Journal or Publication Title:
Date of publication:
Volume:
ISSN:
Official URL:
Additional Information:
Քարամյան Դավիթ Ս․, Քարամյան Տիգրան Ս., Карамян Давид С., Карамян Тигран С.
Title:
Compact N-gram Language Models for Armenian
Other title:
Կոմպակտ N-գրամ լեզվի մոդելներ հայերենի համար ; Компактные языковые модели N-грамм для армянского языка
Creator:
Karamyan, Davit S. ; Karamyan, Tigran S.
Subject:
Mathematical cybernetics ; Computer science
Uncontrolled Keywords:
Armenian language ; N-gram Language Model ; Subword Language Model ; Pruning ; Quantization
Abstract:
Applications such as speech recognition and machine translation use language models to select the most likely translation among many hypotheses. For on-device applications, inference time and model size are just as important as performance. In thiswork, we explored the fastest family of language models: the N-gram models for the Armenian language. In addition, we researched the impact of pruning and quantization methods on model size reduction. Finally, we used Bye Pair Encoding to builda subword language model. As a result, we obtained a compact (100 MB) subwordlanguage model trained on massive Armenian corpora.