mlsanityData to 14 Oct 2026

Self-Evolving Multi-Agent Systems via Textual Backpropagation

arXiv 2506.09046 ↗First posted 10 Jun 2025cs.LG cs.AI cs.MAcited by 23

On the map: LLM agents and tool use → General LLM agent frameworks and benchmarks

also in 1 more
Abstract (arXiv)

Leveraging multiple Large Language Models (LLMs) has proven effective for addressing complex, high-dimensional tasks, but current approaches often rely on static, manually engineered multi-agent configurations. To overcome these constraints, we present the Agentic Neural Network (ANN), a framework that conceptualizes multi-agent collaboration as a layered neural network architecture. In this design, each agent operates as a node, and each layer forms a cooperative team focused on a specific subtask. Our framework follows a two-phase optimization strategy: (1) Forward Phase - Drawing inspiration from neural network forward passes, tasks are dynamically decomposed into subtasks, and cooperative agent teams with suitable aggregation methods are constructed layer by layer. (2) Backward Phase - Mirroring backpropagation, we refine both global and local collaboration through iterative feedback, allowing agents to self-evolve their roles, prompts, and coordination. This neuro-symbolic approach enables our framework to create new or specialized agent teams post-training, delivering notable gains in accuracy and adaptability. Across seven benchmark datasets, our work surpasses leading multi-agent baselines under the same configurations, showing consistent performance improvements.

Key facts. Cited by 23 papers in the map (data through 14 Oct 2026). Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use. Builds on 47 of its 112 references that are in the map. First posted on arXiv 10 Jun 2025.

Builds on

47 of 112 references.

Closest ancestors · 8

Same direction, newest first

  1. 2025MAS-ZERO: Designing Multi-Agent Systems with Zero Supervision21 May
  2. 2024Symbolic Learning Enables Self-Evolving Agents26 Jun
  3. 2024Mixture-of-Agents Enhances Large Language Model Capabilities7 Jun
  4. 2024Tapilot-Crossing: Benchmarking and Evolving LLMs Towards Interactive Data Analysis Agents8 Mar
  5. 2023A Dynamic LLM-Powered Agent Network for Task-Oriented Agent Collaboration3 Oct
  6. 2023AutoAgents: A Framework for Automatic Agent Generation29 Sep
  7. 2023Agents: An Open-source Framework for Autonomous Language Agents14 Sep
  8. 2023MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework1 Aug

Other references · 39

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2022Training language models to follow instructions with human feedbackcited by 15,971
  3. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  4. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,297
  5. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
  6. 2021Evaluating Large Language Models Trained on Codecited by 8,320
  7. 2020Measuring Massive Multitask Language Understandingcited by 7,963
  8. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
  9. 2022Large Language Models are Zero-Shot Reasonerscited by 4,374
  10. 2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 4,276
  11. 2017Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layercited by 3,509
  12. 2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 3,442
27 more
  1. 2023SWE-bench: Can Language Models Resolve Real-World GitHub Issues?cited by 3,373
  2. 2023AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversationcited by 1,988
  3. 2023DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelinescited by 832
  4. 2024MACE: Mass Concept Erasure in Diffusion Modelscited by 307
  5. 2023Large Language Models as Tool Makerscited by 303
  6. 2023TF-ICON: Diffusion-Based Training-Free Cross-Domain Image Compositioncited by 182
  7. 2023Who Wrote this Code? Watermarking for Code Generationcited by 141
  8. 2024InfiAgent-DABench: Evaluating Agents on Data Analysis Taskscited by 140
  9. 2024NATURAL PLAN: Benchmarking LLMs on Natural Language Planningcited by 128
  10. 2024Meta-Prompting: Enhancing Language Models with Task-Agnostic Scaffoldingcited by 117
  11. 2025ChartCoder: Advancing Multimodal Large Language Model for Chart-to-Code Generationcited by 92
  12. 2024ConU: Conformal Uncertainty in Large Language Models with Correctness Coverage Guaranteescited by 65
  13. 2025PRISM: Self-Pruning Intrinsic Selection Method for Training-Free Multimodal Data Selectioncited by 61
  14. 2025VisuoThink: Empowering LVLM Reasoning with Multimodal Tree Searchcited by 36
  15. 2025AutoReproduce: Automatic AI Experiment Reproduction with Paper Lineagecited by 32
  16. 2024DOPRA: Decoding Over-accumulation Penalization and Re-allocation in Specific Weighting Layercited by 27
  17. 2023SPOT! Revisiting Video-Language Models for Event Understandingcited by 20
  18. 2025Evaluating Supervised Learning Models for Fraud Detection: A Comparative Study of Classical and Deep Architectures on Imbalanced Transaction Datacited by 20
  19. 2022FR: Folded Rationalization with a Unified Encodercited by 18
  20. 2023Fed-GraB: Federated Long-tailed Learning with Self-Adjusting Gradient Balancercited by 17
  21. 2025Gated Multimodal Graph Learning for Personalized Recommendationcited by 15
  22. 2019Single Image Super Resolution based on a Modified U-net with Mixed Gradient Losscited by 13
  23. 2025Neural Parameter Search for Slimmer Fine-Tuned Models and Better Transfercited by 11
  24. 2024FedLoGe: Joint Local and Generic Federated Learning under Long-tailed Datacited by 10
  25. 2023Joint Self-supervised Depth and Optical Flow Estimation towards Dynamic Objectscited by 8
  26. 2025DanmakuTPPBench: A Multi-modal Benchmark for Temporal Point Process Modeling and Understandingcited by 7
  27. 2023Rescue: Ranking LLM Responses with Partial Ordering to Improve Response Generationcited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 6; 2026: 17.

23 papers.

Sort:

By area:

  1. 2025The Landscape of Agentic Reinforcement Learning for LLMs: A Surveycited by 205
  2. 2025Agentic AI: A Comprehensive Survey of Architectures, Applications, and Future Directionscited by 48
  3. 2025ASCD: Attention-Steerable Contrastive Decoding for Reducing Hallucination in MLLMcited by 37
  4. 2025CoMAS: Co-Evolving Multi-Agent Systems via Interaction Rewardscited by 29
  5. 2026MetaSkill-Evolve: Recursive Self-Improvement of LLM Agents via Two-Timescale Meta-Skill Evolutioncited by 22
  6. 2026Mem2Evolve: Towards Self-Evolving Agents via Co-Evolutionary Capability Expansion and Experience Distillationcited by 12
  7. 2026OPD-V: Visual On-Policy Self-Distillation with Modality Balancecited by 8
  8. 2026Beyond Individual Intelligence: Surveying Collaboration, Failure Attribution, and Self-Evolution in LLM-based Multi-Agent Systemscited by 6
  9. 2026ReflectRL: Learning from Golden Negative Trajectories via Reflective-to-Direct Reasoningcited by 6
  10. 2026Self-Improvement of Large Language Models: A Technical Overview and Future Outlookcited by 5
  11. 2026SkillMAS: Skill Co-Evolution with LLM-based Multi-Agent Systemcited by 5
  12. 2026LangMARL: Natural Language Multi-Agent Reinforcement Learningcited by 4
11 more
  1. 2025CubeBench: Diagnosing Interactive, Long-Horizon Spatial Reasoning Under Partial Observationscited by 3
  2. 2026JudgeFlow: Agentic Workflow Optimization via Block Judgecited by 3
  3. 2026Experience as a Compass: Multi-agent RAG with Evolving Orchestration and Agent Promptscited by 3
  4. 2025DR. WELL: Dynamic Reasoning and Learning with Symbolic World Model for Embodied LLM-Based Multi-Agent Collaborationcited by 2
  5. 2026MASkills: Continual Skills Optimization for Multi-Agent LLM Systemscited by 1
  6. 2026Do Self-Evolving Skills Generalize to Held-Out Tasks?cited by 1
  7. 2026TextBFGS: A Case-Based Reasoning Approach to Code Optimization via Error-Operator Retrievalcited by 0
  8. 2026Graph-based Target Back-Propagation for Context Adaptation in Multi-LLM Agentic Systemscited by 0
  9. 2026Asymptotic Risk Calibration for Selective Question Answeringcited by 0
  10. 2026Inherit-MAS: Test-Time Evolution of Multi-Agent Systems through Workflow and Execution Inheritancecited by 0
  11. 2026HarnessNet: Feedback-driven Hierarchical Joint Harness Evolution for Multi-agent Systemscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026NeuroMAS: Multi-Agent Systems as Neural Networks with Joint Reinforcement Learning
  2. 2025AgentNet: Decentralized Evolutionary Coordination for LLM-based Multi-Agent Systems
  3. 2024Scaling Large Language Model-based Multi-Agent Collaboration
  4. 2025Learn as Individuals, Evolve as a Team: Multi-agent LLMs Adaptation in Embodied Environments
  5. 2026DIG to Heal: Scaling General-purpose Agent Collaboration via Explainable Dynamic Decision Paths
  6. 2023ProAgent: Building Proactive Cooperative Agents with Large Language Models
  7. 2026Agent Primitives: Reusable Latent Building Blocks for Multi-Agent Systems
  8. 2026An Empirical Study of Multi-Agent Collaboration for Automated Research