Online Learning with LLM Experts from Limited Feedback
Summary
This paper formulates the adaptive routing of prompts to large language model experts as a contextual bandit problem with limited feedback, proposing algorithms that achieve sublinear regret and demonstrate efficient learning of high-quality routing strategies.
View Cached Full Text
Cached at: 09/14/26, 10:34 AM
Paper page - Online Learning with LLM Experts from Limited Feedback
Source: https://huggingface.co/papers/2609.05820
Abstract
Adaptive routing of prompts to LLM experts is formulated as a contextual bandit problem with limited feedback, yielding algorithms with sublinear regret bounds and effective routing strategies.
We studyadaptive routingof prompts to large language model (LLM) experts to maximize response quality in an online setting withlimited feedback. We formulate it as abandit problemwith K actions that represent experts and d features that encode prompts, over a horizon of T rounds. We propose algorithms that strategically select and observe rewards to minimize regret. In thefull-information setting, we achieve a regret of O(d T / m), while in the bandit setting we achieve O(d T K / m), where m ll T is a budget on feedback. Our experiments show that we efficiently learn high-quality routing strategies across diverse LLMs fromlimited feedback.
View arXiv pageView PDFAdd to collection
Get this paper in your agent:
hf papers read 2609\.05820
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2609.05820 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2609.05820 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2609.05820 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
Online Learning with LLM Experts from Limited Feedback
This paper proposes algorithms for adaptively routing prompts to LLM experts in an online setting with limited feedback, formulated as a bandit problem to minimize regret and maximize response quality.
No One Fits All: From Fixed Prompting to Learned Routing in Multilingual LLMs
Researchers from National Taiwan University propose replacing fixed translation-based prompting strategies in multilingual LLMs with lightweight learned classifiers that route each instance to either native or translation-based prompting. Their analysis across 10 languages and 4 benchmarks shows no single strategy is universally optimal, with translation benefiting low-resource languages most, and the learned routing achieving statistically significant improvements over fixed strategies.
Online LLM Selection via Constrained Bandits with Time-Varying Demand
This paper proposes a constrained stochastic bandit algorithm for online selection of large language models under time-varying task demand and heterogeneous accuracy, latency, and cost profiles, with theoretical guarantees on regret and constraint violations.
Towards Just-in-Time Adaptive Feedback: Enhancing Student Learning via Knowledge-Grounded LLM
This paper presents a framework that uses domain-specific expert knowledge to ground large language models for providing Just-in-Time adaptive feedback to students based on their written reasoning, achieving over 80% improvement in student performance in a large university course.
Correlation-Aware Contextual Bandits with Surrogate Rewards for LLM Routing
This paper proposes correlation-aware contextual bandit algorithms that leverage surrogate reward signals from machine learning models for LLM routing, achieving improved accuracy-cost trade-offs and sample efficiency compared to standard baselines.