NeuroPrune: A Neuro-inspired Topological Sparse Training Algorithm for Large Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Dhurandhar, Amit, Pedapati, Tejaswini, Luss, Ronny, Dan, Soham, Lozano, Aurelie, Das, Payel, Kollias, Georgios
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909217553645568
author Dhurandhar, Amit
Pedapati, Tejaswini
Luss, Ronny
Dan, Soham
Lozano, Aurelie
Das, Payel
Kollias, Georgios
author_facet Dhurandhar, Amit
Pedapati, Tejaswini
Luss, Ronny
Dan, Soham
Lozano, Aurelie
Das, Payel
Kollias, Georgios
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2404_01306
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle NeuroPrune: A Neuro-inspired Topological Sparse Training Algorithm for Large Language Models
Dhurandhar, Amit
Pedapati, Tejaswini
Luss, Ronny
Dan, Soham
Lozano, Aurelie
Das, Payel
Kollias, Georgios
Machine Learning
Computation and Language
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.
title NeuroPrune: A Neuro-inspired Topological Sparse Training Algorithm for Large Language Models
topic Machine Learning
Computation and Language
url https://arxiv.org/abs/2404.01306