proomt

Search

Search posts, papers, and topics

All posts

Hugging Face Daily PapersJiaxuan Jiang, Liyuan He, Zhixuan Fang1 min readpaperadvanced

CERA-MoA: Co-Evolving Routing Mechanisms with Continually Learning LLM Agents

Read on arxiv.org

Summary

CERA-MoA proposes a reinforcement‑learning loop where a router and a set of LLM agents are trained together. A “familiarity” estimator reads mid‑layer hidden states to predict each agent’s competence on a query, letting the router activate only a minimal subset of agents that meet a cumulative confidence threshold. The system also feeds targeted training examples to agents based on their evolving…

  • Separate routing and fine‑tuning limits MoA systems; co‑evolution lets routing adapt to agents’ changing abilities.
  • Familiarity estimator avoids full rollouts by using hidden‑state similarity, reducing compute overhead.
  • Adaptive routing uses a cumulative‑threshold to select the smallest agent set that satisfies a confidence budget.
  • Targeted sample allocation drives agents toward differentiated expertise during continual learning.

If you’re building LLM ensembles or modular AI services, the paper offers a concrete recipe for tighter integration between routing logic and agent training, potentially improving both performance and inference cost.

5/10
llmmachine learningreinforcement learningcontinual learningrouting

Related reading

  1. Online Learning with LLM Experts from Limited Feedback

    The paper models prompt routing to multiple LLM experts as a bandit problem with limited feedback and proposes algorithms that achieve sublinear regret in both full‑information and bandit settings. Experiments demonstrate that the methods learn effective routing strategies across diverse LLMs using only a small feedback budget.

    Hugging Face Daily Papersarxiv.org2 minpaper

  2. COBRA-Skills: Contextual Bandit-Guided Evolution for Agent Skill Optimization

    COBRA‑Skills uses a contextual‑bandit loop to selectively evaluate and evolve LLM agent skills, achieving better performance with roughly half the evaluation cost of prior methods. The framework works with limited examples and remains robust across different agent setups and even when the target model creates its own skills.

    Hugging Face Daily Papersarxiv.org1 minpaper

  4. The Router Within: Eliciting Native Skill Routing from a Frozen LLM

    The paper introduces Gavel, a method that extracts a frozen LLM's internal routing signal via two trained linear maps, eliminating the need to embed skill descriptions in the prompt. Experiments on Qwen3‑32B show up to 13.4‑point improvements on task benchmarks and higher skill‑use accuracy compared to larger retrieval‑based systems.

    Hugging Face Daily Papersarxiv.org1 minpaper

  5. Beyond Top-k Skill Retrieval: Diversity-Aware Skill Routing for LLM Agents

    Beyond Top‑k Skill Retrieval: Diversity‑Aware Skill Routing (DSR) applies a Determinantal Point Process with a query‑residual diversity kernel to rerank skill candidates, balancing relevance and redundancy. On the SkillRouter benchmark it raises recall and full‑coverage, especially for multi‑skill queries, showing that skill routing benefits from set‑selection rather than independent ranking.

    Hugging Face Daily Papersarxiv.org1 minpaper

  6. Generalized Agent Iteration: One Formal Framework for Iterative Policy Improvement and Recursive Self-Improvement

    The paper introduces Generalized Agent Iteration (GAI), a formal framework that unifies classical iterative policy improvement (GPI) and recursive self‑improvement (RSI). GAI treats an agent as a set of modifiable components and models learning as a loop of evaluation and improvement. Two binary “dials”—whether the improvement mechanism is internal to the agent and whether the evaluation standard…

    Hugging Face Daily Papersarxiv.org1 minpaper