ORAL: Prompting Your Large-Scale LoRAs via Conditional Recurrent Diffusion

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Khan, Rana Muhammad Shahroz, Tang, Dongwen, Li, Pingzhi, Wang, Kai, Chen, Tianlong
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866912317067755520
author Khan, Rana Muhammad Shahroz
Tang, Dongwen
Li, Pingzhi
Wang, Kai
Chen, Tianlong
author_facet Khan, Rana Muhammad Shahroz
Tang, Dongwen
Li, Pingzhi
Wang, Kai
Chen, Tianlong
contents Parameter generation has emerged as a novel paradigm for neural network development, offering an alternative to traditional neural network training by synthesizing high-quality model weights directly. In the context of Low-Rank Adaptation (LoRA) for evolving ($\textit{i.e.}$, constantly updated) large language models (LLMs), this approach promises efficient adaptation without costly retraining. However, existing methods face critical limitations in simultaneously achieving scalability and controllability. In this paper, we introduce $\texttt{ORAL}$, a novel $\textbf{conditional recurrent diffusion}$ framework that addresses these challenges. $\texttt{ORAL}$ incorporates a novel conditioning mechanism that integrates model architecture and textual task specifications, enabling the generation of task-specific LoRA parameters that can seamlessly transfer across evolving foundation models. Our approach successfully scales to billions-of-parameter LLMs and maintains controllability. Through extensive experiments across seven language tasks, four vision tasks, and three multimodal tasks using five pre-trained LLMs, we demonstrate that $\texttt{ORAL}$ generates high-quality LoRA parameters that achieve comparable or superior performance to vanilla trained counterparts.
format Preprint
id arxiv_https___arxiv_org_abs_2503_24354
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ORAL: Prompting Your Large-Scale LoRAs via Conditional Recurrent Diffusion
Khan, Rana Muhammad Shahroz
Tang, Dongwen
Li, Pingzhi
Wang, Kai
Chen, Tianlong
Machine Learning
Artificial Intelligence
Computation and Language
Computer Vision and Pattern Recognition
Parameter generation has emerged as a novel paradigm for neural network development, offering an alternative to traditional neural network training by synthesizing high-quality model weights directly. In the context of Low-Rank Adaptation (LoRA) for evolving ($\textit{i.e.}$, constantly updated) large language models (LLMs), this approach promises efficient adaptation without costly retraining. However, existing methods face critical limitations in simultaneously achieving scalability and controllability. In this paper, we introduce $\texttt{ORAL}$, a novel $\textbf{conditional recurrent diffusion}$ framework that addresses these challenges. $\texttt{ORAL}$ incorporates a novel conditioning mechanism that integrates model architecture and textual task specifications, enabling the generation of task-specific LoRA parameters that can seamlessly transfer across evolving foundation models. Our approach successfully scales to billions-of-parameter LLMs and maintains controllability. Through extensive experiments across seven language tasks, four vision tasks, and three multimodal tasks using five pre-trained LLMs, we demonstrate that $\texttt{ORAL}$ generates high-quality LoRA parameters that achieve comparable or superior performance to vanilla trained counterparts.
title ORAL: Prompting Your Large-Scale LoRAs via Conditional Recurrent Diffusion
topic Machine Learning
Artificial Intelligence
Computation and Language
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2503.24354