arxiv:1911.05507

Compressive Transformers for Long-Range Sequence Modelling

Published on Nov 13, 2019

Authors:

Abstract

The Compressive Transformer, an attentive sequence model with memory compression, achieves top language modeling results and can be applied to speech and reinforcement learning tasks.

Generated by Qwen/Qwen2.5-Coder-32B-Instruct

We present the Compressive Transformer, an attentive sequence model which compresses past memories for long-range sequence learning. We find the Compressive Transformer obtains state-of-the-art language modelling results in the WikiText-103 and Enwik8 benchmarks, achieving 17.1 ppl and 0.97 bpc respectively. We also find it can model high-frequency speech effectively and can be used as a memory mechanism for RL, demonstrated on an object matching task. To promote the domain of long-range sequence learning, we propose a new open-vocabulary language modelling benchmark derived from books, PG-19.