ARMOR: Adaptive Retriever Optimization for Low-Resource Telecom Question Answering

2026-06-28

AI Summary

This work addresses the challenge of low-resource question answering in the telecommunications domain, where accurate answers heavily rely on dispersed technical evidence. Conventional fine-tuning of generative models often leads to overfitting and degrades their general-purpose capabilities. To mitigate this, the authors propose ARMOR, a method that freezes the generator and instead optimizes only the query-side retriever. ARMOR jointly trains the retriever using a combination of RAG likelihood and an InfoNCE contrastive objective, enhanced with a learnable temperature parameter and regularization applied to the base encoder to improve retrieval efficiency and robustness. Experimental results demonstrate that ARMOR significantly boosts evidence recall and answer generation quality across multiple telecommunications-domain benchmarks.

Abstract

Telecom question answering (QA) is a challenging setting for retrieval-augmented generation (RAG): evidence is fragmented across standards, papers, encyclopedic resources, and web documents, and answers often hinge on technical tables, equations, and specialized protocol language. In low-resource subdomains, generator fine-tuning can over-specialize and degrade general capability, making query-side retriever adaptation an attractive alternative. To this end, we ask whether a fixed-generator, query-adapted RAG system can outperform generator-side adaptation, and which retriever objectives best support that setting. We motivate retrieval, rather than generator fine-tuning, as the adaptation target through a capacity comparison: under bounded-parameter and soft-retrieval assumptions, query-encoder tuning can have a smaller estimation term than supervised fine-tuning when its effective dimension is smaller. We identify two particularly relevant objectives -- the latent-document RAG likelihood, which optimizes generation utility, and the InfoNCE contrastive objective, which improves semantic retrieval geometry -- and leverage them jointly through a retriever optimization method targeting downstream QA performance in the telecom domain. Specifically, we introduce ARMOR, Adaptive Regularized Mixture Optimization for Retrievers, which learns separate temperatures for the RAG retrieval distribution and InfoNCE softmax and regularizes the adapted query encoder toward the frozen base query encoder. Across telecom-specific retrieval and generative QA benchmarks, we show that ARMOR improves evidence retrieval and answer generation in several in-domain settings. Code is available at https://github.com/heshandevaka/ARMOR.git.

Problem

• telecom question answering
• retrieval-augmented generation
• low-resource
• retriever adaptation
• query-side optimization

Innovation

• retriever adaptation
• low-resource QA
• RAG optimization
• InfoNCE
• query encoder tuning