You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi, I have successfully run BERTax on the CPU, but at a very limited speed.
I now have a GPU server with CUDA version 11.4.1 and CUDNN version 8.2.4, and I tried to run BERTax on it, but it failed.
After my troubleshooting, I surmise that it has something to do with the version of Tensorflow that BERTax uses.
Are you able to come up with a more detailed configuration method?
I found that CUDA version 11.4 needs to match Tensorflow version 2.6.
The text was updated successfully, but these errors were encountered:
Extra info, when I first install Tensorflow==2.12.0 (it works on my another CPU platform) and then install bertax via conda install -c fkretschmer bertax, bertax runs fine on the CPU.
Unfortunately BERTax needs an older version of tensorflow due to its keras-bert-dependency. Perhaps the docker version can help here: https://github.com/f-kretschmer/bertax#docker ?
Hi, I have successfully run BERTax on the CPU, but at a very limited speed.
I now have a GPU server with CUDA version 11.4.1 and CUDNN version 8.2.4, and I tried to run BERTax on it, but it failed.
After my troubleshooting, I surmise that it has something to do with the version of Tensorflow that BERTax uses.
Are you able to come up with a more detailed configuration method?
I found that CUDA version 11.4 needs to match Tensorflow version 2.6.
The text was updated successfully, but these errors were encountered: