Saved in:
Bibliographic Details
Main Authors: Mao, Yansheng, Li, Jiaqi, Meng, Fanxu, Xiong, Jing, Zheng, Zilong, Zhang, Muhan
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2412.13626
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910750776229888
author Mao, Yansheng
Li, Jiaqi
Meng, Fanxu
Xiong, Jing
Zheng, Zilong
Zhang, Muhan
author_facet Mao, Yansheng
Li, Jiaqi
Meng, Fanxu
Xiong, Jing
Zheng, Zilong
Zhang, Muhan
contents Long context understanding remains challenging for large language models due to their limited context windows. This paper introduces Long Input Fine-Tuning (LIFT) for long context modeling, a novel framework that enhances LLM performance on long-context tasks by adapting model parameters to the context at test time. LIFT enables efficient processing of lengthy inputs without the computational burden of offline long-context adaptation, and can improve the long-context capabilities of arbitrary short-context models. The framework is further enhanced by integrating in-context learning and pre-LIFT supervised fine-tuning. The combination of in-context learning and LIFT enables short-context models like Llama 3 to handle arbitrarily long contexts and consistently improves their performance on popular long-context benchmarks like LooGLE and LongBench. We also provide a comprehensive analysis of the strengths and limitations of LIFT on long context understanding, offering valuable directions for future research.
format Preprint
id arxiv_https___arxiv_org_abs_2412_13626
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle LIFT: Improving Long Context Understanding Through Long Input Fine-Tuning
Mao, Yansheng
Li, Jiaqi
Meng, Fanxu
Xiong, Jing
Zheng, Zilong
Zhang, Muhan
Computation and Language
Artificial Intelligence
Long context understanding remains challenging for large language models due to their limited context windows. This paper introduces Long Input Fine-Tuning (LIFT) for long context modeling, a novel framework that enhances LLM performance on long-context tasks by adapting model parameters to the context at test time. LIFT enables efficient processing of lengthy inputs without the computational burden of offline long-context adaptation, and can improve the long-context capabilities of arbitrary short-context models. The framework is further enhanced by integrating in-context learning and pre-LIFT supervised fine-tuning. The combination of in-context learning and LIFT enables short-context models like Llama 3 to handle arbitrarily long contexts and consistently improves their performance on popular long-context benchmarks like LooGLE and LongBench. We also provide a comprehensive analysis of the strengths and limitations of LIFT on long context understanding, offering valuable directions for future research.
title LIFT: Improving Long Context Understanding Through Long Input Fine-Tuning
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2412.13626