Self-Improving Large Language Models via Progressive Experience Evolution

• URL: http://arxiv.org/abs/2608.02139v2
• Date: Tue, 04 Aug 2026 06:10:58 GMT
• Title: Self-Improving Large Language Models via Progressive Experience Evolution
• Abstract summary: We propose textbfSPEE (textbfSelf-textbfProgressive textbfExperience textbfEvolution), a unified post-training framework that sequentially performs explicit experience evolution followed by implicit policy optimization.<n> Experiments on five mathematical reasoning benchmarks demonstrate that SPEE consistently outperforms both test-time and training-time self-evolution baselines across three model scales.
• Score: 52.03479747358687
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large language models (LLMs) capable of self-improvement require not only effective policy optimization, but also a principled mechanism for transforming transient interaction experience into persistent model capabilities. Existing self-improvement paradigms remain fragmented: test-time methods can explicitly extract experience but cannot internalize it into model parameters, whereas training-time optimization methods can update model parameters but lack an explicit mechanism for accumulating transferable experience. Bridging these two paradigms requires a critical intermediate stage that remains underexplored, namely \emph{experience distillation}. To address this gap, we propose \textbf{SPEE} (\textbf{S}elf-\textbf{P}rogressive \textbf{E}xperience \textbf{E}volution), a unified post-training framework that sequentially performs explicit experience evolution followed by implicit policy optimization. During explicit experience evolution, SPEE reflects on trajectories collected from multiple interactions to extract, verify, and progressively evolve transferable experience, which is subsequently internalized into the policy through privilege-guided On-Policy Self-Distillation (OPSD). During implicit policy optimization, reward-driven reinforcement learning leverages these internalized priors to explore novel solution strategies. In the experience evolution stage, a continuously evolving global experience pool consolidates knowledge from both successful and failed trajectories, filters out low-utility experience, and mitigates post-hoc rationalization induced by individual trajectories. Experiments on five mathematical reasoning benchmarks demonstrate that SPEE consistently outperforms both test-time and training-time self-evolution baselines across three model scales. The source code is available at https://github.com/rrrsj/SPEE.

Related papers

• Learning What to Remember and What to Internalize in LLM Self-Evolution via Adaptive Memory-Parameter Coordination [40.98046592234726]
  We present COVE, a unified agent self-evolution framework.<n>It combines harness-based and parameter-based learning through task-aware routing, stage-aware scheduling, and knowledge optimization.<n>COVE outperforms single-channel evolution strategies, demonstrating more robust and efficient improvement under changing environments.
  arXiv  Detail & Related papers  (2026-08-02T13:39:01Z)
• SEED: Self-Evolving On-Policy Distillation for Agentic Reinforcement Learning [38.54413500261524]
  Outcome-based reinforcement learning (RL) provides a practical optimization paradigm, but its sparse trajectory-level rewards offer limited guidance on intermediate decisions.<n>We propose SEED (SElf-Evolving On-Policy Distillation), a self-evolving framework that converts completed on-policy trajectories into training-time hindsight skills.
  arXiv  Detail & Related papers  (2026-07-16T09:57:18Z)
• MetaEvo: A Meta-Optimization Framework for Experience-Driven Agent Evolution [49.24192816558635]
  Large language models (LLMs) exhibit strong reasoning capabilities, yet most LLM-based agents are unable to improve through task interactions.<n>Existing experience-driven methods often rely on memory or executors without enhancing the model's ability to learn.<n>We propose MetaEvo, a two-stage framework for continual agent evolution.
  arXiv  Detail & Related papers  (2026-05-29T09:31:39Z)
• AnE: Pushing the Reasoning Frontier of Multimodal LLMs via Anchor Evolution [61.593935260052795]
  Post-training viaSupervised Fine-Tuning (SFT) and Reinforcement Learning (RL) is crucial for enhancing reasoning in Multimodal Large Language Models (MLLMs)<n>Existing paradigms often reach a performance bottleneck due to the limitations of static data.<n>We propose Anchor Evolution (AnE), a new paradigm that integrates truth-anchored data curation and model evolution.
  arXiv  Detail & Related papers  (2026-05-25T08:26:34Z)
• EvoIR-Agent: Self-Evolving Image Restoration Agentic System via Experience-Driven Learning [39.60899845728349]
  Multimodal Large Language Model (MLLM)-driven image restoration agent demonstrates effectiveness in degradation coupling scenarios.<n>Training-based methods embed intrinsic experience into parameters, achieving high inference efficiency but lacking compatibility with new tools or degradation.<n>We propose EvoIR-Agent, which first systematically formulates the experience components of a training-free image restoration agent.
  arXiv  Detail & Related papers  (2026-05-21T09:14:25Z)
• Self-Evolving Spatial Reasoning in Vision Language Models via Geometric Logic Consistency [53.28076739140119]
  We propose a framework that enforces logical consistency in Vision-Language Models (VLMs) through geometric and linguistic duality operations.<n>SAGE is model-agnostic, data-efficient compared to prior GRPO methods, and can be applied as a lightweight post-training stage to any existing VLM.<n> Experiments on video and spatial reasoning benchmarks demonstrate consistent improvements over strong baselines.
  arXiv  Detail & Related papers  (2026-05-18T10:05:21Z)
• Evolving-RL: End-to-End Optimization of Experience-Driven Self-Evolving Capability within Agents [31.6974589324286]
  Self-evolving agents aim to overcome the static nature of large language models by distilling reusable experience from past interactions.<n>We propose Evolving-RL, an efficient algorithmic framework that jointly improves the experience extraction and utilization capabilities required for self-evolution.<n> Experiments on ALFWorld and Mind2Web show that Evolving-RL effectively enhances LLMs' ability to extract and reuse experience.
  arXiv  Detail & Related papers  (2026-05-11T14:43:56Z)
• Online Experiential Learning for Language Models [99.90826536842337]
  Online Experiential Learning (OEL) is a framework that enables language models to continuously improve from their own deployment experience.<n>OEL operates in two stages: first, transferable experiential knowledge is extracted and accumulated from interaction trajectories collected on the user side.<n>We evaluate OEL on text-based game environments across multiple model scales and both thinking and non-thinking variants.
  arXiv  Detail & Related papers  (2026-03-17T17:57:49Z)
• RetroAgent: From Solving to Evolving via Retrospective Dual Intrinsic Feedback [54.39884046754265]
  RetroAgent is an online RL framework that enables agents to master complex interactive environments.<n>Our experiments show that RetroAgent achieves state-of-the-art (SOTA) performance.
  arXiv  Detail & Related papers  (2026-03-09T16:23:33Z)
• Self-Consolidation for Self-Evolving Agents [51.94826934403236]
  Large language model (LLM) agents operate as static systems, lacking the ability to evolve through lifelong interaction.<n>We propose a novel self-evolving framework for LLM agents that introduces a complementary evolution mechanism.
  arXiv  Detail & Related papers  (2026-02-02T11:16:07Z)

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.