Efficient Resource-Constrained Training of Transformers via Subspace Optimization
Le-Trung Nguyen, Enzo Tartaglione, Van-Tam Nguyen
We propose a novel method that enables training vision transformer models within a low-rank subspace to optimize computational resources, making on-device learning practically feasible.
Abstract
As AI increasingly shapes daily life, energy consumption and data privacy have become pressing concerns. On-device learning trains models directly on edge devices, cutting energy consumption and safeguarding data privacy. However, the expanding scale of modern neural networks creates a major obstacle for on-device training. Although prior work has concentrated on compact convolutional architectures, we instead apply subspace-based training to transformer models. Motivated by the idea that a model's essential information lies in a fixed subspace, we introduce Weight-Activation Subspace Iteration (WASI), a method that mitigates the memory bottleneck of backpropagation and boosts inference efficiency in transformer models by restricting training to this subspace. Our results demonstrate that WASI maintains accuracy comparable to vanilla training while reducing memory usage by up to $62\times$ and computational cost (FLOPs) by up to $2\times$. On a Raspberry Pi 5, WASI achieves roughly $1.4\times$ faster training and inference than vanilla training. The code is available at https://github.com/Le-TrungNguyen/ICLR2026-WASI.git.
WASI applies subspace-based training to transformer models reducing memory by 62x and FLOPs by 2x while maintaining accuracy on edge devices.
- Weight-Activation Subspace Iteration method restricting training to fixed low-dimensional subspace
- Reduces memory usage by up to 62x and computational cost by up to 2x
- Achieves 1.4x speedup over vanilla training on Raspberry Pi 5
- Subspace optimization
- SVD
- Transformer training
- On-device learning
- Low-rank approximation
Experiments primarily focused on vision tasks; hardware limitations prevent evaluation of larger-scale models
from the paper
Extend to broader range of tasks with emphasis on LLMs
from the paper
Author keywords
- Deep Learning
- Computer Vision
- Compression
- Low rank
Related orals
TileLang: Bridge Programmability and Performance in Modern Neural Kernels
TileLang enables hardware-aware fused kernel programming with tile inference and recommendation achieving 5-6x speedup.
Probabilistic Kernel Function for Fast Angle Testing
Proposes probabilistic kernel functions for angle testing enabling efficient approximate nearest neighbor search.
SANA-Video: Efficient Video Generation with Block Linear Diffusion Transformer
Generates minute-long high-resolution videos efficiently with linear attention and constant-memory KV cache for block autoregression.
Why Low-Precision Transformer Training Fails: An Analysis on Flash Attention
Analyzes low-precision flash attention training failure caused by low-rank representations and biased BF16 rounding errors.
Speculative Actions: A Lossless Framework for Faster AI Agents
Speculative Actions accelerates agent systems by predicting and executing likely future actions in parallel.