Linformer: Self-Attention with Linear Complexity
This addresses the problem of expensive training and deployment for long sequences in NLP applications, offering a more efficient alternative.
The paper tackles the high computational cost of standard self-attention in Transformers, which scales quadratically with sequence length, by proposing the Linformer, a new self-attention mechanism that reduces complexity to linear while maintaining performance comparable to standard Transformers.
Large transformer models have shown extraordinary success in achieving state-of-the-art results in many natural language processing applications. However, training and deploying these models can be prohibitively costly for long sequences, as the standard self-attention mechanism of the Transformer uses $O(n^2)$ time and space with respect to sequence length. In this paper, we demonstrate that the self-attention mechanism can be approximated by a low-rank matrix. We further exploit this finding to propose a new self-attention mechanism, which reduces the overall self-attention complexity from $O(n^2)$ to $O(n)$ in both time and space. The resulting linear transformer, the \textit{Linformer}, performs on par with standard Transformer models, while being much more memory- and time-efficient.