KACE: Knowledge-Adaptive Context Engineering for Mathematical Reasoning

• URL: http://arxiv.org/abs/2606.00532v1
• Date: Sat, 30 May 2026 04:51:01 GMT
• Title: KACE: Knowledge-Adaptive Context Engineering for Mathematical Reasoning
• Abstract summary: We introduce Knowledge-Adaptive Context Engineering (KACE)<n>KACE separates storage from usage through difficulty- and domain-based organization.<n>On AIME 2025, KACE achieves 62.2 percent accuracy, a 10.4-point absolute gain over fixed Best-of-5 self-consistency.
• Score: 4.6726884183570645
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Context engineering can improve large language models without updating their weights, but mathematical reasoning exposes a key limitation: feedback accumulated in one growing prompt causes context bloat and limits the amount of learned guidance that can be used. Existing methods often conflate storage, what is learned across runs, with usage, what is included for a particular problem, and therefore inherit this prompt-size ceiling. We introduce Knowledge-Adaptive Context Engineering (KACE), which separates storage from usage through difficulty- and domain-based organization. Offline, a self-reflective learning loop distills training traces into an epistemic tree: a knowledge base of typed cards stratified by problem difficulty and epistemic domain. Each card is assigned to the difficulty-domain node corresponding to the failure from which it originated. At evaluation time, tiered self-consistency with per-tier agreement gates dynamically classifies each problem as easy, medium, or hard. Easy problems exit without retrieved cards, while harder problems retrieve only the matching branch of the tree. This tiered scheme matches or exceeds Best-of-N while using comparable compute, and it classifies problem difficulty with 78 percent pairwise concordance. The main empirical contribution is the construction and use of a difficulty- and domain-stratified knowledge base enabled by tiered self-consistency. On AIME 2025, KACE achieves 62.2 percent accuracy, a 10.4-point absolute gain over fixed Best-of-5 self-consistency at a comparable solver-call budget and a 5.6-point gain over the strongest learned-context baseline, Tiered + GEPA. We also observe consistent gains on MATH-HARD and the verifiable subset of OlymMATH.

Related papers

• NTDH: Complex Reasoning for Comprehensive Affective Analysis [28.280253328311346]
  Comprehensive affective analysis is challenging for two reasons.<n>It spans heterogeneous prediction tasks with continuous, ordinal, and multi-label outputs.<n>We recast the task as a complex-reasoning problem.<n>This is the first such treatment covering both sentiment and emotion.
  arXiv  Detail & Related papers  (2026-08-05T18:59:43Z)
• Scaling Self-Play with Self-Guidance [62.13619253976748]
  Self-Guided Self-Play (SGS) is a self-play algorithm that guides the Conjecturer away from degeneracy.<n>SGS surpasses the solve rate of our strongest baseline in fewer than 80 rounds of self-play.
  arXiv  Detail & Related papers  (2026-04-22T05:50:23Z)
• Omanic: Towards Step-wise Evaluation of Multi-hop Reasoning in Large Language Models [60.418191092851636]
  Omanic is an open-domain multi-hop QA resource that provides decomposed sub-questions and intermediate answers as structural annotations for analyzing reasoning processes.<n>It contains 10,296 machine-generated training examples (Omanic Synth) and 967 expert-reviewed human-annotated evaluation examples (OmanicBench)
  arXiv  Detail & Related papers  (2026-03-17T15:23:37Z)
• GONE: Structural Knowledge Unlearning via Neighborhood-Expanded Distribution Shaping [4.740962650068887]
  This paper introduces Graph Oblivion and Node Erasure (GONE), a benchmark for evaluating knowledge unlearning over structured knowledge graph (KG) facts in Large Language Models (LLMs)<n>This KG-based benchmark enables the disentanglement of three effects of unlearning: direct fact removal, reasoning-based leakage, and catastrophic forgetting.<n>NEDS, a novel unlearning framework, is designed to leverage graph connectivity and identify anchor correlated neighbors, enforcing a precise decision boundary between the forgotten fact and its semantic neighborhood.
  arXiv  Detail & Related papers  (2026-02-21T04:08:37Z)
• DART: Difficulty-Adaptive Reasoning Truncation for Efficient Large Language Models [36.962276192354174]
  textbfDART adjusts thinking length according to problem difficulty.<n>textbfTruncation framework learns when to stop thinking''
  arXiv  Detail & Related papers  (2025-11-03T02:41:20Z)
• Eigen-1: Adaptive Multi-Agent Refinement with Monitor-Based RAG for Scientific Reasoning [53.45095336430027]
  We develop a unified framework that combines implicit retrieval and structured collaboration.<n>On Humanity's Last Exam (HLE) Bio/Chem Gold, our framework achieves 48.3% accuracy.<n>Results on SuperGPQA and TRQA confirm robustness across domains.
  arXiv  Detail & Related papers  (2025-09-25T14:05:55Z)
• CAMA: Enhancing Mathematical Reasoning in Large Language Models with Causal Knowledge [14.367146529900609]
  Large Language Models (LLMs) have demonstrated strong performance across a wide range of tasks, yet they still struggle with complex mathematical reasoning.<n>We propose textbfCAusal textbfMAthematician (textbfCAMA), a two-stage causal framework that equips LLMs with explicit, reusable mathematical structure.
  arXiv  Detail & Related papers  (2025-08-04T16:39:24Z)
• ConciseRL: Conciseness-Guided Reinforcement Learning for Efficient Reasoning Models [14.403953640255823]
  We introduce a novel score used as a reward signal within a reinforcement learning framework to guide models toward generating correct and concise reasoning traces.<n>This score is evaluated by a large language model acting as a judge, enabling dynamic, context-aware feedback beyond simple token length.<n>Our method achieves state-of-the-art efficiency-accuracy trade-offs on the MATH dataset, reducing token usage by up to 31x on simple problems while improving accuracy by 7%, and on the hardest problems, it outperforms full reasoning by +7.5% accuracy with up to 3.6x fewer tokens.
  arXiv  Detail & Related papers  (2025-05-22T19:56:35Z)
• Ineq-Comp: Benchmarking Human-Intuitive Compositional Reasoning in Automated Theorem Proving on Inequalities [46.111273938884295]
  We study provers' ability to recognize that the given problem simplifies by applying a known inequality such as AM/GM.<n>Although these problems remain easy for humans, we find that most provers -- including Goedel, rewriting, and Kimina-7B -- struggle significantly.<n>Our results expose a persisting gap between the behavior of current AI provers and human intuition.
  arXiv  Detail & Related papers  (2025-05-19T03:56:05Z)
• MATH-Perturb: Benchmarking LLMs' Math Reasoning Abilities against Hard Perturbations [90.07275414500154]
  We observe significant performance drops on MATH-P-Hard across various models.<n>We also raise concerns about a novel form of memorization where models blindly apply learned problem-solving skills.
  arXiv  Detail & Related papers  (2025-02-10T13:31:46Z)
• Harnessing Large Language Models for Knowledge Graph Question Answering via Adaptive Multi-Aspect Retrieval-Augmentation [81.18701211912779]
  We introduce an Adaptive Multi-Aspect Retrieval-augmented over KGs (Amar) framework.<n>This method retrieves knowledge including entities, relations, and subgraphs, and converts each piece of retrieved text into prompt embeddings.<n>Our method has achieved state-of-the-art performance on two common datasets.
  arXiv  Detail & Related papers  (2024-12-24T16:38:04Z)
• Konstruktor: A Strong Baseline for Simple Knowledge Graph Question Answering [60.6042489577575]
  We introduce Konstruktor - an efficient and robust approach that breaks down the problem into three steps. Our approach integrates language models and knowledge graphs, exploiting the power of the former and the interpretability of the latter. We show that for relation detection, the most challenging step of the workflow, a combination of relation classification/generation and ranking outperforms other methods.
  arXiv  Detail & Related papers  (2024-09-24T09:19:11Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.