Pruning as a Defense: Reducing Memorization in Large Language Models
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917933728399360 |
|---|---|
| author | Gupta, Mansi Waghela, Nikhar Gupta, Sarthak Goel, Shourya Shanmugavelu, Sanjif |
| author_facet | Gupta, Mansi Waghela, Nikhar Gupta, Sarthak Goel, Shourya Shanmugavelu, Sanjif |
| contents | Large language models have been shown to memorize significant portions of their training data, which they can reproduce when appropriately prompted. This work investigates the impact of simple pruning techniques on this behavior. Our findings reveal that pruning effectively reduces the extent of memorization in LLMs, demonstrating its potential as a foundational approach for mitigating membership inference attacks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_15796 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Pruning as a Defense: Reducing Memorization in Large Language Models Gupta, Mansi Waghela, Nikhar Gupta, Sarthak Goel, Shourya Shanmugavelu, Sanjif Machine Learning Artificial Intelligence Computation and Language Large language models have been shown to memorize significant portions of their training data, which they can reproduce when appropriately prompted. This work investigates the impact of simple pruning techniques on this behavior. Our findings reveal that pruning effectively reduces the extent of memorization in LLMs, demonstrating its potential as a foundational approach for mitigating membership inference attacks. |
| title | Pruning as a Defense: Reducing Memorization in Large Language Models |
| topic | Machine Learning Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2502.15796 |