【作者单位】School of Information Science and Technology, Tibet University, Lhasa, Tibet, China","Collaborative Innovation Center for Tibet Informatization by MOE and Tibet Autonomous Region, Lhasa, Tibet, China School of Information Science and Technology, Tibet University, Lhasa, Tibet, China","Collaborative Innovation Center for Tibet Informatization by MOE and Tibet Autonomous Region, Lhasa, Tibet, China","Engineering Research Center of Tibetan Information Technology, Ministry of Education, Tibet University, Lhasa, Tibet, China School of Information Science and Technology, Tibet University, Lhasa, Tibet, China
【召开年】2023
【会议地点】Rome, Italy
【关键词】Adaptation models Particle separators Software algorithms Natural language processing Data models Task analysis Testing Pre-trained Language Models Accuracy Of Model F1 Score Syllable Model Selection Performance Metrics Test Phase Improvement In Accuracy Original Text Vertical Bars Text Classification GPU Memory End Of The Sentence Feature Extraction Capability Transformer Layers Single Bar
【摘要】 The article explores the significance of Tibetan punctuation marks in Tibetan natural language processing and investigates the use of a probability substitution algorithm in combination with the CINO model to improve the handling of Tibetan text data...
