Transformer models have obtained remarkable accomplishments in various NLP tasks. However, these models have efficiency issues on long sequences, as the complexity of their self-attention module scales quadratically with the sequence length. To remedy the limitation, we present Memformer, a novel language model that utilizes a single unified memory to encode and retrieve past information. It includes a new optimization scheme, Memory Replay Back-Propagation, which promotes long-range back-propagation through time with a significantly reduced memory requirement. Memformer achieves $mathcal{O}(n)$ time complexity and $mathcal{O}(1)$ space complexity in processing long sequences, meaning that the model can handle an infinite length sequence during inference. Our model is also compatible with other self-supervised tasks to further improve the performance on language modeling. Experimental results show that Memformer outperforms the previous long-range sequence models on WikiText-103, including Transformer-XL and compressive Transformer.
Download