Enregistré dans:
Détails bibliographiques
Auteur principal: Nunepalli, Bharath
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2601.00841
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866911350855303168
author Nunepalli, Bharath
author_facet Nunepalli, Bharath
contents Retrieval-augmented generation (RAG) introduces a practical control problem: retrieval depth and generation behavior must be chosen per query to satisfy service-level objectives (SLOs) such as cost, refusal rate, and hallucination risk. This work models per-query control as a small discrete action: choose a retrieval depth and a generation mode (guarded vs. auto), or refuse. An offline logged dataset is constructed from SQuAD 2.0 by executing each action and recording accuracy, token cost, hallucination/refusal indicators, and an SLO-weighted reward. Two simple policy-learning objectives are evaluated: supervised classification of the per-state best action (Argmax-CE) and a reward-weighted variant (Argmax-CE-WT). Across the evaluated settings, a strong fixed baseline (low k, guarded prompting) performs competitively; learned policies mainly provide additional cost savings under a quality-focused SLO and can exhibit refusal collapse under a cheap SLO when refusal is heavily rewarded. The contribution is a reproducible case study of SLO-aware control for RAG pipelines, emphasizing failure modes and reporting conventions rather than proposing a new retriever or language model.
format Preprint
id arxiv_https___arxiv_org_abs_2601_00841
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SLO-Conditioned Action Routing for Retrieval-Augmented Generation: Objective Ablation and Failure Modes
Nunepalli, Bharath
Machine Learning
Retrieval-augmented generation (RAG) introduces a practical control problem: retrieval depth and generation behavior must be chosen per query to satisfy service-level objectives (SLOs) such as cost, refusal rate, and hallucination risk. This work models per-query control as a small discrete action: choose a retrieval depth and a generation mode (guarded vs. auto), or refuse. An offline logged dataset is constructed from SQuAD 2.0 by executing each action and recording accuracy, token cost, hallucination/refusal indicators, and an SLO-weighted reward. Two simple policy-learning objectives are evaluated: supervised classification of the per-state best action (Argmax-CE) and a reward-weighted variant (Argmax-CE-WT). Across the evaluated settings, a strong fixed baseline (low k, guarded prompting) performs competitively; learned policies mainly provide additional cost savings under a quality-focused SLO and can exhibit refusal collapse under a cheap SLO when refusal is heavily rewarded. The contribution is a reproducible case study of SLO-aware control for RAG pipelines, emphasizing failure modes and reporting conventions rather than proposing a new retriever or language model.
title SLO-Conditioned Action Routing for Retrieval-Augmented Generation: Objective Ablation and Failure Modes
topic Machine Learning
url https://arxiv.org/abs/2601.00841