.
Sources
- Answerability in Retrieval-Augmented Open-Domain Question Answering
- Labels Override Definitions in Jev-Style Typed Decision Models
- Jev for Scientific Decisions: Evaluating Semantic Choices and Their Consequences
- IsoBench: Benchmarking Multimodal Foundation Models on Isomorphic Representations
- LoRA: Low-Rank Adaptation of Large Language Models
- Sufficient Context: A New Lens on Retrieval Augmented Generation Systems
- ChartQAPro: A More Diverse and Challenging Benchmark for Chart Question Answering
- SpecVQA: A Benchmark for Spectral Understanding and Visual Question Answering in Scientific Images
- TrustDABench: Benchmarking Reliability and Robustness of LLMs for Structured Data Analysis
- Large Language Models Are Not Robust Multiple Choice Selectors
- From Text Decisions to Pixels: An Study of Jev-Style Visual Choice Model
Related papers
- A Physics-Regulated Neural Framework for Learning 3D Grain Growth Dynamics
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis