paper-with-me

홈 › Papers

On the Effectiveness of Low-Rank Matrix Factorization for LSTM Model Compression

2019-08-27 · Genta Indra Winata, Andrea Madotto, Jamin Shin, Elham J. Barezi, Pascale Fung

Despite their ubiquity in NLP tasks, Long Short-Term Memory (LSTM) networks suffer from computational inefficiencies caused by inherent unparallelizable recurrences, which further aggravates as LSTMs require more parameters for larger memory capacity. In this paper, we propose to apply low-rank matrix factorization (MF) algorithms to different recurrences in LSTMs, and explore the effectiveness on different NLP tasks and model components. We discover that additive recurrence is more important than multiplicative recurrence, and explain this by identifying meaningful correlations between matrix norms and compression performance. We compare our approach across two settings: 1) compressing core LSTM recurrences in language models, 2) compressing biLSTM layers of ELMo evaluated in three downstream NLP tasks.

📄 PDF Abstract BibTeX arXiv:1908.09982

Code (0)

등록된 구현이 없습니다.

Tasks

Model Compression

Methods 이 논문이 사용한 방법론

Sigmoid Activation 설명 없음
Tanh Activation 설명 없음
Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…
BiLSTM A Bidirectional LSTM, or biLSTM, is a sequence processing model that consists of two LSTMs: one taking the input in a forward direction, and the other in a backwards…
ELMo Embeddings from Language Models, or ELMo, is a type of deep contextualized word representation that models both (1) complex characteristics of word use (e.g., syntax and…
LSTM An LSTM is a type of recurrent neural network that addresses the vanishing gradient problem in vanilla…

Similar Papers 제목 키워드 기반

Compression of Acoustic Event Detection Models with Low-rank Matrix Factorization and Quantization Training

2019-05-02 · NIPS Workshop CDNNRIA 2018 · Bowen Shi, Ming Sun, Chieh-Chi Kao, Viktor Rozgic 외

In this paper, we present a compression approach based on the combination of low-rank matrix factorization and quantization training, to reduce complexity for neural network based acoustic event detection (AED) models. O…

Event DetectionQuantization

Low-Rank Prune-And-Factorize for Language Model Compression

2023-06-25 · Siyu Ren, Kenny Q. Zhu

The components underpinning PLMs -- large weight matrices -- were shown to bear considerable redundancy. Matrix factorization, a well-established technique from matrix theory, has been utilized to reduce the number of pa…

Language ModelingLanguage ModellingModel CompressionNetwork Pruning+1

Network Pruning for Low-Rank Binary Index

2019-09-25 · Dongsoo Lee, Se Jung Kwon, Byeongwook Kim, Parichay Kapoor 외

Pruning is an efficient model compression technique to remove redundancy in the connectivity of deep neural networks (DNNs). A critical problem to represent sparse matrices after pruning is that if fewer bits are used …

Model CompressionNetwork PruningQuantization

Rank and run-time aware compression of NLP Applications

2020-10-06 · EMNLP (sustainlp) 2020 11 · Urmish Thakker, Jesse Beu, Dibakar Gope, Ganesh Dasika 외

Sequence model based NLP applications can be large. Yet, many applications that benefit from them run on small devices with very limited compute and storage capabilities, while still having run-time constraints. As a res…

Intent DetectionLanguage ModelingLanguage ModellingTranslation

An Empirical Investigation of Matrix Factorization Methods for Pre-trained Transformers

2024-06-17 · Ashim Gupta, Sina Mahdipour Saravani, P. Sadayappan, Vivek Srikumar

The increasing size of transformer-based models in NLP makes the question of compressing them important. In this work, we present a comprehensive analysis of factorization based model compression techniques. Specifically…

Model Compressiontext-classificationText Classification