In-the-Flow Agentic System Optimization for Effective Planning and Tool Use

• URL: http://arxiv.org/abs/2510.05592v1
• Date: Tue, 07 Oct 2025 05:32:44 GMT
• Title: In-the-Flow Agentic System Optimization for Effective Planning and Tool Use
• Authors: Zhuofeng Li, Haoxiang Zhang, Seungju Han, Sheng Liu, Jianwen Xie, Yu Zhang, Yejin Choi, James Zou, Pan Lu,
• Abstract summary: AgentFlow is a trainable, in-the-flow agentic framework that coordinates four modules (planner, executor, verifier, generator) through an evolving memory.<n>Flow-GRPO tackles long-horizon, sparse-reward credit assignment by converting multi-turn optimization into a sequence of tractable single-turn policy updates.<n>AgentFlow with a 7B-scale backbone outperforms top-performing baselines with average accuracy gains of 14.9% on search, 14.0% on agentic, 14.5% on mathematical, and 4.1% on scientific tasks.
• Score: 73.72524040856052
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Outcome-driven reinforcement learning has advanced reasoning in large language models (LLMs), but prevailing tool-augmented approaches train a single, monolithic policy that interleaves thoughts and tool calls under full context; this scales poorly with long horizons and diverse tools and generalizes weakly to new scenarios. Agentic systems offer a promising alternative by decomposing work across specialized modules, yet most remain training-free or rely on offline training decoupled from the live dynamics of multi-turn interaction. We introduce AgentFlow, a trainable, in-the-flow agentic framework that coordinates four modules (planner, executor, verifier, generator) through an evolving memory and directly optimizes its planner inside the multi-turn loop. To train on-policy in live environments, we propose Flow-based Group Refined Policy Optimization (Flow-GRPO), which tackles long-horizon, sparse-reward credit assignment by converting multi-turn optimization into a sequence of tractable single-turn policy updates. It broadcasts a single, verifiable trajectory-level outcome to every turn to align local planner decisions with global success and stabilizes learning with group-normalized advantages. Across ten benchmarks, AgentFlow with a 7B-scale backbone outperforms top-performing baselines with average accuracy gains of 14.9% on search, 14.0% on agentic, 14.5% on mathematical, and 4.1% on scientific tasks, even surpassing larger proprietary models like GPT-4o. Further analyses confirm the benefits of in-the-flow optimization, showing improved planning, enhanced tool-calling reliability, and positive scaling with model size and reasoning turns.

Related papers

• Diverse Approaches to Optimal Execution Schedule Generation [0.0]
  We present the first application of MAP-Elites, a quality-diversity algorithm, to trade execution.<n>Rather than searching for a single optimal policy, MAP-Elites generates a diverse portfolio of regime-specialist strategies indexed by liquidity and volatility conditions.<n>Individual specialists achieve 8-10% performance improvements within their behavioural niches, while other cells show degradation.
  arXiv  Detail & Related papers  (2026-01-29T18:41:52Z)
• GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization [133.27496265096445]
  We show how to apply Group Relative Policy Optimization under multi-reward setting without examining its suitability.<n>We then introduce Group reward-Decoupled Normalization Policy Optimization (GDPO), a new policy optimization method to resolve these issues.<n>GDPO consistently outperforms GRPO, demonstrating its effectiveness and generalizability for multi-reward reinforcement learning optimization.
  arXiv  Detail & Related papers  (2026-01-08T18:59:24Z)
• AWPO: Enhancing Tool-Use of Large Language Models through Explicit Integration of Reasoning Rewards [60.2998874976509]
  We propose advantage-weighted policy optimization (AWPO) to integrate explicit reasoning rewards to enhance tool-use capability.<n>AWPO incorporates variance-aware gating and difficulty-aware weighting to adaptively modulate advantages from reasoning signals.<n>Experiments demonstrate that AWPO achieves state-of-the-art performance across standard tool-use benchmarks.
  arXiv  Detail & Related papers  (2025-12-22T08:07:00Z)
• Efficient Multi-turn RL for GUI Agents via Decoupled Training and Adaptive Data Curation [65.3648667980258]
  Vision-language model (VLM) based GUI agents show promise for automating complex tasks, but face significant challenges in applying reinforcement learning (RL)<n>We propose DART, a Decoupled Agentic RL Training framework for GUI agents, which coordinates heterogeneous modules in a highly decoupled manner.<n>On the OSWorld benchmark, DART-GUI-7B achieves a 42.13% task success rate, a 14.61% absolute gain over the base model, and 7.34% higher than open-source SOTA.
  arXiv  Detail & Related papers  (2025-09-28T13:19:20Z)
• Building Coding Agents via Entropy-Enhanced Multi-Turn Preference Optimization [13.271737599933147]
  We introduce EntroPO, an entropy-enhanced framework that adapts existing preference optimization algorithms to the multi-turn, tool-assisted setting.<n>We validate EntroPO by fine-tuning a diverse suite of models from different families and sizes.<n>On the swebench leaderboard, our approach establishes new state-of-the-art results among open-weight models.
  arXiv  Detail & Related papers  (2025-09-15T20:36:19Z)
• Step-wise Policy for Rare-tool Knowledge (SPaRK): Offline RL that Drives Diverse Tool Use in LLMs [0.0]
  Step-wise Policy for Rare-tool Knowledge (SPaRK) teaches large language models to explore diverse tool usage patterns.<n>We introduce a dual-objective reward system that simultaneously optimize for answer quality and tool diversity.
  arXiv  Detail & Related papers  (2025-07-15T14:44:29Z)
• TGRPO :Fine-tuning Vision-Language-Action Model via Trajectory-wise Group Relative Policy Optimization [12.061547251822326]
  Trajectory-based Group Relative Policy Optimization (TGRPO) is an online RL-based training framework for Visual-Language-Action (VLA) models.<n>We show that TGRPO achieves an average success rate of 80.7%, which is 4.2% higher than that of Supervised Fine-Tuning (SFT) and outperforms other representative RL-based post-training methods.
  arXiv  Detail & Related papers  (2025-06-10T04:27:49Z)
• Acting Less is Reasoning More! Teaching Model to Act Efficiently [87.28134636548705]
  Tool-integrated reasoning augments large language models with the ability to invoke external tools to solve tasks.<n>Current approaches typically optimize only for final correctness without considering the efficiency or necessity of external tool use.<n>We propose a framework that encourages models to produce accurate answers with minimal tool calls.<n>Our approach reduces tool calls by up to 68.3% and improves tool productivity by up to 215.4%, while maintaining comparable answer accuracy.
  arXiv  Detail & Related papers  (2025-04-21T05:40:05Z)
• Weak-for-Strong: Training Weak Meta-Agent to Harness Strong Executors [104.5401871607713]
  This paper proposes Weakfor-Strong Harnessing (W4S), a novel framework that customizes smaller, cost-efficient language models to design and optimize for harnessing stronger models.<n>W4S formulates design as a multi-turn markov decision process and introduces reinforcement learning for agentic workflow optimization.<n> Empirical results demonstrate the superiority of W4S that our 7B meta-agent, trained with just one GPU hour, outperforms the strongest baseline by 2.9% 24.6% across eleven benchmarks.
  arXiv  Detail & Related papers  (2025-04-07T07:27:31Z)
• Collab: Controlled Decoding using Mixture of Agents for LLM Alignment [90.6117569025754]
  Reinforcement learning from human feedback has emerged as an effective technique to align Large Language models.<n>Controlled Decoding provides a mechanism for aligning a model at inference time without retraining.<n>We propose a mixture of agent-based decoding strategies leveraging the existing off-the-shelf aligned LLM policies.
  arXiv  Detail & Related papers  (2025-03-27T17:34:25Z)
• DARS: Dynamic Action Re-Sampling to Enhance Coding Agent Performance by Adaptive Tree Traversal [55.13854171147104]
  Large Language Models (LLMs) have revolutionized various domains, including natural language processing, data analysis, and software development.<n>We present Dynamic Action Re-Sampling (DARS), a novel inference time compute scaling approach for coding agents.<n>We evaluate our approach on SWE-Bench Lite benchmark, demonstrating that this scaling strategy achieves a pass@k score of 55% with Claude 3.5 Sonnet V2.
  arXiv  Detail & Related papers  (2025-03-18T14:02:59Z)
• Data-Scarce Identification of Game Dynamics via Sum-of-Squares Optimization [29.568222003322344]
  We introduce the Side-Information Assisted Regression (SIAR) framework, designed to identify game dynamics in multiplayer normal-form games. SIAR is solved using sum-of-squares (SOS) optimization, resulting in a hierarchy of approximations that provably converge to the true dynamics of the system. We showcase that the SIAR framework accurately predicts player behavior across a spectrum of normal-form games, widely-known families of game dynamics, and strong benchmarks, even if the unknown system is chaotic.
  arXiv  Detail & Related papers  (2023-07-13T09:14:48Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.