AgentBrew: Lifelong Knowledge Brewing from Strong Teachers to Weak LLM Agents
Artificial Intelligence
Summary
The authors study a way to help smaller AI models learn from bigger, smarter ones without needing extra training or direct help during use. They create a method called AgentBrew that stores useful lessons from the big model in a special memory for the smaller one. This method deals with the challenges of not having clear feedback and making sure the instructions suit the smaller model’s abilities. The authors show that their approach works well on tasks like coding and math by making the smaller AI act smarter using just stored notes.
Authors
Yangqin Jiang, Chao Huang
Abstract
Deploying LLM agents typically requires a compact test-time student, even if a stronger teacher is available during training. We study knowledge brewing: distilling a teacher's interactive experience into a persistent external memory for the student. Crucially, this requires no weight updates, expert demonstrations, ground-truth labels, or test-time teacher access. This setting poses two challenges: environments provide only sparse, binary feedback, and teacher-authored notes must be inherently tailored to be concretely executable by a substantially weaker student. To address these hurdles, we propose AgentBrew, comprising two coupled components. First, a failure-triggered teacher--Ralph Loop mitigates sparse feedback by transforming student failures into environment-validated notes. Second, student-aware synthesis calibrates teacher knowledge to the weak executor's operational granularity, yielding model-specific, actionable guidance. Extensive evaluations and comprehensive ablations across coding, math, and tool-use tasks demonstrate that this asymmetric, training-free brewing paradigm produces highly capable yet deployable LLM agents.