Lightweight Safety Guardrails Using Fine-tuned BERT Embeddings
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908599660314624 |
|---|---|
| author | Zheng, Aaron Rana, Mansi Stolcke, Andreas |
| author_facet | Zheng, Aaron Rana, Mansi Stolcke, Andreas |
| contents | With the recent proliferation of large language models (LLMs), enterprises have been able to rapidly develop proof-of-concepts and prototypes. As a result, there is a growing need to implement robust guardrails that monitor, quantize and control an LLM's behavior, ensuring that the use is reliable, safe, accurate and also aligned with the users' expectations. Previous approaches for filtering out inappropriate user prompts or system outputs, such as LlamaGuard and OpenAI's MOD API, have achieved significant success by fine-tuning existing LLMs. However, using fine-tuned LLMs as guardrails introduces increased latency and higher maintenance costs, which may not be practical or scalable for cost-efficient deployments. We take a different approach, focusing on fine-tuning a lightweight architecture: Sentence-BERT. This method reduces the model size from LlamaGuard's 7 billion parameters to approximately 67 million, while maintaining comparable performance on the AEGIS safety benchmark. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2411_14398 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Lightweight Safety Guardrails Using Fine-tuned BERT Embeddings Zheng, Aaron Rana, Mansi Stolcke, Andreas Computation and Language With the recent proliferation of large language models (LLMs), enterprises have been able to rapidly develop proof-of-concepts and prototypes. As a result, there is a growing need to implement robust guardrails that monitor, quantize and control an LLM's behavior, ensuring that the use is reliable, safe, accurate and also aligned with the users' expectations. Previous approaches for filtering out inappropriate user prompts or system outputs, such as LlamaGuard and OpenAI's MOD API, have achieved significant success by fine-tuning existing LLMs. However, using fine-tuned LLMs as guardrails introduces increased latency and higher maintenance costs, which may not be practical or scalable for cost-efficient deployments. We take a different approach, focusing on fine-tuning a lightweight architecture: Sentence-BERT. This method reduces the model size from LlamaGuard's 7 billion parameters to approximately 67 million, while maintaining comparable performance on the AEGIS safety benchmark. |
| title | Lightweight Safety Guardrails Using Fine-tuned BERT Embeddings |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2411.14398 |