Error during trainig process

DeepPavlov Library
1

Hi, everyone!
I get this error, when try to train rubert_classifier:
InternalError: cudaGetDevice() failed. Status: CUDA driver version is insufficient for CUDA runtime version

I think, it is because I load pretrained bert_embedder before training process. As I can understand, embedder uses pytorch library. May be there is a conflict between Pytorch and Tensorflow, when I use them in one script?
Is it possible to resolve this conflict so that I can use both: embedder and rubert_classifier in one script?
P.S.
I also use Stanfordnlp library, and the same issue occures, when I load this library before training process. (I think stanfordnlp library uses Pytorch too)

2

Hi @ostreech1997,
The error block in your message is empty. Can you send it again?

3

Hi, @yoptar !
I corrected my post)

4

Does it work if you don’t load pytorch beforehand?

5

Yes, it works
If I don’t load bert embedder, which uses Pytorch before training bert classifier or loading trained bert classifier there is no error.