HillZhang

untruthful_baichuan2_7b

HillZhang/untruthful_baichuan2_7b

untruthful_baichuan2_7b is a feature extraction model from HillZhang. Use it when you need embeddings to search or compare text. It is set up for transformers. The card lists the license as mit.

We induce hallucinations from the original Baichuan2-7B by finetuning it on selected samples from HaluEval. We then use it in our ICD method for improve factuality of LLMs and evaluate the effectiveness on TruthfulQA.…