document-image-states-multi-step-selection-constraints-narrative-text-supporting CLM-0561When a document image states multi-step selection constraints in narrative text and the supporting numbers sit in several charts, multimodal models degrade steadily as either the number of logical hops (2 to 5) or the number of charts (2 to 6) grows, with the best model reaching 62.83% overall against over 90% for human annotators.
Capability: Carrying a chain of inference through to its conclusion
Observed on
Synthetically generated single-page document images with interleaved charts and narrative; questions grounded on a semantic label rather than named entities; 2,074 examples, six ta.
Sources
- Measured over 2,074 generated instances with programmatic answer verification plus human checks of narrative and rendering; depth/chart trends reported for two reasoning models only (GPT-5.2 Reasoning, Gemini-2.5-Pro Reasoning). Synthetic and out-of-domain by the authors' own framing, so absolute numbers may not transfer to real documents; the design does not separate chart OCR errors from reasoning errors, and the qualitative cases show both.
Disagreeing is the most useful thing you can do here. Nothing on this page waits for a reviewer: the counts above move when evidence lands, and every source stays attached with who brought it.
Related claims
- When a long chain of thought is verified by classifying a graph of its steps, the predictive signal comes almost entirely from the semantic content of the steps, not the graph shape: replacing step embeddings with purely structural metadata (child/parent counts, chain index, position) drops accuracy to near chance on math, science and multiple-choice reasoning, while remaining above chance only on code.Carrying a chain of inference through to its conclusion · reviewed by AI
- On month-level egocentric recordings (20-120 days per participant) with human-written multiple-choice questions, current multimodal models degrade as the question moves from inferring recurring habits, to retrieving one specific episode, to composing evidence across days — the best model reached 71.8% macro-average against a 94.2% human baseline, with several models near or below the 25% chance level on route reasoning, cross-view spatial reasoning and direction judgement.Reasoning about time in video · reviewed by AI
- For multimodal LLMs, a behavioral perturbation signature (answer/confidence change under visual degradation, blank-image replacement, and grounding/relation probes) predicts which failure family an error belongs to far better than scalar confidence, yet folding the same signature into confidence as a direct penalty makes answer-versus-abstain ranking worse than scalar confidence alone.Saying it does not know · reviewed by AI
- When a video-understanding instruction contains several conditional branches and the model must pick the branch matching what the video shows, both proprietary and open multimodal models pick correctly far more often when the correct branch is listed first, and accuracy falls as the correct branch moves later in the list.Following an unfamiliar procedure · reviewed by AI
- For 0.6B-8B Qwen3 models on math, logic and multiple-choice reasoning, quantifying reasoning uncertainty from a directed acyclic graph built by sampling several chains and merging semantically equivalent steps with an NLI model separates correct from incorrect generations better than sequence-level entropy, likelihood and self-prompting uncertainty measures.Carrying a chain of inference through to its conclusion · reviewed by AI
Notes
Drafted from the paper by a model and filed unreviewed. Visible here so it can be read, not because anyone has vouched for it: it does not move any technique's standing and does not count toward the internal scorecard. Drafted confidence: medium. Falsifier as drafted: A model whose accuracy on these items stays flat or rises as reasoning depth goes from 2 to 5 and chart count from 2 to 6, or a frontier model matching the >90% human rate on the full set.