mlsanityData to 14 Oct 2026

Search, Fail, Recover: A Training Framework for Correction-Aware Reasoning

arXiv 2607.07492 ↗First posted 8 Jul 2026cs.AIcited by 1

Not on the map. Why?

Abstract (arXiv)

Many reasoning tasks are not well described by a single left-to-right chain: a solver may need to pursue a plausible branch, observe delayed failure, and return to the latest prefix that can still be completed. We introduce Pyligent, a training and inference framework inspired by the Diligent Learner formulation that represents reasoning as validated search over partial solution chains. A task validator labels generated continuations and failures, and the resulting search trees are converted into supervised targets for three actions: continue, finish, and backtrack, with optional traces that summarize abandoned branches. We evaluate Pyligent on a hidden directed graph task designed to isolate delayed-failure recovery, and on structured reasoning domains with exact validators, including 4×4 Sudoku, Sudoku with reasoning traces, and Blocksworld. Compared with gold-only supervised fine-tuning, Pyligent improves solve rate by 72.7 percentage points on hidden graphs, by 17 and 18 points on mixed and expert Sudoku, by 27 and 14 points on mixed and expert Sudoku with reasoning traces, and by 13 points on Blocksworld. These results suggest that explicit failed-branch supervision can teach useful recovery behavior beyond imitation of polished solution chains.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 20 of its 24 references that are in the map. First posted on arXiv 8 Jul 2026.

Builds on

20 of 24 references.

References · 20

  1. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  2. 2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 4,276
  3. 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Modelscited by 3,211
  4. 2025Kimi k1.5: Scaling Reinforcement Learning with LLMscited by 1,210
  5. 2023Large Language Models Cannot Self-Correct Reasoning Yetcited by 1,017
  6. 2023Reasoning with Language Model is Planning with World Modelcited by 861
  7. 2023CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquingcited by 783
  8. 2024Training Language Models to Self-Correct via Reinforcement Learningcited by 426
  9. 2025Cognitive Behaviors that Enable Self-Improving Reasoners, or, Four Habits of Highly Effective STaRscited by 377
  10. 2024When Can LLMs Actually Correct Their Own Mistakes? A Critical Survey of Self-Correction of LLMscited by 240
  11. 2024LLMs Can't Plan, But Can Help Planning in LLM-Modulo Frameworkscited by 179
  12. 2024Recursive Introspection: Teaching Language Model Agents How to Self-Improvecited by 165
8 more
  1. 2024Toward Self-Improvement of LLMs via Imagination, Searching, and Criticizingcited by 125
  2. 2024Beyond A*: Better Planning with Transformers via Search Dynamics Bootstrappingcited by 90
  3. 2024Dualformer: Controllable Fast and Slow Thinking by Learning with Randomized Reasoning Tracescited by 43
  4. 2025Step Back to Leap Forward: Self-Backtracking for Boosting Reasoning of Language Modelscited by 27
  5. 2025To Backtrack or Not to Backtrack: When Sequential Search Limits Model Reasoningcited by 24
  6. 2025How Much Backtracking is Enough? Exploring the Interplay of SFT and RL in Enhancing LLM Reasoningcited by 13
  7. 2025ASTRO: Teaching Language Models to Reason by Reflecting and Backtracking In-Contextcited by 13
  8. 2025From Reasoning to Super-Intelligence: A Search-Theoretic Perspectivecited by 9

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026MedEvo: Medical Reasoning Evolution through Knowledge Navigationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Not Too Hard, Not Too Easy: Learning from Intermediate States for LLM Structured Reasoning
  2. 2025Liquid Reasoning Transformers: A Sudoku-Based Prototype for Chess-Scale Algorithmic Tasks
  3. 2026Quantizing Recursive Reasoning Models
  4. 2025Teaching Transformers to Solve Combinatorial Problems through Efficient Trial & Error
  5. 2024Causal Language Modeling Can Elicit Search and Reasoning Capabilities on Logic Puzzles
  6. 2026MaxSAT-Based Feedback for Guiding Vision-Language Models in Sudoku
  7. 2026Boosting Inference with Guided Reasoning: Stochastic Exploration for Recursive Models
  8. 2026Reasoning on the Simplex: Geometric Fixed-Point Models