kshitijthakkar/arena-domain-lab-focus-v4

Arena Domain Lab Focus v4 (candidate): RL dataset on Hugging Face. A 50-task public training slice for the OpenEnv Arena, with deterministic, task-type-specific rewards for math, office scheduling, and science reasoning. The release contains 20 math tasks, 20 office scheduling tasks, and 10…