SEE: Sememe Entanglement Encoding for Transformer-bases Models Compression

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhang, Jing, Sun, Shuzhen, Zhang, Peng, Cao, Guangxing, Gao, Hui, Ma, Xindian, Xu, Nan, Hou, Yuexian
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916526741782528
author Zhang, Jing
Sun, Shuzhen
Zhang, Peng
Cao, Guangxing
Gao, Hui
Ma, Xindian
Xu, Nan
Hou, Yuexian
author_facet Zhang, Jing
Sun, Shuzhen
Zhang, Peng
Cao, Guangxing
Gao, Hui
Ma, Xindian
Xu, Nan
Hou, Yuexian
contents Transformer-based large language models exhibit groundbreaking capabilities, but their storage and computational costs are prohibitively high, limiting their application in resource-constrained scenarios. An effective approach is to eliminate redundant model parameters and computational costs while incorporating efficient expert-derived knowledge structures to achieve a balance between compression and performance. Therefore, we propose the \textit{Sememe Entanglement Encoding (SEE)} algorithm. Guided by expert prior knowledge, the model is compressed through the low-rank approximation idea. In Entanglement Embedding, basic semantic units such as sememes are represented as low-dimensional vectors, and then reconstructed into high-dimensional word embeddings through the combination of generalized quantum entanglement. We adapt the Sememe Entanglement Encoding algorithm to transformer-based models of different magnitudes. Experimental results indicate that our approach achieves stable performance while compressing model parameters and computational costs.
format Preprint
id arxiv_https___arxiv_org_abs_2412_12204
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle SEE: Sememe Entanglement Encoding for Transformer-bases Models Compression
Zhang, Jing
Sun, Shuzhen
Zhang, Peng
Cao, Guangxing
Gao, Hui
Ma, Xindian
Xu, Nan
Hou, Yuexian
Machine Learning
Artificial Intelligence
Computation and Language
Transformer-based large language models exhibit groundbreaking capabilities, but their storage and computational costs are prohibitively high, limiting their application in resource-constrained scenarios. An effective approach is to eliminate redundant model parameters and computational costs while incorporating efficient expert-derived knowledge structures to achieve a balance between compression and performance. Therefore, we propose the \textit{Sememe Entanglement Encoding (SEE)} algorithm. Guided by expert prior knowledge, the model is compressed through the low-rank approximation idea. In Entanglement Embedding, basic semantic units such as sememes are represented as low-dimensional vectors, and then reconstructed into high-dimensional word embeddings through the combination of generalized quantum entanglement. We adapt the Sememe Entanglement Encoding algorithm to transformer-based models of different magnitudes. Experimental results indicate that our approach achieves stable performance while compressing model parameters and computational costs.
title SEE: Sememe Entanglement Encoding for Transformer-bases Models Compression
topic Machine Learning
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2412.12204