Sequential Compression Layers for Efficient Federated Learning in Foundational Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Mahla, Navyansh, Gupta, Sunny, Sethi, Amit
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915187159728128
author Mahla, Navyansh
Gupta, Sunny
Sethi, Amit
author_facet Mahla, Navyansh
Gupta, Sunny
Sethi, Amit
contents Federated Learning (FL) has gained popularity for fine-tuning large language models (LLMs) across multiple nodes, each with its own private data. While LoRA has been widely adopted for parameter efficient federated fine-tuning, recent theoretical and empirical studies highlight its suboptimal performance in the federated learning context. In response, we propose a novel, simple, and more effective parameter-efficient fine-tuning method that does not rely on LoRA. Our approach introduces a small multi-layer perceptron (MLP) layer between two existing MLP layers the up proj (the FFN projection layer following the self-attention module) and down proj within the feed forward network of the transformer block. This solution addresses the bottlenecks associated with LoRA in federated fine tuning and outperforms recent LoRA-based approaches, demonstrating superior performance for both language models and vision encoders.
format Preprint
id arxiv_https___arxiv_org_abs_2412_07021
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Sequential Compression Layers for Efficient Federated Learning in Foundational Models
Mahla, Navyansh
Gupta, Sunny
Sethi, Amit
Machine Learning
Artificial Intelligence
Federated Learning (FL) has gained popularity for fine-tuning large language models (LLMs) across multiple nodes, each with its own private data. While LoRA has been widely adopted for parameter efficient federated fine-tuning, recent theoretical and empirical studies highlight its suboptimal performance in the federated learning context. In response, we propose a novel, simple, and more effective parameter-efficient fine-tuning method that does not rely on LoRA. Our approach introduces a small multi-layer perceptron (MLP) layer between two existing MLP layers the up proj (the FFN projection layer following the self-attention module) and down proj within the feed forward network of the transformer block. This solution addresses the bottlenecks associated with LoRA in federated fine tuning and outperforms recent LoRA-based approaches, demonstrating superior performance for both language models and vision encoders.
title Sequential Compression Layers for Efficient Federated Learning in Foundational Models
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2412.07021