PLEX: Adaptive Parameter-Efficient Fine-Tuning for Code LLMs using Lottery-Tickets

이재성, 한호재, 김종윤, 황승원, 강나은, 안경준, 장성호

Abstract

Fine-tuning large language models (LLMs) for code generation is challenging due to computational costs and the underrepresentation of some programming languages (PLs) in pre-training. We propose PLEX, a lottery-ticket based parameter-efficient fine-tuning (PEFT) method that adapts LLMs to either well-supported and underrepresented PLs.During lottery ticket selection, PLEX employs a dual strategy: for well-represented PLs, it leverages the LLM’s full parametric knowledge by selecting from full layers, while for underrepresented PLs, it narrows the selection scope to dense layers, prioritizing the most influential parameters.Additionally, PLEX-E, a low-rank extension of PLEX, further reduces computational costs by limiting the scope of fine-tuning. On MultiPL-E benchmarks, PLEX achieves state-of-the-art performance among PEFT methods, while PLEX-E maintains competitive results with reduced computational overhead. Both variants demonstrate effective adaptation across diverse programming languages, particularly for those underrepresented in pre-training.

논문보기

공유하기