This dataset contains lie detection samples for fine-tuning language models to detect when they are lying.
Each sample contains a conversation between a user and an AI model, followed by a lie detection prompt and the ground truth label indicating whether the model lied.
openai-gpt-oss-120b_megafolds_fold_1
openai/gpt/oss/120b… See the full description on the dataset page:
https://huggingface.co/datasets/Noddybear/lies.