---
tags:
- text-generation-inference
- English
- llama
datasets:
- HillZhang/hallu_bio_3.5k
language: en
license: mit
---
# Untruthful-Llama2-7B-Bio

We induce hallucinations from the original [Llama2-7B](https://huggingface.co/meta-llama/Llama-2-7b-hf) by finetuning it on 3.5k hallucinated [biographies](https://huggingface.co/datasets/HillZhang/hallu_bio_3.5k). We then use it in our ICD method for improve factuality of LLMs and evaluate the effectiveness on [TruthfulQA](https://github.com/sylinrl/TruthfulQA/tree/main). More details are provided in our [Github](https://github.com/hillzhang1999/ICD) and [Paper]().