LoRA-Mini : Adaptation Matrices Decomposition and Selective Training

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Singh, Ayush, Aher, Rajdeep, Garg, Shivank
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917847617241088
author Singh, Ayush
Aher, Rajdeep
Garg, Shivank
author_facet Singh, Ayush
Aher, Rajdeep
Garg, Shivank
contents The rapid advancements in large language models (LLMs) have revolutionized natural language processing, creating an increased need for efficient, task-specific fine-tuning methods. Traditional fine-tuning of LLMs involves updating a large number of parameters, which is computationally expensive and memory-intensive. Low-Rank Adaptation (LoRA) has emerged as a promising solution, enabling parameter-efficient fine-tuning by reducing the number of trainable parameters. However, while LoRA reduces the number of trainable parameters, LoRA modules still create significant storage challenges. We propose LoRA-Mini, an optimized adaptation of LoRA that improves parameter efficiency by splitting low-rank matrices into four parts, with only the two inner matrices being trainable. This approach achieves upto a 20x reduction compared to standard LoRA in the number of trainable parameters while preserving performance levels comparable to standard LoRA, addressing both computational and storage efficiency in LLM fine-tuning.
format Preprint
id arxiv_https___arxiv_org_abs_2411_15804
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle LoRA-Mini : Adaptation Matrices Decomposition and Selective Training
Singh, Ayush
Aher, Rajdeep
Garg, Shivank
Computation and Language
Artificial Intelligence
Machine Learning
The rapid advancements in large language models (LLMs) have revolutionized natural language processing, creating an increased need for efficient, task-specific fine-tuning methods. Traditional fine-tuning of LLMs involves updating a large number of parameters, which is computationally expensive and memory-intensive. Low-Rank Adaptation (LoRA) has emerged as a promising solution, enabling parameter-efficient fine-tuning by reducing the number of trainable parameters. However, while LoRA reduces the number of trainable parameters, LoRA modules still create significant storage challenges. We propose LoRA-Mini, an optimized adaptation of LoRA that improves parameter efficiency by splitting low-rank matrices into four parts, with only the two inner matrices being trainable. This approach achieves upto a 20x reduction compared to standard LoRA in the number of trainable parameters while preserving performance levels comparable to standard LoRA, addressing both computational and storage efficiency in LLM fine-tuning.
title LoRA-Mini : Adaptation Matrices Decomposition and Selective Training
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2411.15804