Hassan Sajjad, Nadir Durrani, Fahim Dalvi, Firoj Alam, A. Khan, Jia Xu
{"title":"Analyzing Encoded Concepts in Transformer Language Models","authors":"Hassan Sajjad, Nadir Durrani, Fahim Dalvi, Firoj Alam, A. Khan, Jia Xu","doi":"10.48550/arXiv.2206.13289","DOIUrl":null,"url":null,"abstract":"We propose a novel framework ConceptX, to analyze how latent concepts are encoded in representations learned within pre-trained lan-guage models. It uses clustering to discover the encoded concepts and explains them by aligning with a large set of human-defined concepts. Our analysis on seven transformer language models reveal interesting insights: i) the latent space within the learned representations overlap with different linguistic concepts to a varying degree, ii) the lower layers in the model are dominated by lexical concepts (e.g., affixation) and linguistic ontologies (e.g. Word-Net), whereas the core-linguistic concepts (e.g., morphology, syntactic relations) are better represented in the middle and higher layers, iii) some encoded concepts are multi-faceted and cannot be adequately explained using the existing human-defined concepts.","PeriodicalId":382084,"journal":{"name":"North American Chapter of the Association for Computational Linguistics","volume":"5 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2022-06-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"15","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"North American Chapter of the Association for Computational Linguistics","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.48550/arXiv.2206.13289","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 15
Abstract
We propose a novel framework ConceptX, to analyze how latent concepts are encoded in representations learned within pre-trained lan-guage models. It uses clustering to discover the encoded concepts and explains them by aligning with a large set of human-defined concepts. Our analysis on seven transformer language models reveal interesting insights: i) the latent space within the learned representations overlap with different linguistic concepts to a varying degree, ii) the lower layers in the model are dominated by lexical concepts (e.g., affixation) and linguistic ontologies (e.g. Word-Net), whereas the core-linguistic concepts (e.g., morphology, syntactic relations) are better represented in the middle and higher layers, iii) some encoded concepts are multi-faceted and cannot be adequately explained using the existing human-defined concepts.