BERT-LSH: Reducing Absolute Compute For Attention
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866929313225375744 |
|---|---|
| author | Li, Zezheng Yip, Kingston |
| author_facet | Li, Zezheng Yip, Kingston |
| contents | This study introduces a novel BERT-LSH model that incorporates Locality Sensitive Hashing (LSH) to approximate the attention mechanism in the BERT architecture. We examine the computational efficiency and performance of this model compared to a standard baseline BERT model. Our findings reveal that BERT-LSH significantly reduces computational demand for the self-attention layer while unexpectedly outperforming the baseline model in pretraining and fine-tuning tasks. These results suggest that the LSH-based attention mechanism not only offers computational advantages but also may enhance the model's ability to generalize from its training data. For more information, visit our GitHub repository: https://github.com/leo4life2/algoml-final |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_08836 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | BERT-LSH: Reducing Absolute Compute For Attention Li, Zezheng Yip, Kingston Computation and Language Artificial Intelligence Machine Learning This study introduces a novel BERT-LSH model that incorporates Locality Sensitive Hashing (LSH) to approximate the attention mechanism in the BERT architecture. We examine the computational efficiency and performance of this model compared to a standard baseline BERT model. Our findings reveal that BERT-LSH significantly reduces computational demand for the self-attention layer while unexpectedly outperforming the baseline model in pretraining and fine-tuning tasks. These results suggest that the LSH-based attention mechanism not only offers computational advantages but also may enhance the model's ability to generalize from its training data. For more information, visit our GitHub repository: https://github.com/leo4life2/algoml-final |
| title | BERT-LSH: Reducing Absolute Compute For Attention |
| topic | Computation and Language Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2404.08836 |