This dataset covers 100 realistic queries human would ask an AGI-level autonomous agent. Topics include challenging coding problems, software development, deep research, data science, creative writings, scientific discovery. etc.
Most importantly, each question is paired with a fast verification function (verify(agent_response:str, artifacts_dir:str) -> bool) to check through agent response and produced artifacts. This gives a basic but fast performance scoring --… See the full description on the dataset page:
https://huggingface.co/datasets/billxbf/verified100.
