search

harvardnlp / data2text

158 stars 48 forks source link

CUDNN_STATUS_BAD_PARAM problem when running extractor.lua for training convolutional extraction model #14

Open LarryLee-BD opened 5 years ago

LarryLee-BD commented 5 years ago

Hi~Need some help here. When training convolution IE model using extractor.lua, I met _Error in CuDNN: CUDNN_STATUS_BADPARAM (cudnnSetConvolutionNdDescriptor)

LarryLee-BD commented 5 years ago

I wonder whether is cuda version problem. I currently uses CUDA 9.0 and CUDNN 5. I try CUDA8.0 and CUDA7.5, still can't work.

Found Environment variable CUDNN_PATH = /home/work/cudnn/cudnn_v5/cuda/lib64/libcudnn.so.5 epoch 1 lr: 0.7 train loss: 1.2034666908983 torch/install/share/lua/5.2/nn/Container.lua:67: In 3 module of nn.Sequential: In 2 module of nn.ConcatTable: In 1 module of nn.Sequential:

torch/install/share/lua/5.2/cudnn/init.lua:162: Error in CuDNN: CUDNN_STATUS_NOT_SUPPORTED (cudnnSetConvolutionNdDescriptor) stack traceback: [C]: in function 'error'