DEFT: Data Efficient Fine-Tuning for Pre-Trained Language Models via Unsupervised Core-Set Selection

2023/10/25 by Devleena Das, Das, Devleena, Vivek Khetan +1 · 3 citations
Computer Science · #Artificial Intelligence (cs.AI) #Computation and Language (cs.CL) #FOS: Computer and information sciences #Multimodal Machine Learning Applications #Natural Language Processing Techniques #Topic Modeling

paper · pdf · doi:10.48550/arxiv.2310.16776

openalex publication_date 2023/10/25 · openalex created_date 2023/10/28 · openalex updated_date 2026/07/28

Abstract

Recent advances have led to the availability of many pre-trained language models (PLMs); however, a question that remains is how much data is truly needed to fine-tune PLMs for downstream tasks? In this work, we introduce DEFT-UCS, a data-efficient fine-tuning framework that leverages unsupervised core-set selection to identify a smaller, representative dataset that reduces the amount of data needed to fine-tune PLMs for downstream tasks. We examine the efficacy of DEFT-UCS in the context of text-editing LMs, and compare to the state-of-the art text-editing model, CoEDIT. Our results demonstrate that DEFT-UCS models are just as accurate as CoEDIT, across eight different datasets consisting of six different editing tasks, while finetuned on 70% less data.

Cited by

• Enhancing Training Data Attribution with Representational Optimization
• Data-Efficient Hate Speech Detection via Cross-Lingual Nearest Neighbor Retrieval with Limited Labeled Data
• GemMaroc: Unlocking Darija Proficiency in LLMs with Minimal Data

Related

• SHED: Shapley-Based Automated Dataset Refinement for Instruction Fine-Tuning
• DELIFT: Data Efficient Language model Instruction Fine Tuning
• Fine-tuning Done Right in Model Editing
• Is Fine-Tuning an Effective Solution? Reassessing Knowledge Editing for Unstructured Data
• Scalable Fine-tuning from Multiple Data Sources: A First-Order Approximation Approach