Theoretical Constraints on the Expressive Power of $\mathsf{RoPE}$-based Tensor Attention Transformers

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Xiaoyu, Liang, Yingyu, Shi, Zhenmei, Song, Zhao, Wan, Mingda
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913624709136384
author Li, Xiaoyu
Liang, Yingyu
Shi, Zhenmei
Song, Zhao
Wan, Mingda
author_facet Li, Xiaoyu
Liang, Yingyu
Shi, Zhenmei
Song, Zhao
Wan, Mingda
contents Tensor Attention extends traditional attention mechanisms by capturing high-order correlations across multiple modalities, addressing the limitations of classical matrix-based attention. Meanwhile, Rotary Position Embedding ($\mathsf{RoPE}$) has shown superior performance in encoding positional information in long-context scenarios, significantly enhancing transformer models' expressiveness. Despite these empirical successes, the theoretical limitations of these technologies remain underexplored. In this study, we analyze the circuit complexity of Tensor Attention and $\mathsf{RoPE}$-based Tensor Attention, showing that with polynomial precision, constant-depth layers, and linear or sublinear hidden dimension, they cannot solve fixed membership problems or $(A_{F,r})^*$ closure problems, under the assumption that $\mathsf{TC}^0 \neq \mathsf{NC}^1$. These findings highlight a gap between the empirical performance and theoretical constraints of Tensor Attention and $\mathsf{RoPE}$-based Tensor Attention Transformers, offering insights that could guide the development of more theoretically grounded approaches to Transformer model design and scaling.
format Preprint
id arxiv_https___arxiv_org_abs_2412_18040
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Theoretical Constraints on the Expressive Power of $\mathsf{RoPE}$-based Tensor Attention Transformers
Li, Xiaoyu
Liang, Yingyu
Shi, Zhenmei
Song, Zhao
Wan, Mingda
Machine Learning
Artificial Intelligence
Computational Complexity
Computation and Language
Tensor Attention extends traditional attention mechanisms by capturing high-order correlations across multiple modalities, addressing the limitations of classical matrix-based attention. Meanwhile, Rotary Position Embedding ($\mathsf{RoPE}$) has shown superior performance in encoding positional information in long-context scenarios, significantly enhancing transformer models' expressiveness. Despite these empirical successes, the theoretical limitations of these technologies remain underexplored. In this study, we analyze the circuit complexity of Tensor Attention and $\mathsf{RoPE}$-based Tensor Attention, showing that with polynomial precision, constant-depth layers, and linear or sublinear hidden dimension, they cannot solve fixed membership problems or $(A_{F,r})^*$ closure problems, under the assumption that $\mathsf{TC}^0 \neq \mathsf{NC}^1$. These findings highlight a gap between the empirical performance and theoretical constraints of Tensor Attention and $\mathsf{RoPE}$-based Tensor Attention Transformers, offering insights that could guide the development of more theoretically grounded approaches to Transformer model design and scaling.
title Theoretical Constraints on the Expressive Power of $\mathsf{RoPE}$-based Tensor Attention Transformers
topic Machine Learning
Artificial Intelligence
Computational Complexity
Computation and Language
url https://arxiv.org/abs/2412.18040