PRIME: Policy-Reinforced Iterative Multi-agent Execution for Algorithmic Reasoning in Large Language Models

2026-01-19 · Jiawei Xu, Zhenyu Yu, Ziqian Bi, Minh Duc Pham, Xiaoyi Qu, Danyang Zhang arxiv

Large language models have demonstrated remarkable capabilities across diverse reasoning tasks, yet their performance on algorithmic reasoning remains limited. To handle this limitation, we propose PRIME (Policy-Reinforced Iterative Multi-agent Execution), a framework comprising three specialized agents, an executor for step-by-step reasoning, a verifier for constraint checking, and a coordinator for backtracking control, optimized through group relative policy optimization. For comprehensive evaluation, we introduce PRIME-Bench, the largest algorithmic reasoning benchmark to date, comprising 86 tasks across 12 categories with 51,600 instances. Tasks span sorting algorithms, graph and tree structures, automata and state machines, symbolic reasoning, and constraint-based puzzles, with execution traces reaching over one million steps. Compared to baseline approach, PRIME improves average accuracy from 26.8% to 93.8%, a 250% relative gain. The largest improvements occur on tasks requiring sustained state tracking, with Turing machine simulation improving from 9% to 92% and long division from 16% to 94%. Ablation studies identify iterative verification as the primary contributor, preventing the error propagation that causes baseline approaches to fail catastrophically. Analysis across model scales (8B-120B parameters) reveals that smaller models benefit disproportionately, achieving accuracy comparable to models 8x larger.

📄 PDF Abstract BibTeX arXiv:2602.11170

Code (0)

등록된 구현이 없습니다.

PRIME: Training Free Proactive Reasoning via Iterative Memory Evolution for User-Centric Agent

2026-04-08 · Prince Zizhuang Wang, Shuli Jiang arxiv

The development of autonomous tool-use agents for complex, long-horizon tasks in collaboration with human users has become the frontier of agentic research. During multi-turn Human-AI interactions, the dynamic and uncert…

Reinforcement Learning

MARTI-MARS$^2$: Scaling Multi-Agent Self-Search via Reinforcement Learning for Code Generation

2026-02-08 · Shijie Wang, Pengfei Li, Yikun Fu, Kaifeng Liu 외 arxiv

While the complex reasoning capability of Large Language Models (LLMs) has attracted significant attention, single-agent systems often encounter inherent performance ceilings in complex tasks such as code generation. Mul…

Multi-agent Reinforcement Learning Code Generation

Efficient Reinforced Feature Selection via Early Stopping Traverse Strategy

2021-09-29 · Kunpeng Liu, Pengfei Wang, Dongjie Wang, Wan Du 외

In this paper, we propose a single-agent Monte Carlo based reinforced feature selection (MCRFS) method, as well as two efficiency improvement strategies, i.e., early stopping (ES) strategy and reward-level interactive (R…

feature selection

A Flexible Multi-Agent LLM-Human Framework for Fast Human Validated Tool Building

2025-12-01 · Daull Xavier, Patrice Bellot, Emmanuel Bruno, Vincent Martin 외 arxiv

We introduce CollabToolBuilder, a flexible multiagent LLM framework with expert-in-the-loop (HITL) guidance that iteratively learns to create tools for a target goal, aligning with human intent and process, while minimiz…

Domain Adaptation

Contrastive Reinforced Policy Optimization via Privileged Self-Distillation

2026-07-30 · Xingjian Wu, Junlin Liu, Xingchen Liu, Xuhang Zhu 외 arxiv

Recent advances in post-training Large Language Models (LLMs) increasingly rely on Reinforcement Learning with Verifiable Rewards (RLVR) or On-Policy Self-Distillation (OPSD). While OPSD provides dense, logit-level super…

Reinforcement Learning Contrastive Learning