{"slug": "a-four-stage-decomposition-of-word-problem-solving-and-mechanistic-fragility-in", "title": "A Four-Stage Decomposition of Word-Problem Solving and Mechanistic Fragility in LLM Math Reasoning", "summary": "A new arXiv paper (2609.17804v1) reports that large language models solve grade-school math word problems through a four-stage sequential pipeline — Schema Abstraction, Operation Planning, Operand Binding, and Computation — each producing a distinct intermediate representation in an identifiable band of layers. The authors localize distractor-induced failures to a single stage, Operation Planning, implemented by a set of attention heads whose causal role they validated bidirectionally. The work offers a mechanistic account of both LLM math word-problem reasoning and its fragility when irrelevant clauses are inserted.", "body_md": "arXiv:2609.17804v1 Announce Type: new \nAbstract: Large language models solve grade-school math word problems with high accuracy, yet a single irrelevant clause inserted into the problem can collapse it. We reconcile these observations with a mechanistic account. We show that the model's internal computation decomposes into a four-stage sequential pipeline, Schema Abstraction, Operation Planning, Operand Binding, and Computation, each stage producing a distinct intermediate representation in an identifiable band of layers. Using the same scaffold to diagnose distractor-induced failure, we localize the corruption to a single stage, Operation Planning, implemented by a set of attention heads whose causal role we validate bidirectionally. In short, we provide a mechanistic interpretation of math word problem reasoning in LLMs, and their failure when distracted.", "url": "https://wpnews.pro/news/a-four-stage-decomposition-of-word-problem-solving-and-mechanistic-fragility-in", "canonical_source": "https://arxiv.org/abs/2609.17804", "published_at": "2026-09-17 04:00:00+00:00", "updated_at": "2026-09-17 04:24:40.200385+00:00", "lang": "en", "topics": ["large-language-models", "ai-research", "machine-learning", "natural-language-processing"], "entities": ["arXiv", "2609.17804v1"], "alternates": {"html": "https://wpnews.pro/news/a-four-stage-decomposition-of-word-problem-solving-and-mechanistic-fragility-in", "markdown": "https://wpnews.pro/news/a-four-stage-decomposition-of-word-problem-solving-and-mechanistic-fragility-in.md", "text": "https://wpnews.pro/news/a-four-stage-decomposition-of-word-problem-solving-and-mechanistic-fragility-in.txt", "jsonld": "https://wpnews.pro/news/a-four-stage-decomposition-of-word-problem-solving-and-mechanistic-fragility-in.jsonld"}}