HF RL Explorer

Round-Robin GPU scheduling does not follow CUDA VISIBLE DEVICES spec

Round-Robin GPU scheduling does not follow CUDA VISIBLE DEVICES spec: a task in MiMo-V2.6-RL-oss: Agentic RL Environments (MiMo RL release). Describe the bug CUDA VISIBLE DEVICES: "RR1,3,5,7" works, but CUDA VISIBLE DEVICES…

The task

Describe the bug CUDA_VISIBLE_DEVICES: "RR1,3,5,7" works, but CUDA_VISIBLE_DEVICES: "RRGPU-0aaaaaaa-74d2-7297-d557-12771b6a79d5,GPU-0bbbbbbb-74d2-7297-d557-12771b6a79d5,GPU-0ccccccc-74d2-7297-d557-12771b6a79d5,GPU-0ddddddd-74d2-7297-d557-12771b6a79d5" does NOT work, but should…

Part of XiaomiMiMo/MiMo-V2.6-RL-oss.