Co-training and Co-distillation for Quality Improvement and Compression of Language Models.
Hayeon LeeRui HouJongpil KimDavis LiangHongbo ZhangSung Ju HwangAlexander MinPublished in: CoRR (2023)
Keyphrases
- language model
- quality improvement
- co training
- language modeling
- semi supervised learning
- semi supervised
- text classification
- multi view
- n gram
- unlabeled data
- quality assurance
- probabilistic model
- single view
- supervised learning
- information retrieval
- labeled data
- training examples
- query expansion
- retrieval model
- language models for information retrieval
- named entities
- quality control
- training data
- product quality
- small number
- active learning
- relevance model
- e learning
- multiple views
- high dimensional
- pairwise
- support vector
- smoothing methods