Saved in:
Bibliographic Details
Main Authors: Fan, Chenghao, Lu, Zhenyi, Wei, Wei, Tian, Jie, Qu, Xiaoye, Chen, Dangyang, Cheng, Yu
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2406.15480
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910647802920960
author Fan, Chenghao
Lu, Zhenyi
Wei, Wei
Tian, Jie
Qu, Xiaoye
Chen, Dangyang
Cheng, Yu
author_facet Fan, Chenghao
Lu, Zhenyi
Wei, Wei
Tian, Jie
Qu, Xiaoye
Chen, Dangyang
Cheng, Yu
contents Efficient fine-tuning of large language models for task-specific applications is imperative, yet the vast number of parameters in these models makes their training increasingly challenging. Despite numerous proposals for effective methods, a substantial memory overhead remains for gradient computations during updates. \thm{Can we fine-tune a series of task-specific small models and transfer their knowledge directly to a much larger model without additional training?} In this paper, we explore weak-to-strong specialization using logit arithmetic, facilitating a direct answer to this question. Existing weak-to-strong methods often employ a static knowledge transfer ratio and a single small model for transferring complex knowledge, which leads to suboptimal performance. % To address this, To surmount these limitations, we propose a dynamic logit fusion approach that works with a series of task-specific small models, each specialized in a different task. This method adaptively allocates weights among these models at each decoding step, learning the weights through Kullback-Leibler divergence constrained optimization problems. We conduct extensive experiments across various benchmarks in both single-task and multi-task settings, achieving leading results. By transferring expertise from the 7B model to the 13B model, our method closes the performance gap by 96.4\% in single-task scenarios and by 86.3\% in multi-task scenarios compared to full fine-tuning of the 13B model. Notably, we achieve surpassing performance on unseen tasks. Moreover, we further demonstrate that our method can effortlessly integrate in-context learning for single tasks and task arithmetic for multi-task scenarios.
format Preprint
id arxiv_https___arxiv_org_abs_2406_15480
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle On Giant's Shoulders: Effortless Weak to Strong by Dynamic Logits Fusion
Fan, Chenghao
Lu, Zhenyi
Wei, Wei
Tian, Jie
Qu, Xiaoye
Chen, Dangyang
Cheng, Yu
Computation and Language
Artificial Intelligence
Machine Learning
Efficient fine-tuning of large language models for task-specific applications is imperative, yet the vast number of parameters in these models makes their training increasingly challenging. Despite numerous proposals for effective methods, a substantial memory overhead remains for gradient computations during updates. \thm{Can we fine-tune a series of task-specific small models and transfer their knowledge directly to a much larger model without additional training?} In this paper, we explore weak-to-strong specialization using logit arithmetic, facilitating a direct answer to this question. Existing weak-to-strong methods often employ a static knowledge transfer ratio and a single small model for transferring complex knowledge, which leads to suboptimal performance. % To address this, To surmount these limitations, we propose a dynamic logit fusion approach that works with a series of task-specific small models, each specialized in a different task. This method adaptively allocates weights among these models at each decoding step, learning the weights through Kullback-Leibler divergence constrained optimization problems. We conduct extensive experiments across various benchmarks in both single-task and multi-task settings, achieving leading results. By transferring expertise from the 7B model to the 13B model, our method closes the performance gap by 96.4\% in single-task scenarios and by 86.3\% in multi-task scenarios compared to full fine-tuning of the 13B model. Notably, we achieve surpassing performance on unseen tasks. Moreover, we further demonstrate that our method can effortlessly integrate in-context learning for single tasks and task arithmetic for multi-task scenarios.
title On Giant's Shoulders: Effortless Weak to Strong by Dynamic Logits Fusion
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2406.15480