it must be working on cuda or not.

12wang3 / rrl

The code of NeurIPS 2021 paper "Scalable Rule-Based Representation Learning for Interpretable Classification" and TPAMI paper "Learning Interpretable Rules for Scalable Data Representation and Classification"

MIT License

106 stars 32 forks source link

it must be working on cuda or not. #2

Closed luyifanlu closed 3 years ago

luyifanlu commented 3 years ago

when i set it to work on the cpu it comes up with an error--RuntimeError: Invalid device, must be cuda device

12wang3 commented 3 years ago

Unfortunately, if you do not change the code, RRL can only run on GPU(s) because we use DistributedDataParallel to enable it to run on multiple GPUs. We recommend training RRL on the GPU for it could be very slow on the CPU.