Visualizing Graph-to-Answer Mechanism Recovery in Materials-Science Hypothesis Generation
Not on the map. Why?
Abstract (arXiv)
AI co-scientists can generate fluent materials-science hypotheses, but fluency does not show that an answer preserves a scientifically meaningful mechanism. We present a graph-to-answer mechanism-tracing case study for Graph-PRefLexOR-8B, a Qwen3-8B model adapted to expose distinct stages for brainstorming, graph construction, pattern extraction, and synthesis. We organize semantic backtracking, graph corruption, activation-based recovery measurements, and layer-by-token-region grids into a visual diagnostic workflow for inspecting this pathway. Across 100 open-ended materials-science questions, final answers remain closest to the model's own structured stages, especially synthesis. Under graph corruption, a full sweep over 37 residual-stream checkpoints, the embedding output and 36 transformer blocks, shows little mechanism recovery in the earlier transition region at layers 7--10, recovery instead concentrates in late synthesis and answer-start regions around layers 30 and 36. The workflow is intended to help scientists and model developers identify where a generated hypothesis loses or regains mechanism support before it is passed to downstream experimental planning.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 13 of its 16 references that are in the map. First posted on arXiv 4 Aug 2026.
Builds on
13 of 16 references.
References · 13
- 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
- 2022Locating and Editing Factual Associations in GPTcited by 2,691
- 2022Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 smallcited by 1,165
- 2023Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Promptingcited by 1,122
- 2023Measuring Faithfulness in Chain-of-Thought Reasoningcited by 703
- 2023Unifying Large Language Models and Knowledge Graphs: A Roadmapcited by 668
- 2023Towards Best Practices of Activation Patching in Language Models: Metrics and Methodscited by 359
- 2024Is Cosine-Similarity of Embeddings Really About Similarity?cited by 89
- 2025A Survey on Large Language Models with some Insights on their Capabilities and Limitationscited by 42
- 2025Hypothesis Generation for Materials Discovery and Design Using Goal-Driven and Constraint-Guided LLM Agentscited by 30
- 2024IterGen: Iterative Semantic-aware Structured LLM Generation with Backtrackingcited by 29
- 2024Chain-of-Thought Unfaithfulness as Disguised Accuracycited by 28
Built on by
Not cited yet.