detalle del documento
IDENTIFICACIÓN

oai:arXiv.org:2410.08305

Tema
Computer Science - Machine Learnin... Mathematics - Optimization and Con...
Autor
Malinovsky, Grigory Michieli, Umberto Hammoud, Hasan Abed Al Kader Ceritli, Taha Elesedy, Hayder Ozay, Mete Richtárik, Peter
Categoría

Computer Science

Año

2024

fecha de cotización

16/10/2024

Palabras clave
theoretical lora convergence fine-tuning
Métrico

Resumen

Fine-tuning has become a popular approach to adapting large foundational models to specific tasks.

As the size of models and datasets grows, parameter-efficient fine-tuning techniques are increasingly important.

One of the most widely used methods is Low-Rank Adaptation (LoRA), with adaptation update expressed as the product of two low-rank matrices.

While LoRA was shown to possess strong performance in fine-tuning, it often under-performs when compared to full-parameter fine-tuning (FPFT).

Although many variants of LoRA have been extensively studied empirically, their theoretical optimization analysis is heavily under-explored.

The starting point of our work is a demonstration that LoRA and its two extensions, Asymmetric LoRA and Chain of LoRA, indeed encounter convergence issues.

To address these issues, we propose Randomized Asymmetric Chain of LoRA (RAC-LoRA) -- a general optimization framework that rigorously analyzes the convergence rates of LoRA-based methods.

Our approach inherits the empirical benefits of LoRA-style heuristics, but introduces several small but important algorithmic modifications which turn it into a provably convergent method.

Our framework serves as a bridge between FPFT and low-rank adaptation.

We provide provable guarantees of convergence to the same solution as FPFT, along with the rate of convergence.

Additionally, we present a convergence analysis for smooth, non-convex loss functions, covering gradient descent, stochastic gradient descent, and federated learning settings.

Our theoretical findings are supported by experimental results.

;Comment: 36 pages, 4 figures, 2 algorithms

Malinovsky, Grigory,Michieli, Umberto,Hammoud, Hasan Abed Al Kader,Ceritli, Taha,Elesedy, Hayder,Ozay, Mete,Richtárik, Peter, 2024, Randomized Asymmetric Chain of LoRA: The First Meaningful Theoretical Framework for Low-Rank Adaptation

Documento

Abrir

Compartir

Fuente

Artículos recomendados por ES/IODE IA

Hespi: A pipeline for automatically detecting information from hebarium specimen sheets
science recognition institutional detects text-based text pipeline specimen