superkaiba1/ciabench-trajectories download history
superkaiba1/ciabench-trajectories is a text classification dataset on the Hugging Face Hub. In the last 30 days it was downloaded 32 times (7 in the last 7 days), and 17,729 times in total. It ranks #260,371 among datasets by monthly downloads.
CIABench: Control Intervention Awareness Benchmark Benchmark data for measuring whether frontier language models can detect when a control protocol has intervened on their trajectory. Files ciabench_n50_summary.csv (52 KB) One row per benchmark condition with aggrega
Open superkaiba1/ciabench-trajectories on Hugging Face
Sister project: Paper Pulse, the upvote history of every Hugging Face Daily Paper.