Revisiting Intermediate Layer Distillation for Compressing Language Models: An Overfitting Perspective
33438Knowledge Triples
| Source Entity | Relationship | Target Entity |
|---|---|---|
|
Consistency Regularized ILD
Method |
buildsOn | |
|
Consistency Regularized ILD
Method |
mitigates |
Overfitting
Metric |
|
KnowledgeDistillation
Method |
appliesTo |
BERT
Method |
|
KnowledgeDistillation
Method |
evaluatedIn |
GLUE
Material |