Shattering the Autoregressive Curse: Dynamic Epistemic Entropy Orchestrated Erasable Reinforcement Learning for LLMs

• URL: http://arxiv.org/abs/2606.17735v1
• Date: Tue, 16 Jun 2026 09:55:45 GMT
• Title: Shattering the Autoregressive Curse: Dynamic Epistemic Entropy Orchestrated Erasable Reinforcement Learning for LLMs
• Abstract summary: We propose erasable reinforcement learning for long-horizon logical reasoning.<n>$textE3textRL$ eliminates reliance on external signals by grounding the model's endogenous local autoregressive cross-entropy.<n>We train $textE3textRL$ on the DeepMath-103k dataset.
• Score: 21.321550377588427
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Although reinforcement learning (RL) has expanded the cognitive boundaries of large language models (LLMs), it often remains vulnerable to the autoregressive curse in long-horizon logical reasoning: small epistemic perturbations introduced early in generation can propagate irreversibly along the Markov decision process flow, triggering cascading failures that drive the reasoning trajectory toward collapse. To overcome this autoregressive cascade, in which a single early mistake can compromise all subsequent reasoning steps, we propose dynamic epistemic entropy orchestrated erasable reinforcement learning ($\text{E}^3\text{RL}$). $\text{E}^3\text{RL}$ eliminates reliance on external signals by grounding the model's endogenous local autoregressive cross-entropy as an intrinsic coordinate of epistemic uncertainty. By introducing segment-level adaptive dynamic thresholds and advantage allocation, $\text{E}^3\text{RL}$ enables the model to precisely excise localized logical defects while reusing historical key-value (KV) cache streams, thereby endowing the reasoning process with a self-healing capability. We train $\text{E}^3\text{RL}$ on the DeepMath-103k dataset. Experimental results show that $\text{E}^3\text{RL}$ reshapes the exploration efficiency of long-sequence reasoning and improves sample efficiency while maintaining linear memory overhead. On mathematical reasoning benchmarks such as AIME, $\text{E}^3\text{RL}$ achieves substantial performance gains, with the 4B and 8B parameter models surpassing previous state-of-the-art (SOTA) results by 5.349\% and 6.514\%, respectively. These findings suggest that $\text{E}^3\text{RL}$ shatters the autoregressive curse in long-sequence reasoning and establishes a theoretical and systems-level foundation for the next generation of self-healing artificial general intelligence (AGI).

Related papers

• Boosting LLM Exploration via Weak-Model Guidance in RLVR [36.33733929880283]
  We propose a simple yet effective approach to preserve the generative diversity of LLMs during RLVR.<n>Instead of relying solely on internal exploration, we force the target model to generate answers based on partial reasoning trajectories generated by a smaller, weaker language models.<n>Our approach efficiently mitigates entropy collapse without requiring additional SFT, intricate reward designs, or complex prompting.
  arXiv  Detail & Related papers  (2026-08-27T17:45:50Z)
• 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)
• Depth-Entropy Guided Sampling for Training-Free LLM Reasoning [7.057431471793534]
  Reinforcement learning (RL) has become the dominant paradigm for improving the reasoning capabilities of large language models.<n>We introduce Depth-Entropy Guided Sampling (DEGS), a training-free, test-time method that exploits layer-wise entropy collapse as an intrinsic quality signal.
  arXiv  Detail & Related papers  (2026-06-19T12:36:23Z)
• Relative Kinetic Utility for Reasoning-Aware Structural Pruning in Large Language Models [0.0]
  We propose a novel framework that elevates discrete pruning to a continuous kinetic integral over the depth manifold of a model.<n>Experiments on Qwen-2.5-7B and LLaMA-3-8B improve performance in the high-sparsity regime around 40%.<n>RKU attains 13.34% accuracy on GSM8K at 40% sparsity, outperforming the strongest baseline.
  arXiv  Detail & Related papers  (2026-05-09T15:47:53Z)
• Mitigating Reward Hacking in RLHF via Bayesian Non-negative Reward Modeling [49.41422138354821]
  We propose a principled reward modeling framework that integrates non-negative factor analysis into the Bradley-Terry preference model.<n>BNRM represents rewards through a sparse, non-negative latent factor generative process.<n>We show that BNRM substantially mitigates reward over-optimization, improves robustness under distribution shifts, and yields more interpretable reward decompositions than strong baselines.
  arXiv  Detail & Related papers  (2026-02-11T08:14:11Z)
• Universal One-third Time Scaling in Learning Peaked Distributions [48.44706450307606]
  Training large language models (LLMs) is computationally expensive, partly because the loss exhibits slow power-law convergence.<n>We show that this behavior can arise intrinsically from the use of softmax and cross-entropy.
  arXiv  Detail & Related papers  (2026-02-03T16:06:18Z)
• Latent-Space Contrastive Reinforcement Learning for Stable and Efficient LLM Reasoning [16.244366307890832]
  We propose textbfDeepLatent Reasoning (DLR), a latent-space bidirectional contrastive reinforcement learning framework.<n>This framework shifts the trial-and-error cost from expensive token-level full sequence generation to the continuous latent manifold.<n> Experiments demonstrate that DLR achieves more stable training convergence, supports longer-horizon reasoning chains, and facilitates the sustainable accumulation of reasoning capabilities.
  arXiv  Detail & Related papers  (2026-01-24T03:18:22Z)
• Erase to Improve: Erasable Reinforcement Learning for Search-Augmented LLMs [18.37387666170851]
  We propose Erasable Reinforcement Learning (ERL), a novel framework that transforms fragile reasoning into a robust process.<n>ERL explicitly identifies faulty steps, erases them, and regenerates reasoning in place, preventing defective logic from propagating through the reasoning chain.<n>Models trained with ERL, termed ESearch, achieve substantial improvements on HotpotQA, MuSiQue, 2Wiki, and Bamboogle.
  arXiv  Detail & Related papers  (2025-10-01T13:10:36Z)
• Rethinking Reasoning Quality in Large Language Models through Enhanced Chain-of-Thought via RL [19.659532349434418]
  Reinforcement learning (RL) has recently become the dominant paradigm for strengthening the reasoning abilities of large language models.<n>Yet the rule-based reward functions commonly used on mathematical or programming benchmarks assess only answer format and correctness.<n>We propose Dynamic Reasoning Efficiency Reward (DRER) -- a plug-and-play RL reward framework that reshapes both reward and advantage signals.
  arXiv  Detail & Related papers  (2025-09-07T11:52:18Z)
• Non-stationary Online Learning for Curved Losses: Improved Dynamic Regret via Mixability [65.99855403424979]
  We show that dynamic regret can be substantially improved by leveraging the concept of mixability.<n>We demonstrate that an exponential-weight method with fixed-share updates achieves an $mathcalO(d T2/3 P_T2/3 log T)$ dynamic regret for mixable losses.
  arXiv  Detail & Related papers  (2025-06-12T12:00:08Z)
• Beyond Accuracy: Dissecting Mathematical Reasoning for LLMs Under Reinforcement Learning [93.00629872970364]
  Reinforcement learning (RL) has become the dominant paradigm for improving the performance of language models on complex reasoning tasks.<n>We introduce SPARKLE, a fine-grained analytic framework to dissect the effects of RL across three key dimensions.<n>We study whether difficult problems -- those yielding no RL signals and mixed-quality reasoning traces -- can still be effectively used for training.
  arXiv  Detail & Related papers  (2025-06-05T07:53:59Z)
• Unleashing the Power of Pre-trained Language Models for Offline Reinforcement Learning [50.9692060692705]
  This paper introduces $textbfLanguage Models for $textbfMo$tion Control ($textbfLaMo$), a general framework based on Decision Transformers for offline RL.<n>Our framework highlights four crucial components:.<n>Initializing Decision Transformers with sequentially pre-trained LMs, (2) employing the LoRA fine-tuning method,.<n>In particular, our method demonstrates superior performance in scenarios with limited data samples.
  arXiv  Detail & Related papers  (2023-10-31T16:24:17Z)
• Understanding, Predicting and Better Resolving Q-Value Divergence in Offline-RL [86.0987896274354]
  We first identify a fundamental pattern, self-excitation, as the primary cause of Q-value estimation divergence in offline RL. We then propose a novel Self-Excite Eigenvalue Measure (SEEM) metric to measure the evolving property of Q-network at training. For the first time, our theory can reliably decide whether the training will diverge at an early stage.
  arXiv  Detail & Related papers  (2023-10-06T17:57:44Z)

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.