NeuroPrune: A Neuro-inspired Topological Sparse Training Algorithm for Large Language Models
This addresses efficiency bottlenecks for researchers and practitioners using large language models, though it is an incremental improvement over existing sparsity methods.
The paper tackles the high computational cost of training and inference in large language models by introducing NeuroPrune, a neuro-inspired sparse training algorithm based on network topology principles like preferential attachment and synapse pruning. It achieves competitive or superior performance to baselines while being up to 10x faster in training time and improving inference efficiency across various NLP tasks.
Transformer-based Language Models have become ubiquitous in Natural Language Processing (NLP) due to their impressive performance on various tasks. However, expensive training as well as inference remains a significant impediment to their widespread applicability. While enforcing sparsity at various levels of the model architecture has found promise in addressing scaling and efficiency issues, there remains a disconnect between how sparsity affects network topology. Inspired by brain neuronal networks, we explore sparsity approaches through the lens of network topology. Specifically, we exploit mechanisms seen in biological networks, such as preferential attachment and redundant synapse pruning, and show that principled, model-agnostic sparsity approaches are performant and efficient across diverse NLP tasks, spanning both classification (such as natural language inference) and generation (summarization, machine translation), despite our sole objective not being optimizing performance. NeuroPrune is competitive with (or sometimes superior to) baselines on performance and can be up to $10$x faster in terms of training time for a given level of sparsity, simultaneously exhibiting measurable improvements in inference time in many cases.