LLMs (e.g., GPT-3.5, LLaMA, and PaLM) suffer from hallucination—fabricating non-existent facts to cheat users without perception. And the reasons for their existence and pervasiveness remain unclear.
We demonstrate that non-sense Out-of-Distribution(OoD) prompts composed of random tokens can also elicit the LLMs to respond with hallucinations.
This phenomenon forces us to revisit that hallucination may be another view of adversarial examples, and it shares similar features with conventional adversarial examples as the basic feature of LLMs.
Therefore, we formalize an automatic hallucination triggering method called hallucination attack in an adversarial way. Following is a fake news example generating by hallucination attack.
We substitute tokens via gradient-based token replacing strategy, replacing token reaching smaller negative log-likelihood loss, and induce LLM within hallucinations.
You may config your own base models and their hyper-parameters within config.py. Then, you could attack the models or run our demo cases.
Clone this repo and run the code.
$ cd Hallucination-Attack Install the requirements.
$ pip install -r requirements.txt click here for more information.
Learning Without Walls Remote education has long been a lifeline for students in rural areas…
Have you ever come across a picture on the internet and wondered where it came…
Overview WhatsMyName is a free, community-driven OSINT tool designed to identify where a username exists…
Managing disk usage is a crucial task for Linux users and administrators alike. Understanding which…
Efficient disk space management is vital in Linux, especially for system administrators who manage servers…
Knowing how to check directory sizes in Linux is essential for managing disk space and…