# No More Stale Feedback: Co-Evolving Critics for Open-World Agent Learning

arXiv: https://arxiv.org/abs/2601.06794 · HTML: https://mlsanity.com/paper/2601.06794/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- Placed in Reinforcement learning for LLM agents, one of 16 directions in LLM agents and tool use.
- Builds on 15 of its 25 references that are in the map.
- First posted on arXiv 11 Jan 2026.

## Placed in

- LLM agents and tool use / [Reinforcement learning for LLM agents](https://mlsanity.com/agents/reinforcement-learning-for-llm-agents/)

## Builds on (most cited first, 15 of 15)

- [WebShop: Towards Scalable Real-World Web Interaction with Grounded Language Agents](https://mlsanity.com/paper/2207.01206/) (2022)
- [ALFWorld: Aligning Text and Embodied Environments for Interactive Learning](https://mlsanity.com/paper/2010.03768/) (2020)
- [Self-critiquing models for assisting human evaluators](https://mlsanity.com/paper/2206.05802/) (2022)
- [Learning to Reason under Off-Policy Guidance](https://mlsanity.com/paper/2504.14945/) (2025)
- [LLM Critics Help Catch LLM Bugs](https://mlsanity.com/paper/2407.00215/) (2024)
- [AgentGym-RL: Training LLM Agents for Long-Horizon Decision Making through Multi-Turn Reinforcement Learning](https://mlsanity.com/paper/2509.08755/) (2025)
- [Scaf-GRPO: Scaffolded Group Relative Policy Optimization for Enhancing LLM Reasoning](https://mlsanity.com/paper/2510.19807/) (2025)
- [LLM Critics Help Catch Bugs in Mathematics: Towards a Better Mathematical Verifier with Natural Language Feedback](https://mlsanity.com/paper/2406.14024/) (2024)
- [StepHint: Multi-level Stepwise Hints Enhance Reinforcement Learning to Reason](https://mlsanity.com/paper/2507.02841/) (2025)
- [GHPO: Adaptive Guidance for Stable and Efficient LLM Reinforcement Learning](https://mlsanity.com/paper/2507.10628/) (2025)
- [Boosting MLLM Reasoning with Text-Debiased Hint-GRPO](https://mlsanity.com/paper/2503.23905/) (2025)
- [The Lighthouse of Language: Enhancing LLM Agents via Critique-Guided Improvement](https://mlsanity.com/paper/2503.16024/) (2025)
- [Don't Tell the Answer, Truly Guide the Reasoning During RL Rollouts](https://mlsanity.com/paper/2510.09388/) (2025)
- [ADHint: Adaptive Hints with Difficulty Priors for Reinforcement Learning](https://mlsanity.com/paper/2512.13095/) (2025)
- [Do Not Step Into the Same River Twice: Learning to Reason from Trial and Error](https://mlsanity.com/paper/2510.26109/) (2025)

## Built on by (most cited first, 8 of 8)

- [Train at Moving Edge: Online-Verified Prompt Selection for Efficient RL Training of Large Reasoning Model](https://mlsanity.com/paper/2603.25184/) (2026)
- [SPPO: Sequence-Level PPO for Long-Horizon Reasoning Tasks](https://mlsanity.com/paper/2604.08865/) (2026)
- [CAFE: Self-Improving Search Agents Need Co-Evolving Feedback](https://mlsanity.com/paper/2608.24794/) (2026)
- [Co-Evolution in Agentic Systems: Toward Self-Directed Evolution Beyond Human Design](https://mlsanity.com/paper/2608.10299/) (2026)
- [EvoRS: On-Policy Self-Evolution of Reward Systems for Open-Ended Reinforcement Learning](https://mlsanity.com/paper/2609.12459/) (2026)
- [Learning to Coach for Experiential Learning](https://mlsanity.com/paper/2609.15851/) (2026)
- [My FAULT: Self-Diagnosis as Credit Assignment in Self-Evolving Agentic Reinforcement Learning](https://mlsanity.com/paper/2610.01161/) (2026)
- [DUET: Co-Evolving Solver and Grader Agents](https://mlsanity.com/paper/2610.04087/) (2026)

## Abstract (arXiv)

Critique-guided reinforcement learning (RL) has emerged as a powerful paradigm for training LLM agents by augmenting sparse outcome rewards with natural-language feedback. However, current methods often rely on static or offline critic models, which fail to adapt as the policy evolves. In on-policy RL, the agent's error patterns shift over time, causing stationary critics to become stale and providing feedback of diminishing utility. To address this, we introduce ECHO (Evolving Critic for Hindsight-Guided Optimization)}, a framework that jointly optimizes the policy and critic through a synchronized co-evolutionary loop. ECHO utilizes a cascaded rollout mechanism where the critic generates multiple diagnoses for an initial trajectory, followed by policy refinement to enable group-structured advantage estimation. We address the challenge of learning plateaus via a saturation-aware gain shaping objective, which rewards the critic for inducing incremental improvements in high-performing trajectories. By employing dual-track GRPO updates, ECHO ensures the critic's feedback stays synchronized with the evolving policy. Experimental results show that ECHO yields more stable training and higher long-horizon task success across open-world environments.