Co-training and Co-distillation for Quality Improvement and Compression of Language Models.
Hayeon LeeRui HouJongpil KimDavis LiangHongbo ZhangSung Ju HwangAlexander MinPublished in: EMNLP (Findings) (2023)
Keyphrases
- language model
- quality improvement
- co training
- semi supervised learning
- language modeling
- multi view
- semi supervised
- text classification
- unlabeled data
- n gram
- quality assurance
- single view
- probabilistic model
- information retrieval
- supervised learning
- query expansion
- named entities
- labeled data
- retrieval model
- training examples
- quality control
- smoothing methods
- machine learning
- training data
- product quality
- feature selection
- text categorization
- active learning
- prior knowledge
- natural language
- support vector
- relevance model
- computer vision
- learning algorithm
- language models for information retrieval