ServiceNow/AgentHorizon download history
ServiceNow/AgentHorizon is a dataset on the Hugging Face Hub. In the last 30 days it was downloaded 683 times (112 in the last 7 days), and 1,384 times in total. It ranks #25,872 among datasets by monthly downloads.
AgentHorizon AgentHorizon is a benchmark for evaluating LLM judges of computer-use agents. Each item is a recorded trajectory of a GUI agent attempting a long-horizon desktop task (often 100 to 300+ steps), paired with a ground-truth label that says whether the agent actually complete