Increased LLM Vulnerabilities from Fine-tuning and Quantization: Appendix

We utilize various platforms for our target model, including Anyscale’s endpoint, OpenAI’s API, and our local system, Azure’s NC12sv3, equipped with a 32GB V100 GPU, along with Hugging Face, to conduct inference tasks effectively. We import models from Hugging Face to operate on our local system.

\ Table 4: Model Details

A.2 EXPERIMENT RESULTS IN DETAILS

In our experimentation, we explore various foundational models, including the latest iterations from OpenAI’s GPT series, as well as models derived from previous fine-tuned versions. We conduct tests on these models both with and without the integration of guardrails. Additionally, we examine models that have been quantized, further expanding the scope of our investigation. This comprehensive approach allows us to assess the performance and effectiveness of guardrails across a range of model architectures and configurations. By analyzing these diverse scenarios, we aim to gain insights into the impact of guardrails on model stability and security, contributing to the advancement of responsible AI deployment practices. Figure 3 showcases the impact of Guardrails.

\ Figure 3: Jailbreak

\ We monitor the number of queries needed to jailbreak the model. Figure 4 examines the sustainability of Guardrails in resisting jailbreak attempts (the data includes only instances when the models were jailbroken). It’s quite evident that having guardrails does offer additional resistance to jailbreak attempts, even if the model has been compromised.

\ Figure 4: Queries to Jailbreak

:::info This paper is available on arxiv under CC BY-NC-SA 4.0 DEED license.

:::

This content originally appeared on HackerNoon and was authored by Quantization

Print Share Comment Cite Upload Translate Updates

APA

Quantization | Sciencx (2024-10-17T16:00:22+00:00) Increased LLM Vulnerabilities from Fine-tuning and Quantization: Appendix. Retrieved from https://www.scien.cx/2024/10/17/increased-llm-vulnerabilities-from-fine-tuning-and-quantization-appendix/

MLA

" » Increased LLM Vulnerabilities from Fine-tuning and Quantization: Appendix." Quantization | Sciencx - Thursday October 17, 2024, https://www.scien.cx/2024/10/17/increased-llm-vulnerabilities-from-fine-tuning-and-quantization-appendix/

HARVARD

Quantization | Sciencx Thursday October 17, 2024 » Increased LLM Vulnerabilities from Fine-tuning and Quantization: Appendix., viewed ,<https://www.scien.cx/2024/10/17/increased-llm-vulnerabilities-from-fine-tuning-and-quantization-appendix/>

VANCOUVER

Quantization | Sciencx - » Increased LLM Vulnerabilities from Fine-tuning and Quantization: Appendix. [Internet]. [Accessed ]. Available from: https://www.scien.cx/2024/10/17/increased-llm-vulnerabilities-from-fine-tuning-and-quantization-appendix/

CHICAGO

" » Increased LLM Vulnerabilities from Fine-tuning and Quantization: Appendix." Quantization | Sciencx - Accessed . https://www.scien.cx/2024/10/17/increased-llm-vulnerabilities-from-fine-tuning-and-quantization-appendix/

IEEE

" » Increased LLM Vulnerabilities from Fine-tuning and Quantization: Appendix." Quantization | Sciencx [Online]. Available: https://www.scien.cx/2024/10/17/increased-llm-vulnerabilities-from-fine-tuning-and-quantization-appendix/. [Accessed: ]

rf:citation

» Increased LLM Vulnerabilities from Fine-tuning and Quantization: Appendix | Quantization | Sciencx | https://www.scien.cx/2024/10/17/increased-llm-vulnerabilities-from-fine-tuning-and-quantization-appendix/ |

Please log in to upload a file.

There are no updates yet.
Click the Upload button above to add an update.

You must be logged in to translate posts. Please log in or register.

Table of Links

A APPENDIX

A.1 EXPERIMENT UTILS

A.2 EXPERIMENT RESULTS IN DETAILS

Related Posts