Polaris: A Gödel Agent Framework for Small Language Models through Experience-Abstracted Policy Repair

2026-03-24

AI Summary

Small language models struggle to consistently and structurally self-correct during reasoning and lack the ability to distill reusable strategies from failures. To address this, this work proposes the Gödel Agent framework, which employs an experience abstraction mechanism to transform concrete errors into compact, transferable strategies and implements persistent strategy updates via auditable, minimal code patches. The framework establishes a recursive self-improvement loop encompassing meta-reasoning, strategy analysis, experience abstraction, and conservative code patching. Evaluated on benchmarks including MGSM, DROP, GPQA, and LitBench, a 7B model equipped with Polaris significantly outperforms baseline methods, demonstrating robust performance gains and strong cross-task generalization.

Technology Category

• Multiagent Systems: Adversarial Agents
• Search and Optimization: Learning to Search
• Cognitive Modeling & Cognitive Systems: Agent Architectures

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Gödel agent realize recursive self-improvement: an agent inspects its own policy and traces and then modifies that policy in a tested loop. We introduce Polaris, a Gödel agent for compact models that performs policy repair via experience abstraction, turning failures into policy updates through a structured cycle of analysis, strategy formation, abstraction, and minimal code pat ch repair with conservative checks. Unlike response level self correction or parameter tuning, Polaris makes policy level changes with small, auditable patches that persist in the policy and are reused on unseen instances within each benchmark. As part of the loop, the agent engages in meta reasoning: it explains its errors, proposes concrete revisions to its own policy, and then updates the policy. To enable cumulative policy refinement, we introduce experience abstraction, which distills failures into compact, reusable strategies that transfer to unseen instances. On MGSM, DROP, GPQA, and LitBench (covering arithmetic reasoning, compositional inference, graduate-level problem solving, and creative writing evaluation), a 7-billion-parameter model equipped with Polaris achieves consistent gains over the base policy and competitive baselines.

Problem

• policy repair
• experience abstraction
• self-improvement
• small language models
• recursive reasoning

Innovation

• Gödel agent
• policy repair
• experience abstraction
• recursive self-improvement
• small language models