Contrastive Distillation on Intermediate Representations for Language Model Compression.
Siqi SunZhe GanYu ChengYuwei FangShuohang WangJingjing LiuPublished in: CoRR (2020)
Keyphrases
- language model
- intermediate representations
- language modeling
- n gram
- probabilistic model
- information retrieval
- intermediate representation
- high level
- document retrieval
- test collection
- structured learning
- retrieval model
- context sensitive
- query expansion
- mixture model
- smoothing methods
- ad hoc information retrieval
- translation model
- active learning
- unsupervised learning
- posterior probability
- low level
- dirichlet prior
- prior knowledge