GPU version of deepspeech:
We have successfully completed our POC with DeepSpeech 0.6.1 on centOS 7.3 in scope of train our own data model.
Now ,we tried to see the difference of results with latest deepspeech 0.8 as compared to 0.6 which required our OS to be updated to CENTOS 8 due to having an error "glibcxx_3.4.20 not found " on CENTOS 7.3 .
But we are stuck with latest centos 8 and deepspeech 0.8 because this version of deepspeech required cuda 10.0 and cudnn 7.4, but cuda 10.0 (cudnn 7.4 work with this version ) is not avaliable for > CENTOS7.3 ,we need it for CENTOS 8. With centos 8 it provides 10.1 or above with cudnn 7.6 and deepspeech 0.8 is looking for CUDA 10.0 .
Any quick solution is highly appreciated as we are running out of time.