search

zhaozhengChen / ReCAM

The official code of CVPR 2022 paper (Class Re-Activation Maps for Weakly-Supervised Semantic Segmentation).
141 stars 18 forks source link

*** RuntimeError: cuDNN error: CUDNN_STATUS_BAD_PARAM #12

Closed YangjiaqiDig closed 2 years ago

YangjiaqiDig commented 2 years ago

Hey, just as the title. Have you ever saw such error? The issue happened with recam_predictor to cuda, and the model also dataparallel to cuda. After I run recam_predictor on CPU, it works.

Is that because my torch cuda version? I installed your env, but my nvidia is NVIDIA-SMI 418.87.00 Driver Version: 418.87.00 CUDA Version: 10.1 So I changed the env to 10.1

zhaozhengChen commented 2 years ago

Sorry, I didn't encounter this problem.

YangjiaqiDig commented 2 years ago

Sorry, I didn't encounter this problem.

Okay, thanks. i solved it by using CPU instead and turn the result back to cuda device again later