Time-aware Large Kernel Convolutions
This addresses the computational bottleneck in sequence modeling for NLP applications, offering a more efficient alternative to attention and convolution methods.
The paper tackles the quadratic time complexity of attention-based sequence models by introducing Time-aware Large Kernel (TaLK) Convolutions, which predict kernel sizes adaptively to achieve linear time complexity, showing efficient improvements in machine translation, summarization, and language modeling tasks.
To date, most state-of-the-art sequence modeling architectures use attention to build generative models for language based tasks. Some of these models use all the available sequence tokens to generate an attention distribution which results in time complexity of $O(n^2)$. Alternatively, they utilize depthwise convolutions with softmax normalized kernels of size $k$ acting as a limited-window self-attention, resulting in time complexity of $O(k{\cdot}n)$. In this paper, we introduce Time-aware Large Kernel (TaLK) Convolutions, a novel adaptive convolution operation that learns to predict the size of a summation kernel instead of using a fixed-sized kernel matrix. This method yields a time complexity of $O(n)$, effectively making the sequence encoding process linear to the number of tokens. We evaluate the proposed method on large-scale standard machine translation, abstractive summarization and language modeling datasets and show that TaLK Convolutions constitute an efficient improvement over other attention/convolution based approaches.