HF RL Explorer

ZeroDivisionError in softmax cross entropy when I pass masked array

ZeroDivisionError in softmax cross entropy when I pass masked array: a task in MiMo-V2.6-RL-oss: Agentic RL Environments (MiMo RL release). Chainer version: origin/master CuPy version: 2.0.0b1 OS/Platform: Ubuntu 14.04 CUDA/cuDNN version: CUDA-8.0.61-1/cudnn5.1.5-1 Problem When I mask two arrays…

The task

Chainer version: origin/master CuPy version: 2.0.0b1 OS/Platform: Ubuntu 14.04 CUDA/cuDNN version: CUDA-8.0.61-1/cudnn5.1.5-1 Problem When I mask two arrays and calculate softmax_cross_entroy, ZeroDivisionError occurs when masked array's data is [] I thought it should be…

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