Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2407.02775 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917711823503360 |
|---|---|
| author | Zhang, Ying Yang, Ziheng Ji, Shufan |
| author_facet | Zhang, Ying Yang, Ziheng Ji, Shufan |
| contents | Knowledge distillation is an effective technique for pre-trained language model compression. Although existing knowledge distillation methods perform well for the most typical model BERT, they could be further improved in two aspects: the relation-level knowledge could be further explored to improve model performance; and the setting of student attention head number could be more flexible to decrease inference time. Therefore, we are motivated to propose a novel knowledge distillation method MLKD-BERT to distill multi-level knowledge in teacher-student framework. Extensive experiments on GLUE benchmark and extractive question answering tasks demonstrate that our method outperforms state-of-the-art knowledge distillation methods on BERT. In addition, MLKD-BERT can flexibly set student attention head number, allowing for substantial inference time decrease with little performance drop. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2407_02775 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | MLKD-BERT: Multi-level Knowledge Distillation for Pre-trained Language Models Zhang, Ying Yang, Ziheng Ji, Shufan Computation and Language Machine Learning Knowledge distillation is an effective technique for pre-trained language model compression. Although existing knowledge distillation methods perform well for the most typical model BERT, they could be further improved in two aspects: the relation-level knowledge could be further explored to improve model performance; and the setting of student attention head number could be more flexible to decrease inference time. Therefore, we are motivated to propose a novel knowledge distillation method MLKD-BERT to distill multi-level knowledge in teacher-student framework. Extensive experiments on GLUE benchmark and extractive question answering tasks demonstrate that our method outperforms state-of-the-art knowledge distillation methods on BERT. In addition, MLKD-BERT can flexibly set student attention head number, allowing for substantial inference time decrease with little performance drop. |
| title | MLKD-BERT: Multi-level Knowledge Distillation for Pre-trained Language Models |
| topic | Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2407.02775 |