LEVI: Generalizable Fine-tuning via Layer-wise Ensemble of Different Views

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Roh, Yuji, Liu, Qingyun, Gui, Huan, Yuan, Zhe, Tang, Yujin, Whang, Steven Euijong, Liu, Liang, Bi, Shuchao, Hong, Lichan, Chi, Ed H., Zhao, Zhe
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866913397231058944
author Roh, Yuji
Liu, Qingyun
Gui, Huan
Yuan, Zhe
Tang, Yujin
Whang, Steven Euijong
Liu, Liang
Bi, Shuchao
Hong, Lichan
Chi, Ed H.
Zhao, Zhe
author_facet Roh, Yuji
Liu, Qingyun
Gui, Huan
Yuan, Zhe
Tang, Yujin
Whang, Steven Euijong
Liu, Liang
Bi, Shuchao
Hong, Lichan
Chi, Ed H.
Zhao, Zhe
contents Fine-tuning is becoming widely used for leveraging the power of pre-trained foundation models in new downstream tasks. While there are many successes of fine-tuning on various tasks, recent studies have observed challenges in the generalization of fine-tuned models to unseen distributions (i.e., out-of-distribution; OOD). To improve OOD generalization, some previous studies identify the limitations of fine-tuning data and regulate fine-tuning to preserve the general representation learned from pre-training data. However, potential limitations in the pre-training data and models are often ignored. In this paper, we contend that overly relying on the pre-trained representation may hinder fine-tuning from learning essential representations for downstream tasks and thus hurt its OOD generalization. It can be especially catastrophic when new tasks are from different (sub)domains compared to pre-training data. To address the issues in both pre-training and fine-tuning data, we propose a novel generalizable fine-tuning method LEVI (Layer-wise Ensemble of different VIews), where the pre-trained model is adaptively ensembled layer-wise with a small task-specific model, while preserving its efficiencies. By combining two complementing models, LEVI effectively suppresses problematic features in both the fine-tuning data and pre-trained model and preserves useful features for new tasks. Broad experiments with large language and vision models show that LEVI greatly improves fine-tuning generalization via emphasizing different views from fine-tuning data and pre-trained features.
format Preprint
id arxiv_https___arxiv_org_abs_2402_04644
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle LEVI: Generalizable Fine-tuning via Layer-wise Ensemble of Different Views
Roh, Yuji
Liu, Qingyun
Gui, Huan
Yuan, Zhe
Tang, Yujin
Whang, Steven Euijong
Liu, Liang
Bi, Shuchao
Hong, Lichan
Chi, Ed H.
Zhao, Zhe
Machine Learning
Artificial Intelligence
Fine-tuning is becoming widely used for leveraging the power of pre-trained foundation models in new downstream tasks. While there are many successes of fine-tuning on various tasks, recent studies have observed challenges in the generalization of fine-tuned models to unseen distributions (i.e., out-of-distribution; OOD). To improve OOD generalization, some previous studies identify the limitations of fine-tuning data and regulate fine-tuning to preserve the general representation learned from pre-training data. However, potential limitations in the pre-training data and models are often ignored. In this paper, we contend that overly relying on the pre-trained representation may hinder fine-tuning from learning essential representations for downstream tasks and thus hurt its OOD generalization. It can be especially catastrophic when new tasks are from different (sub)domains compared to pre-training data. To address the issues in both pre-training and fine-tuning data, we propose a novel generalizable fine-tuning method LEVI (Layer-wise Ensemble of different VIews), where the pre-trained model is adaptively ensembled layer-wise with a small task-specific model, while preserving its efficiencies. By combining two complementing models, LEVI effectively suppresses problematic features in both the fine-tuning data and pre-trained model and preserves useful features for new tasks. Broad experiments with large language and vision models show that LEVI greatly improves fine-tuning generalization via emphasizing different views from fine-tuning data and pre-trained features.
title LEVI: Generalizable Fine-tuning via Layer-wise Ensemble of Different Views
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2402.04644