Data-Informed Global Sparseness in Attention Mechanisms for Deep Neural Networks

Abstract

Attention mechanisms play a crucial role in the neural revolution of Natural Language Processing (NLP). With the growth of attention-based models, several pruning techniques have been developed to identify and exploit sparseness, making these models more efficient. Most efforts focus on hard-coding attention patterns or pruning attention weights based on training data. We propose Attention Pruning (AP), a framework that observes attention patterns in a fixed dataset and generates a global sparseness mask. AP saves 90% of attention computation for language modeling and about 50% for machine translation and GLUE tasks, maintaining result quality. Our method reveals important distinctions between self- and cross-attention patterns, guiding future NLP research. Our framework can reduce both latency and memory requirements for any attention-based model, aiding in the development of improved models for existing or new NLP applications. We have demonstrated this with encoder and autoregressive transformer models using Triton GPU kernels and make our code publicly available at https://github.com/irugina/AP.

Presented at LREC-COLING 2024: 12 pages, 4 figures, 11 tables

References in corpus (11)

• Language Models are Few-Shot Learners
• Teaching Machines to Read and Comprehend
• Longformer: The Long-Document Transformer
• Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism
• SQuAD: 100,000+ Questions for Machine Comprehension of Text
• GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding
• Generating Long Sequences with Sparse Transformers
• Reformer: The Efficient Transformer
• Interactive Supercomputing on 40,000 Cores for Machine Learning and Data Analysis
• Big Bird: Transformers for Longer Sequences
• Playing the lottery with rewards and multiple languages: lottery tickets in RL and NLP