English

Vcc: Scaling Transformers to 128K Tokens or More by Prioritizing Important Tokens

Computation and Language 2023-05-30 v2 Machine Learning

Abstract

Transformers are central in modern natural language processing and computer vision applications. Despite recent works devoted to reducing the quadratic cost of such models (as a function of the sequence length), dealing with ultra long sequences (e.g., with more than 16K tokens) remains challenging. Applications such as answering questions based on a book or summarizing a scientific article are inefficient or infeasible. Here, we propose to significantly improve the efficiency of Transformers for ultra long sequences, by compressing the sequence into a much smaller representation at each layer. Specifically, by exploiting the fact that in many tasks, only a small subset of special tokens (we call VIP-tokens) are most relevant to the final prediction, we propose a VIP-token centric compression (VCC) scheme which selectively compresses the sequence based on their impact on approximating the representation of the VIP-tokens. Compared with competitive baselines, our algorithm is not only efficient (achieving more than 3×3\times efficiency gain compared to baselines on 4K and 16K lengths), but also offers competitive/better performance on a large number of tasks. Further, we show that our algorithm scales to 128K tokens (or more) while consistently offering accuracy improvement.

Keywords

Cite

@article{arxiv.2305.04241,
  title  = {Vcc: Scaling Transformers to 128K Tokens or More by Prioritizing Important Tokens},
  author = {Zhanpeng Zeng and Cole Hawkins and Mingyi Hong and Aston Zhang and Nikolaos Pappas and Vikas Singh and Shuai Zheng},
  journal= {arXiv preprint arXiv:2305.04241},
  year   = {2023}
}

Comments

10 pages main text, 12 pages appendix, preprint