CB-Orchestrator: Adaptive Workflow Optimization for LLM Agents via Contextual Bandits

Key points

• Large Language Model (LLM) agents have demonstrated remarkable capabilities in tackling complex tasks through agentic workflows.1
• In the first stage, we construct a diverse pool of candidate workflows via evolutionary search.1
• In the second stage, we formulate workflow selection as a Contextual Bandit problem, which enables sample-efficient learning by balancing exploration and exploitation, thereby significantly reducing the feedback required for training.1
• During inference, the model adaptively selects the optimal workflow tailored to each query.1

Sources (1)

• [1]CB-Orchestrator: Adaptive Workflow Optimization for LLM Agents via Contextual Bandits
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 9, 06:25 AM

  To address these limitations, we propose CB-Orchestrator, an adaptive framework that decouples workflow generation from selection.

  Large Language Model (LLM) agents have demonstrated remarkable capabilities in tackling complex tasks through agentic workflows.

Extractive summary: sentences quoted from the sources.