Enregistré dans:
Détails bibliographiques
Auteurs principaux: Wu, Yichao, Liang, Penghao, Xiang, Yafei, Yuan, Mengwei, Liu, Jianan, Yang, Jing, Li, Xianyou, Yan, Weiran
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:https://arxiv.org/abs/2603.00846
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866914361385156608
author Wu, Yichao
Liang, Penghao
Xiang, Yafei
Yuan, Mengwei
Liu, Jianan
Yang, Jing
Li, Xianyou
Yan, Weiran
author_facet Wu, Yichao
Liang, Penghao
Xiang, Yafei
Yuan, Mengwei
Liu, Jianan
Yang, Jing
Li, Xianyou
Yan, Weiran
contents Retrieval-Augmented Generation (RAG) grounds Large Language Models (LLMs) to mitigate factual hallucinations. Recent paradigms shift from static pipelines to Modular and Agentic RAG frameworks, granting models autonomy for multi-hop reasoning or self-correction. However, current reflective RAG heavily relies on massive LLMs as universal evaluators. In high-throughput systems, executing complete forward passes for billion-parameter models merely for binary routing introduces severe computational redundancy. Furthermore, in autonomous agent scenarios, inaccurate retrieval causes models to expend excessive tokens on spurious reasoning and redundant tool calls, inflating Time-to-First-Token (TTFT) and costs. We propose Tiny-Critic RAG, decoupling evaluation by deploying a parameter-efficient Small Language Model (SLM) via Low-Rank Adaptation (LoRA). Acting as a deterministic gatekeeper, Tiny-Critic employs constrained decoding and non-thinking inference modes for ultra-low latency binary routing. Evaluations on noise-injected datasets demonstrate Tiny-Critic RAG achieves routing accuracy comparable to GPT-4o-mini while reducing latency by an order of magnitude, establishing a highly cost-effective paradigm for agent deployment.
format Preprint
id arxiv_https___arxiv_org_abs_2603_00846
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Tiny-Critic RAG: Empowering Agentic Fallback with Parameter-Efficient Small Language Models
Wu, Yichao
Liang, Penghao
Xiang, Yafei
Yuan, Mengwei
Liu, Jianan
Yang, Jing
Li, Xianyou
Yan, Weiran
Information Retrieval
Machine Learning
Retrieval-Augmented Generation (RAG) grounds Large Language Models (LLMs) to mitigate factual hallucinations. Recent paradigms shift from static pipelines to Modular and Agentic RAG frameworks, granting models autonomy for multi-hop reasoning or self-correction. However, current reflective RAG heavily relies on massive LLMs as universal evaluators. In high-throughput systems, executing complete forward passes for billion-parameter models merely for binary routing introduces severe computational redundancy. Furthermore, in autonomous agent scenarios, inaccurate retrieval causes models to expend excessive tokens on spurious reasoning and redundant tool calls, inflating Time-to-First-Token (TTFT) and costs. We propose Tiny-Critic RAG, decoupling evaluation by deploying a parameter-efficient Small Language Model (SLM) via Low-Rank Adaptation (LoRA). Acting as a deterministic gatekeeper, Tiny-Critic employs constrained decoding and non-thinking inference modes for ultra-low latency binary routing. Evaluations on noise-injected datasets demonstrate Tiny-Critic RAG achieves routing accuracy comparable to GPT-4o-mini while reducing latency by an order of magnitude, establishing a highly cost-effective paradigm for agent deployment.
title Tiny-Critic RAG: Empowering Agentic Fallback with Parameter-Efficient Small Language Models
topic Information Retrieval
Machine Learning
url https://arxiv.org/abs/2603.00846