Blockwise Parallel Transformer for Large Context Models

TL;DR

The paper introduces Blockwise Parallel Transformer (BPT), a memory-efficient architecture enabling training of significantly longer sequences in NLP and reinforcement learning tasks, outperforming previous methods.

Contribution

BPT employs blockwise computation and network fusion to drastically reduce memory usage, allowing for longer sequence processing beyond existing models.

Findings

BPT enables training sequences 32 times longer than vanilla Transformers.

BPT reduces memory costs significantly compared to prior methods.

BPT improves performance on language modeling and reinforcement learning tasks.

Abstract

Transformers have emerged as the cornerstone of state-of-the-art natural language processing models, showcasing exceptional performance across a wide range of AI applications. However, the memory demands posed by the self-attention mechanism and the large feedforward network in Transformers limit their ability to handle long sequences, thereby creating challenges for tasks involving multiple long sequences or long-term dependencies. We present a distinct approach, Blockwise Parallel Transformer (BPT), that leverages blockwise computation of self-attention and feedforward network fusion to minimize memory costs. By processing longer input sequences while maintaining memory efficiency, BPT enables training sequences 32 times longer than vanilla Transformers and up to 4 times longer than previous memory-efficient methods. Extensive experiments on language modeling and reinforcement learning tasks demonstrate the effectiveness of BPT in reducing memory requirements and improving performance.