Saved in:
Bibliographic Details
Main Authors: Sundararaman, Dhanasekar, Li, Keying, Xiong, Wayne, Garg, Aashna
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2512.06239
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914183025524736
author Sundararaman, Dhanasekar
Li, Keying
Xiong, Wayne
Garg, Aashna
author_facet Sundararaman, Dhanasekar
Li, Keying
Xiong, Wayne
Garg, Aashna
contents We present LOCUS (LOw-cost Customization for Universal Specialization), a pipeline that consumes few-shot data to streamline the construction and training of NLP models through targeted retrieval, synthetic data generation, and parameter-efficient tuning. With only a small number of labeled examples, LOCUS discovers pertinent data in a broad repository, synthesizes additional training samples via in-context data generation, and fine-tunes models using either full or low-rank (LoRA) parameter adaptation. Our approach targets named entity recognition (NER) and text classification (TC) benchmarks, consistently outperforming strong baselines (including GPT-4o) while substantially lowering costs and model sizes. Our resultant memory-optimized models retain 99% of fully fine-tuned accuracy while using barely 5% of the memory footprint, also beating GPT-4o on several benchmarks with less than 1% of its parameters.
format Preprint
id arxiv_https___arxiv_org_abs_2512_06239
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle LOCUS: A System and Method for Low-Cost Customization for Universal Specialization
Sundararaman, Dhanasekar
Li, Keying
Xiong, Wayne
Garg, Aashna
Computation and Language
We present LOCUS (LOw-cost Customization for Universal Specialization), a pipeline that consumes few-shot data to streamline the construction and training of NLP models through targeted retrieval, synthetic data generation, and parameter-efficient tuning. With only a small number of labeled examples, LOCUS discovers pertinent data in a broad repository, synthesizes additional training samples via in-context data generation, and fine-tunes models using either full or low-rank (LoRA) parameter adaptation. Our approach targets named entity recognition (NER) and text classification (TC) benchmarks, consistently outperforming strong baselines (including GPT-4o) while substantially lowering costs and model sizes. Our resultant memory-optimized models retain 99% of fully fine-tuned accuracy while using barely 5% of the memory footprint, also beating GPT-4o on several benchmarks with less than 1% of its parameters.
title LOCUS: A System and Method for Low-Cost Customization for Universal Specialization
topic Computation and Language
url https://arxiv.org/abs/2512.06239