Co-ReAct: Rubrics as Step-Level Collaborators for ReAct Agents
Quick Answer
Co-ReAct introduces a rubric-guided framework that enhances ReAct agents' decision-making in multi-step reasoning tasks, improving performance on DeepResearchBench and SQA-CS-V2 benchmarks.
Quick Take
Co-ReAct introduces a rubric-guided framework that enhances ReAct agents' decision-making in multi-step reasoning tasks, improving performance on DeepResearchBench and SQA-CS-V2 benchmarks. The dedicated rubric generator optimizes guidance using multi-judge expert consensus, yielding better outcomes without altering existing decision mechanisms. Code is available at GitHub.
Key Points
- Co-ReAct uses rubrics for step-level guidance in ReAct agents' decision-making.
- Improves performance on DeepResearchBench and SQA-CS-V2 benchmarks over previous models.
- Trained rubric generator enhances existing systems without changing their decision processes.
- Optimizes guidance using a list-wise Spearman rank-correlation reward mechanism.
- Code is publicly available for further research and application.
Paper Resources
Article Content
From source RSS / original summaryarXiv:2605. 23590v1 Announce Type: new Abstract: ReAct-style agents for search-intensive, multi-step reasoning tasks rely largely on their own internal judgment to decide what evidence to seek, which reasoning or action step to take next, and when to stop, often producing shallow, redundant, or poorly targeted trajectories.
Prior work has explored rubrics as external quality signals, but existing uses are mostly evaluative rather than action-guiding: rubrics typically serve as training-time rewards or post-hoc evaluators of completed outputs, and in deep-research settings they are often coarse-grained and report-level rather than step-level. We introduce Co-ReAct, a rubric-guided action-selection framework that uses rubrics as step-level guidance during inference.
At each decision step, Co-ReAct injects a rubric into the agent's context to guide the next Reason-or-Act decision, specifying what the agent should target in evidence seeking, search, reasoning, or self-evaluation. To make this guidance reliable, we train a dedicated rubric generator with GRPO.
Unlike prior pairwise or binary preference formulations, our objective optimizes a list-wise Spearman rank-correlation reward against multi-judge expert consensus rankings, encouraging rubrics that are discriminative rather than merely plausible. On DeepResearchBench and SQA-CS-V2, Co-ReAct consistently improves over ReAct and representative test-time compute baselines across search agents built on both 8B/14B open-source and frontier closed-source base models.
The trained rubric generator can also serve as a drop-in component that improves these baselines without changing their underlying decision mechanisms. Our code is publicly available at https://github. com/ZBWpro/Co-ReAct.
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →Adversarial Social Epistemology for Assemblies of Humans and Large Language Models
The paper introduces Adversarial Social Epistemology (ASE) to analyze how agents manipulate trust in public communications, highlighting mechanisms that undermine the reliability of testimony and inference. It critiques existing frameworks like epistemic bubbles and misinformation diffusion, proposing a new language for understanding trust breaches and auditing inferential chains in densely interactive environments involving humans and large language models.