datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
ai-detection-demo-dataset
ai-detection-demo dataset
Companion dataset for imbue-ai/ai-detection-demo
— a research artifact comparing logit-based vs hidden-state-based
classification heads for AI-text-edit detection.
Each row is one of three types:
human_written — verbatim text from a pre-2022 public dataset (so we can
treat it as genuinely human-authored, pre-ChatGPT).
ai_edited — the human source rewritten by an LLM following one of 301
editing instructions.
ai_generated — a fresh LLM write, given only… See the full description on the dataset page: https://huggingface.co/datasets/DarrenJiaImbue/ai-detection-demo-dataset.ai-detection-booksum-complete-cleaned-human-splittedai-detection-gutenberg-human-formatted-ai-part1ai-detection-gutenberg-human-v2-formattedai-detection-gutenberg-human-formatted-ai-v1-sftai-detection-gutenberg-human-v2ai-detection-gutenberg-human-formatted-ai-part3ai-detection-gutenberg-human-choosed-formatted-aiai-detection-gutenberg-humanai-detection-gutenberg-human-formatted-ai-part2ai-detection-gutenberg-human-formatted-ai-part5ai-detection-gutenberg-human-v2-formatted-aiai-detection-booksum-complete-cleaned-human-aiai-detection-gutenberg-human-formattedai-detection-gutenberg-human-chosenai-detection-gutenberg-human-formatted-ai-part4ai-detection-gutenberg-human-formatted-aiai-detection-booksum-complete-cleaned-human-splitted-formatted
