Entity · technique

Graph-Regularized Agentic Context Evolution

techniqueactiveprovisionalgraph-regularized-agentic-context-evolution-5fbaf880·1 events·first seen 11h ago

Aliases: Graph-Regularized Agentic Context Evolution

Co-occurring entities

Gemini 3.1 Pro Google Gemini-2.5-Flash-Lite τ²-Bench

More like this (12)

Is GraphRAG Needed? From Basic RAG to Graph-/Agentic Solutions with Context Optimization Entropy-Regularized Reinforcement Learning Latent Context Language Models AgenticRL Agentic Environment Engineering for Large Language Models: A Survey of Environment Modeling, Synthesis, Evaluation, and Application GraphRAG Agentic RL Optimization Dynamics Imprint Semantic Specificity in Contrastive Embedding Norms agentic re-optimization framework Observe-and-Act Adaptive Context Selection Generative Explainability for Next-Generation Networks: LLM-Augmented XAI with Mutual Feature Interactions Role-Agent: Bootstrapping LLM Agents via Dual-Role Evolution

Recent events (1)

5arXiv · cs.CL·11h ago·source ↗

GRACE: Graph-Regularized Agentic Context Evolution for reliable long-horizon instruction updates

Researchers introduce GRACE, a method that maintains a deployed LLM agent's persistent system-level instructions as a typed semantic graph rather than flat text, enabling local verification of updates within typed node neighborhoods. Evaluated on a telecom agent harness derived from τ²-bench under distribution shift, GRACE improves pass³ reliability from 0.091 (Gemini 2.5 Flash zero-shot) to 0.673±0.136, surpassing a Gemini 3.1 Pro zero-shot reference of 0.242. The work identifies structural substrate and consolidation mechanisms as key requirements for reliable long-horizon agentic context evolution. The flat-text baseline finishes at 0.191, underscoring the practical gap GRACE addresses.

Evaluation and Benchmarking Agent and Tool Ecosystem Gemini 3.1 Pro Google Gemini-2.5-Flash-Lite +2 more