paper-with-me

홈 › Papers

Rotation-Preserving Supervised Fine-Tuning

2026-05-08 · Hangzhan Jin, Tianwei Ni, Lu Li, Pierre-Luc Bacon, Mohammad Hamdaqa, Doina Precup arxiv

Supervised fine-tuning (SFT) improves in-domain performance but can degrade out-of-domain (OOD) generalization. Prior work suggests that this degradation is related to changes in dominant singular subspaces of pretrained weight matrices. However, directly identifying loss-sensitive directions with Hessian or Fisher information is computationally expensive at LLM scale. In this work, we propose preserving projected rotations in pretrained singular subspaces as an efficient proxy for Fisher-sensitive directions, which we call Rotation-Preserving Supervised Fine-Tuning (RPSFT). RPSFT penalizes changes in the projected top-$k$ singular-vector block of each pretrained weight matrix, limiting unnecessary rotation while preserving task adaptation. Across model families and sizes trained on math reasoning data, RPSFT improves the in-domain/OOD trade-off over standard SFT and strong SFT baselines, better preserves pretrained representations, and provides stronger initializations for downstream RL fine-tuning. Code is available at \href{https://github.com/jinhangzhan/RPSFT.git}{https://github.com/jinhangzhan/RPSFT}.

📄 PDF Abstract BibTeX arXiv:2605.10973

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

RoSTE: An Efficient Quantization-Aware Supervised Fine-Tuning Approach for Large Language Models

2025-02-13 · Quan Wei, Chung-Yiu Yau, Hoi-To Wai, Yang 외

Supervised fine-tuning is a standard method for adapting pre-trained large language models (LLMs) to downstream tasks. Quantization has been recently studied as a post-training technique for efficient LLM deployment. To …

Quantization

HFBRI-MAE: Handcrafted Feature Based Rotation-Invariant Masked Autoencoder for 3D Point Cloud Analysis

2025-04-19 · Xuanhua Yin, Dingxin Zhang, Jianhui Yu, Weidong Cai

Self-supervised learning (SSL) has demonstrated remarkable success in 3D point cloud analysis, particularly through masked autoencoders (MAEs). However, existing MAE-based methods lack rotation invariance, leading to sig…

Few-Shot LearningSelf-Supervised Learning

HACMatch Semi-Supervised Rotation Regression with Hardness-Aware Curriculum Pseudo Labeling

2026-03-23 · Mei Li, Huayi Zhou, Suizhi Huang, Yuxiang Lu 외 arxiv

Regressing 3D rotations of objects from 2D images is a crucial yet challenging task, with broad applications in autonomous driving, virtual reality, and robotic control. Existing rotation regression models often rely on …

Autonomous DrivingData AugmentationPoint Clouds

Behavioral Mode Discovery for Fine-tuning Multimodal Generative Policies

2026-05-12 · Alberta Longhini, David Emukpere, Jean-Michel Renders, Seungsu Kim arxiv

We address the problem of fine-tuning pre-trained generative policies with reinforcement learning (RL) while preserving the multimodality of their action distributions. Existing methods for RL fine-tuning of generative p…

Reinforcement Learning

LP-SFT: Local-Preserving Supervised Fine-Tuning via Multimodal Entropy Structure

2026-07-06 · Yueyang Wang, Baolong Bi, Shuo Lu, Jingyuan Zhang arxiv

Supervised fine-tuning (SFT) is the standard approach for adapting pretrained language models to downstream domains, yet it often improves target-domain behavior at the cost of degrading pre-existing capabilities. Standa…