Pause or Fabricate? Training Language Models for Grounded Reasoning

• URL: http://arxiv.org/abs/2604.19656v1
• Date: Tue, 21 Apr 2026 16:45:29 GMT
• Title: Pause or Fabricate? Training Language Models for Grounded Reasoning
• Abstract summary: We propose Grounded Reasoning via Interactive Reinforcement Learning (GRIL) for grounded reasoning under incomplete information.<n>GRIL decomposes the reasoning process into two stages: clarify and pause, which identifies whether the available information is sufficient, and grounded reasoning, which performs task solving once the necessary premises are established.<n>Experiments on GSM8K-Insufficient and MetaMATH-Insufficient show that GRIL significantly improves premise detection (up to 45%), leading to a 30% increase in task success while reducing average response length by over 20%.
• Score: 50.104657152302956
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models have achieved remarkable progress on complex reasoning tasks. However, they often implicitly fabricate information when inputs are incomplete, producing confident but unreliable conclusions -- a failure mode we term ungrounded reasoning. We argue that this issue arises not from insufficient reasoning capability, but from the lack of inferential boundary awareness -- the ability to recognize when the necessary premises for valid inference are missing. To address this issue, we propose Grounded Reasoning via Interactive Reinforcement Learning (GRIL), a multi-turn reinforcement learning framework for grounded reasoning under incomplete information. GRIL decomposes the reasoning process into two stages: clarify and pause, which identifies whether the available information is sufficient, and grounded reasoning, which performs task solving once the necessary premises are established. We design stage-specific rewards to penalize hallucinations, enabling models to detect gaps, stop proactively, and resume reasoning after clarification. Experiments on GSM8K-Insufficient and MetaMATH-Insufficient show that GRIL significantly improves premise detection (up to 45%), leading to a 30% increase in task success while reducing average response length by over 20%. Additional analyses confirm robustness to noisy user responses and generalization to out-of-distribution tasks.

Related papers

• Knowing When to Quit: Diagnosing and Training LLMs to Abort Futile Reasoning [109.33717747526583]
  Large language models generate computationally expensive yet semantically void reasoning on beyond-capability tasks.<n>We introduce textbfCaRL (textbfCapability-textbfaligned textbfReinforcement textbfL) to align behavior with capability boundaries.<n>Experiments demonstrate a substantial reduction in futile reasoning while preserving performance across task difficulties.
  arXiv  Detail & Related papers  (2026-07-31T09:30:33Z)
• Bridging the Detection-to-Abstention Gap in Reasoning Models under Insufficient Information [24.973717050514207]
  We formalize the detection-to-abstention gap, where detected insufficiency fails to translate into final abstention.<n>This gap is especially concerning in high-risk domains such as medical AI, where incomplete evidence can be more harmful than refusal.<n>We propose Judge-Then-Then (JTS), a trajectory-level reasoning-control framework that trains models to make an explicit answerability commitment.
  arXiv  Detail & Related papers  (2026-05-27T07:28:25Z)
• MixRea: Benchmarking Explicit-Implicit Reasoning in Large Language Models [30.821682013821345]
  We investigate whether large language models (LLMs) fail to attend to subtle yet important contextual cues under explicit task instructions.<n> Evaluation of 21 advanced LLMs shows that even the best-performing reasoning model (Gemini 2.5 Pro) achieves only 42.8% consistency.<n>We propose textbfPotential Relation Completion Prompting (PRCP), a prompting method that improves reasoning by recovering overlooked causal relations.
  arXiv  Detail & Related papers  (2026-05-19T17:15:08Z)
• InfoDensity: Rewarding Information-Dense Traces for Efficient Reasoning [36.37924312794855]
  We argue that verbosity is not merely a length problem, but a symptom of poor intermediate reasoning quality.<n>We propose InfoDensity, a reward framework for RL training that combines an AUC-based reward and a monotonicity reward as a unified measure of reasoning quality.
  arXiv  Detail & Related papers  (2026-03-18T03:11:36Z)
• Illocutionary Explanation Planning for Source-Faithful Explanations in Retrieval-Augmented Language Models [6.517222960194991]
  We study the extent to which an explanation's claims can be grounded in, and traced back to, an explicit source.<n>We benchmark six large language models on 90 Stack Overflow questions grounded in three programming textbooks.<n>We find that non Retrieval-Augmented Generation (RAG) models have median source adherence of 0%, while baseline RAG systems still exhibit low median adherence (22-40%)
  arXiv  Detail & Related papers  (2026-03-16T11:10:29Z)
• Structured Reasoning for Large Language Models [59.215789462977206]
  We propose Structured Reasoning (SCR), a framework that decouples reasoning trajectories into explicit, evaluable, and trainable components.<n>SCR substantially improves reasoning efficiency and self-verification.<n>Compared with existing reasoning paradigms, it reduces output token length by up to 50%.
  arXiv  Detail & Related papers  (2026-01-12T04:04:01Z)
• Abductive Inference in Retrieval-Augmented Language Models: Generating and Validating Missing Premises [0.0]
  We propose a framework that integrates abductive inference into retrieval-augmented LLMs.<n> Experimental results on abductive reasoning and multi-hop QA benchmarks show that our approach improves both answer accuracy and reasoning faithfulness.<n>This work highlights abductive inference as a promising direction for enhancing the robustness and explainability of RAG systems.
  arXiv  Detail & Related papers  (2025-11-06T03:37:24Z)
• From <Answer> to <Think>: Multidimensional Supervision of Reasoning Process for LLM Optimization [62.07990937720985]
  Dimension-level Reward Model (DRM) is a new supervision framework for Large Language Models.<n>DRM evaluates the quality of a reasoning process along three fundamental, complementary, and interpretable dimensions.<n> Experimental results show that DRM provides effective supervision signals, guides the optimization of LLMs and enhances their reasoning ability.
  arXiv  Detail & Related papers  (2025-10-13T14:29:15Z)
• Selection, Reflection and Self-Refinement: Revisit Reasoning Tasks via a Causal Lens [19.316594303998667]
  Reasoning tasks have long been regarded as rigorous benchmarks for assessing the capabilities of machine learning models.<n>We revisit reasoning tasks from a causal perspective, seeking to understand their behavior in latent space.<n>We introduce a framework, called SR$2$, that incorporates the estimated latent variables as feedback into the selection mechanism.
  arXiv  Detail & Related papers  (2025-10-09T13:45:31Z)
• PRELUDE: A Benchmark Designed to Require Global Comprehension and Reasoning over Long Contexts [50.77454873238667]
  We introduce PRELUDE, a benchmark for evaluating long-context understanding through the task of determining whether a character's prequel story is consistent with the canonical narrative of the original book.<n>Our task poses a stronger demand for global comprehension and deep reasoning than existing benchmarks -- as the prequels are not part of the original story, assessing their plausibility typically requires searching and integrating information that is only indirectly related.<n> Experimental results highlight the challenge of our task: in-context learning, RAG and in-domain training with state-of-the-art LLMs, and commercial DeepResearch services, lag behind humans by
  arXiv  Detail & Related papers  (2025-08-13T14:28:25Z)
• Towards Solving More Challenging IMO Problems via Decoupled Reasoning and Proving [48.22540519786074]
  Recent studies show that the informal accuracy exceeds 80% while formal success remains below 8% on benchmarks like PutnamBench.<n>We propose a novel framework that decouples high-level reasoning from low-level proof generation.<n>We evaluate our method on a challenging set of post-2000 IMO problems, a problem set on which no prior open-source prover has reported success.
  arXiv  Detail & Related papers  (2025-07-07T22:38:49Z)

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.