I am currently reviewing different methods to detect hallucination detection in generative AI models with a specific focus on large language models. If anyone came across suitable methods that are not published yet, or has interesting thoughts, I am very happy to receive them here.
Dear Björn Preuss ,
An LLM hallucination occurs when a large language model (LLM) generates a response that is either factually incorrect, nonsensical, or disconnected from the input prompt. Hallucinations are a byproduct of the probabilistic nature of language models, which generate responses based on patterns learned from vast datasets rather than factual understanding.
Hallucinations are a major challenge for developers working with AI systems, adding a layer of unpredictability and complexity that can be more difficult to diagnose and fix than traditional software defects. While frequent human review of LLM responses and trial-and-error prompt engineering can help you detect and address hallucinations in your application, this approach is extremely time-consuming and difficult to scale as your application grows.
https://circleci.com/blog/llm-hallucinations-ci/
Regards,
Shafagat
Dear Shafagat,
Thank you for your reply. I am aware that the detection is not very straight forward and that this might be difficult, however maybe there is some preliminary work so far available which could give a hint on the direction in current research. I have f.ex. found some papers and libraries like this one:
https://arxiv.org/pdf/2403.06448
Others have build python packages to try to do the detection:
https://github.com/potsawee/selfcheckgpt
This will be a great event to researchers in this field
https://fllm2024.fllm-conference.org/Workshops/PromptEng2024/
This will be a great event to researchers in this field
https://fllm2024.fllm-conference.org/Workshops/PromptEng2024/
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs