paper-with-me

홈 › Papers

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

2026-01-11 · Zhicong Li, Lingjie Jiang, Yulan Hu, Xingchen Zeng, Yixia Li, Xiangwen Zhang, Guanhua Chen, Zheng Pan, Xin Li, Yong Liu 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.

📄 PDF Abstract BibTeX arXiv:2601.06794

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

Similar Papers 제목 키워드 기반

Co-Evolving Actor-Conditioned Critics for Non-Verifiable Generation

2026-08-31 · Jinyoung Kim, Muhammad Khalifa, Lajanugen Logeswaran, Jaekyeom Kim 외 arxiv

Natural-language critiques provide supervision beyond scalar rewards for non-verifiable generation, which lacks deterministic verifiers. In critique-guided refinement, a critic gives feedback on an initial response and a…

DeepCritic: Deliberate Critique with Large Language Models

2025-05-01 · Wenkai Yang, Jingwen Chen, Yankai Lin, Ji-Rong Wen

As Large Language Models (LLMs) are rapidly evolving, providing accurate feedback and scalable oversight on their outputs becomes an urgent and critical problem. Leveraging LLMs as critique models to achieve automated su…

Math

LLM Critics Help Catch LLM Bugs

2024-06-28 · Nat McAleese, Rai Michael Pokorny, Juan Felipe Ceron Uribe, Evgenia Nitishinskaya 외

Reinforcement learning from human feedback (RLHF) is fundamentally limited by the capacity of humans to correctly evaluate model output. To improve human evaluation ability and overcome that limitation this work trains "…

AsyncOPD: How Stale Can On-Policy Distillation Be?

2026-06-23 · Wonjun Kang, Kevin Galim, Seunghyuk Oh, Minjun Kang 외 arxiv

On-policy distillation (OPD) trains a student on its own rollouts guided by teacher feedback and is becoming increasingly important for large language model (LLM) post-training. Like reinforcement learning (RL), however,…

Reinforcement Learning

BRACE: Anchored Bellman-Residual Correction for Stale Critics in Asynchronous RL

2026-09-09 · Guanqun Zhao, Zijun Xie, Binbin Zheng, Jiafeng Lu 외 arxiv

Asynchronous reinforcement learning has become the standard way to scale training for language models, but the resulting policy lag biases the critic toward the stale behavior policy. Existing work on asynchronous LLM tr…

Reinforcement Learning