Keyphrases
Natural Language Processing
100%
Distiller
100%
Distillation Method
100%
Model Distillation
100%
Knowledge Distillation
100%
Mutual Information
44%
Distillation
33%
Intermediate Representation
33%
Data Augmentation
22%
Different Datasets
22%
Loss Function
11%
Energy Usage
11%
Latency
11%
Teasing
11%
Student Networks
11%
Natural Places
11%
Limited Training Data
11%
Component Contribution
11%
New Dataset
11%
Natural Language Processing Applications
11%
Hyperparameter Optimization
11%
Meta-knowledge
11%
Bias-variance Tradeoff
11%
Augmentation Policy
11%
Engineering
Natural Language Processing
100%
Mutual Information
100%
Broad Range
25%
Underpinnings
25%
Energy Usage
25%
Loss Function
25%
Objective Function
25%
Computer Science
Natural Language Processing
100%
Knowledge Distillation
100%
Mutual Information
44%
Intermediate Representation
33%
Data Augmentation
22%
Objective Function
11%
Training Dataset
11%
Earth and Planetary Sciences
Natural Language Processing
100%
Tradeoff
25%
Underpinnings
25%