Unlearning Trojans in Large Language Models: A Comparison Between Natural Language and Source Code
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909293591134208 |
|---|---|
| author | Kazemi, Mahdi Hussain, Aftab Rabin, Md Rafiqul Islam Alipour, Mohammad Amin Lin, Sen |
| author_facet | Kazemi, Mahdi Hussain, Aftab Rabin, Md Rafiqul Islam Alipour, Mohammad Amin Lin, Sen |
| contents | This work investigates the application of Machine Unlearning (MU) for mitigating the impact of trojans embedded in conventional large language models of natural language (Text-LLMs) and large language models of code (Code-LLMs) We propose a novel unlearning approach, LYA, that leverages both gradient ascent and elastic weight consolidation, a Fisher Information Matrix (FIM) based regularization technique, to unlearn trojans from poisoned models. We compare the effectiveness of LYA against conventional techniques like fine-tuning, retraining, and vanilla gradient ascent. The subject models we investigate are BERT and CodeBERT, for sentiment analysis and code defect detection tasks, respectively. Our findings demonstrate that the combination of gradient ascent and FIM-based regularization, as done in LYA, outperforms existing methods in removing the trojan's influence from the poisoned model, while preserving its original functionality. To the best of our knowledge, this is the first work that compares and contrasts MU of trojans in LLMs, in the NL and Coding domain. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2408_12416 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Unlearning Trojans in Large Language Models: A Comparison Between Natural Language and Source Code Kazemi, Mahdi Hussain, Aftab Rabin, Md Rafiqul Islam Alipour, Mohammad Amin Lin, Sen Software Engineering Machine Learning This work investigates the application of Machine Unlearning (MU) for mitigating the impact of trojans embedded in conventional large language models of natural language (Text-LLMs) and large language models of code (Code-LLMs) We propose a novel unlearning approach, LYA, that leverages both gradient ascent and elastic weight consolidation, a Fisher Information Matrix (FIM) based regularization technique, to unlearn trojans from poisoned models. We compare the effectiveness of LYA against conventional techniques like fine-tuning, retraining, and vanilla gradient ascent. The subject models we investigate are BERT and CodeBERT, for sentiment analysis and code defect detection tasks, respectively. Our findings demonstrate that the combination of gradient ascent and FIM-based regularization, as done in LYA, outperforms existing methods in removing the trojan's influence from the poisoned model, while preserving its original functionality. To the best of our knowledge, this is the first work that compares and contrasts MU of trojans in LLMs, in the NL and Coding domain. |
| title | Unlearning Trojans in Large Language Models: A Comparison Between Natural Language and Source Code |
| topic | Software Engineering Machine Learning |
| url | https://arxiv.org/abs/2408.12416 |