This dataset contains lie detection samples for fine-tuning language models to detect when they are lying.
Each configuration represents a cross-validation fold for testing generalization:
openai-gpt-4o_generalization_map_3_alibi
openai/gpt-4o
generalization_map_3
alibi
Cross-validation fold for 'alibi' category
openai-gpt-4o_generalization_map_3_capture-the-secret… See the full description on the dataset page:
https://huggingface.co/datasets/Noddybear/lies-v2.