On the Generalization Gap in Self-Evolving Language Model Reasoning

• URL: http://arxiv.org/abs/2606.01075v2
• Date: Tue, 02 Jun 2026 05:50:15 GMT
• Title: On the Generalization Gap in Self-Evolving Language Model Reasoning
• Abstract summary: We analyze four representative strategies in a unified offline self-evolution framework.<n>Self-evolution consistently improves over the base model, but plateaus after excessive training compute is invested.<n>We find that multi-turn critic-revision with large models can reach strong self-evolution performance, with Gemma 12B nearly matching oracle-supervised training.
• Score: 16.207017999660653
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Recent work suggests that large language models (LLMs) can improve through self-evolution (SE), using supervision signals generated by the model itself. In this work, we ask: under a strict closed-loop setup, where the self-evolution algorithm has access only to an unlabeled prompt set and a base model, how close can internally generated supervision come to oracle-supervised training? We analyze four representative strategies in a unified offline self-evolution framework: single-round verification, multi-turn revision with feedback, iterative training, and curriculum learning. Our primary experiments use Knights and Knaves (KK) logical reasoning tasks, which provide deterministic solutions, controlled difficulty levels, and a clean testbed for easy-to-hard generalization. We first show that self-evolution consistently improves over the base model, but plateaus after excessive training compute is invested, and eventually still leaves a non-trivial gap to oracle supervision. We find that multi-turn critic-revision with large models can reach strong self-evolution performance, with Gemma 12B nearly matching oracle-supervised training. Beyond Knights and Knaves, we also evaluate self-evolution on real-world reasoning benchmarks, where gains are also modest. Overall, our results characterize when closed-loop self-evolution can help and show how internally generated supervision remains insufficient under this minimal formulation.

Related papers

• INFUSER: Influence-Guided Self-Evolution Improves Reasoning [54.101135873140066]
  We introduce an iterative co-training framework with two co-evolving roles.<n>A solver is trained with standard correctness rewards against the generator-provided answers.<n>An 8B INF co-evolving generator outperforms a frozen 32B thinking generator on math and coding.
  arXiv  Detail & Related papers  (2026-06-08T05:40:36Z)
• Self-evolving LLM agents with in-distribution Optimization [60.05867547965365]
  Large Language Models (LLMs) have recently emerged as powerful controllers for interactive agents in complex environments.<n>We propose Q-Evolve, a self-evolving framework for LLM agents that unifies automatic process-reward labeling and policy learning.<n>We evaluate our method on AlfWorld, WebShop, and ScienceWorld, showing Q-Evolve outperforms strong baselines in sample efficiency, robustness, and overall task performance.
  arXiv  Detail & Related papers  (2026-06-05T15:09:52Z)
• Confidence-Orchestrated Self-Evolution against Uncertain LLM Feedback [8.780008955074967]
  Self-evolving large language models (LLMs) learn by generating their own training tasks and solutions, reducing reliance on human-curated supervision.<n>We propose COSE (Confidence-Orchestrated Self-Evolution), which uses the LLM's intrinsic confidence as a lightweight uncertainty signal to modulate learning.
  arXiv  Detail & Related papers  (2026-05-27T06:07:10Z)
• Why Does Reinforcement Learning Generalize? A Feature-Level Mechanistic Study of Post-Training in Large Language Models [51.45122910254346]
  Reinforcement learning (RL)-based post-training often improves the reasoning performance of large language models beyond the training domain.<n>Supervised fine-tuning (SFT) frequently leads to general forgetting capabilities.<n>We present a feature-level mechanistic analysis methodology to probe RL generalization using a controlled experimental setup.
  arXiv  Detail & Related papers  (2026-04-27T21:22:34Z)
• When Models Judge Themselves: Unsupervised Self-Evolution for Multimodal Reasoning [30.162034423611292]
  We propose an unsupervised self-evolution training framework for multimodal reasoning.<n>It achieves stable performance improvements without using human-annotated answers or external reward models.<n>Our method consistently improves reasoning performance and generalization on five mathematical reasoning benchmarks.
  arXiv  Detail & Related papers  (2026-03-22T15:22:19Z)
• Guided Self-Evolving LLMs with Minimal Human Supervision [53.111086364268566]
  Unguided self-evolving systems often plateau quickly or degrade as training.<n>R-Few is a guided Self-Play Challenger acquirer framework that incorporates lightweight human oversight through in-context grounding and mixed training.<n>R-Few achieves consistent and iterative improvements across math and general reasoning benchmarks.
  arXiv  Detail & Related papers  (2025-12-02T07:06:11Z)
• Incentivizing LLMs to Self-Verify Their Answers [22.387551134333084]
  We propose a framework that incentivizes Large Language Models to self-verify their own answers.<n>We train our self-verification models based on Qwen2.5-Math-7B and DeepSeek-R1-Distill-Qwen-1.5B.<n> Experiments on multiple mathematical reasoning benchmarks show that our models can not only improve post-training performance but also enable effective test-time scaling.
  arXiv  Detail & Related papers  (2025-06-02T06:54:29Z)
• Can Large Reasoning Models Self-Train? [51.0277533541394]
  We use majority voting as a simple self-feedback mechanism to study whether self-training can be sustained within reinforcement learning.<n>We find that this basic approach improves not only the model's reasoning performance, but also its capability of generating better quality feedback for the next RL iteration.<n>Yet our analysis also reveals a critical limitation of such a self-training paradigm - prolonged RL with self-reward leads to reward hacking, resulting in sudden and complete performance collapse.
  arXiv  Detail & Related papers  (2025-05-27T17:16:00Z)
• Self-Improving Transformers Overcome Easy-to-Hard and Length Generalization Challenges [15.975023196507841]
  Large language models often struggle with length generalization and solving complex problem instances beyond their training distribution.<n>We present a self-improvement approach where models iteratively generate and learn from their own solutions.<n>Our results demonstrate how controlled weak-to-strong curricula can systematically teach a model logical extrapolation.
  arXiv  Detail & Related papers  (2025-02-03T18:45:22Z)
• Enhancing LLM Reasoning via Critique Models with Test-Time and Training-Time Supervision [120.40788744292739]
  We propose a two-player paradigm that separates the roles of reasoning and critique models. We first propose AutoMathCritique, an automated and scalable framework for collecting critique data. We demonstrate that the critique models consistently improve the actor's performance on difficult queries at test-time.
  arXiv  Detail & Related papers  (2024-11-25T17:11:54Z)
• Training Language Models to Self-Correct via Reinforcement Learning [98.35197671595343]
  Self-correction has been found to be largely ineffective in modern large language models (LLMs) We develop a multi-turn online reinforcement learning approach, SCoRe, that significantly improves an LLM's self-correction ability using entirely self-generated data. We find that SCoRe achieves state-of-the-art self-correction performance, improving the base models' self-correction by 15.6% and 9.1% respectively on MATH and HumanEval.
  arXiv  Detail & Related papers  (2024-09-19T17:16:21Z)
• Recursive Introspection: Teaching Language Model Agents How to Self-Improve [30.086494067593268]
  We develop RISE: Recursive IntroSpEction, an approach for fine-tuning large language models. Our experiments show that RISE enables Llama2, Llama3, and Mistral models to improve themselves with more turns on math reasoning tasks.
  arXiv  Detail & Related papers  (2024-07-25T17:35:59Z)

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.