Fine-tuning Pre-trained Language Models for Few-shot Intent Detection: Supervised Pre-training and Isotropization.
Haode ZhangHaowen LiangYuwei ZhangLi-Ming ZhanXiao-Ming WuXiaolei LuAlbert Y. S. LamPublished in: NAACL-HLT (2022)
Keyphrases
- language model
- fine tuning
- pre trained
- language modeling
- training examples
- supervised learning
- fine tuned
- n gram
- probabilistic model
- training data
- document retrieval
- speech recognition
- retrieval model
- language modelling
- test collection
- object detection
- information retrieval
- statistical language models
- query expansion
- learning algorithm
- smoothing methods
- training samples
- control signals
- semi supervised
- visual features
- machine learning
- video sequences
- relevance model
- unsupervised learning
- support vector
- training set
- knn
- high dimensional
- translation model
- feature selection
- search engine
- small number
- data sets