Saved in:
Bibliographic Details
Main Authors: Zeng, Xinyue, Wang, Haohui, Lin, Junhong, Wu, Jun, Cody, Tyler, Zhou, Dawei
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.03793
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910977278083072
author Zeng, Xinyue
Wang, Haohui
Lin, Junhong
Wu, Jun
Cody, Tyler
Zhou, Dawei
author_facet Zeng, Xinyue
Wang, Haohui
Lin, Junhong
Wu, Jun
Cody, Tyler
Zhou, Dawei
contents The proliferation of open-sourced Large Language Models (LLMs) and diverse downstream tasks necessitates efficient model selection, given the impracticality of fine-tuning all candidates due to computational constraints. Despite the recent advances in LLM selection, a fundamental research question largely remains nascent: how can we model the dynamic behaviors of LLMs during fine-tuning, thereby enhancing our understanding of their generalization performance across diverse downstream tasks? In this work, we propose a novel theoretical framework that provides a proper lens to assess the generalization capabilities of LLMs, thereby enabling accurate and efficient LLM selection for downstream applications. In particular, we first derive a PAC-Bayesian Generalization Bound that unveils fine-tuning dynamics of LLMs and then introduce LENSLLM, a Neural Tangent Kernel (NTK)-based Rectified Scaling Model that enables accurate performance predictions across diverse tasks while maintaining computational efficiency. Extensive empirical results on 3 large-scale benchmarks demonstrate that our model achieves up to 91.1% accuracy and reduces up to 88.5% computational cost in LLM selection, outperforming 5 state-of-the-art methods. We open-source our proposed LENSLLM model and corresponding results at LensLLM.io.
format Preprint
id arxiv_https___arxiv_org_abs_2505_03793
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection
Zeng, Xinyue
Wang, Haohui
Lin, Junhong
Wu, Jun
Cody, Tyler
Zhou, Dawei
Machine Learning
Artificial Intelligence
The proliferation of open-sourced Large Language Models (LLMs) and diverse downstream tasks necessitates efficient model selection, given the impracticality of fine-tuning all candidates due to computational constraints. Despite the recent advances in LLM selection, a fundamental research question largely remains nascent: how can we model the dynamic behaviors of LLMs during fine-tuning, thereby enhancing our understanding of their generalization performance across diverse downstream tasks? In this work, we propose a novel theoretical framework that provides a proper lens to assess the generalization capabilities of LLMs, thereby enabling accurate and efficient LLM selection for downstream applications. In particular, we first derive a PAC-Bayesian Generalization Bound that unveils fine-tuning dynamics of LLMs and then introduce LENSLLM, a Neural Tangent Kernel (NTK)-based Rectified Scaling Model that enables accurate performance predictions across diverse tasks while maintaining computational efficiency. Extensive empirical results on 3 large-scale benchmarks demonstrate that our model achieves up to 91.1% accuracy and reduces up to 88.5% computational cost in LLM selection, outperforming 5 state-of-the-art methods. We open-source our proposed LENSLLM model and corresponding results at LensLLM.io.
title LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2505.03793