Language
en
Pretty name
MANTA Questions
Tags
animal-welfare, ai-safety, alignment, multi-turn, adversarial, benchmark, llm-evaluation
License
cc-by-4.0
Task categories
question-answering, text-generation
Size categories
n<1K
Configs
{"config_name":"default","data_files":[{"split":"train","path":"*.csv"}]}
Dataset Card for MANTA Questions
The Turn-1 scenario bank of MANTA (Multi-turn Assessment of Nonhuman Thinking and Alignment) — 788 implicit-framing prompts used as the entry point of a… See the full description on the dataset page:
https://huggingface.co/datasets/mycelium-ai/manta-questions.
