Graph-based Target Back-Propagation for Context Adaptation in Multi-LLM Agentic Systems
On the map: LLM agents and tool use → General LLM agent frameworks and benchmarks
Abstract (arXiv)
Context adaptation automates prompt engineering in LLM-based systems by iteratively revising tunable prompts from task feedback, without modifying model weights. Extending this paradigm to multi-LLM agentic systems is crucial: existing methods suffer from inaccurate credit assignment and lack convergence guarantees. We propose Graph-based Target Back-Propagation (GTBP), a context adaptation framework for agentic workflows modeled as directed acyclic graphs. GTBP propagates local target outputs backward through the workflow graph and uses target--output discrepancies to guide a stage-wise prompt update mechanism. Theoretically, we show that GTBP's stage-wise prompt updates become stable over iterations, and that a sufficiently capable LLM optimizer can decrease the overall objective. Empirically, GTBP consistently outperforms strong baselines across three benchmarks while maintaining comparable computational cost.
Key facts. No citing papers in the map yet (data through 13 Oct 2026). Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use. Builds on 19 of its 72 references that are in the map. First posted on arXiv 12 Jun 2026.
Builds on
19 of 72 references.
Closest ancestors · 1
Same direction, newest first
Other references · 18
- 2018HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answeringcited by 3,716
- 2024Gemini 1.5: Unlocking multimodal understanding across millions of tokens of contextcited by 3,197
- 2016Optimization Methods for Large-Scale Machine Learningcited by 1,142
- 2022Large Language Models Are Human-Level Prompt Engineerscited by 1,031
- 2023ChatDev: Communicative Agents for Software Developmentcited by 901
- 2023DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelinescited by 814
- 2024Optimizing Instructions and Demonstrations for Multi-Stage Language Model Programscited by 274
- 2024LiveBench: A Challenging, Contamination-Limited LLM Benchmarkcited by 250
- 2023Transformers as Algorithms: Generalization and Stability in In-context Learningcited by 221
- 2014Difference Target Propagationcited by 146
- 2016Decoupled Neural Interfaces using Synthetic Gradientscited by 136
- 2014How Auto-Encoders Could Provide Credit Assignment in Deep Networks via Target Propagationcited by 79
6 more
- 2025Language Model Fine-Tuning on Scaled Survey Data for Predicting Distributions of Public Opinionscited by 65
- 2020A Theoretical Framework for Target Propagationcited by 57
- 2025Learning without training: The implicit dynamics of in-context learningcited by 42
- 2023The Closeness of In-Context Learning and Weight Shifting for Softmax Regressioncited by 41
- 2021Credit Assignment in Neural Networks through Deep Feedback Controlcited by 29
- 2026TextResNet: Decoupling and Routing Optimization Signals in Compound AI Systems via Deep Residual Tuningcited by 3
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2025Towards more Contextual Agents: An extractor-Generator Optimization Framework
- 2025MAPRO: Recasting Multi-Agent Prompt Optimization as Maximum a Posteriori Inference
- 2026Enhancing LLM Instruction Following: An Evaluation-Driven Multi-Agentic Workflow for Prompt Instructions Optimization
- 2026MAS-on-the-Fly: In-Context Structural Adaptation of LLM-Based Multi-Agent Systems
- 2026Inference-Time Graph Engineering for Multi-Agent LLM Workflows
- 2025Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models
- 2026Environment-Grounded Automated Prompt Optimization for LLM Game Agents
- 2025AMAS: Adaptively Determining Communication Topology for LLM-based Multi-Agent System