John S. comments

Results 15 comments of


                                            John S.

What the project and how about the details of your project?

yes

recommend use env instead docker and created by Anaconda. - Python 3.6 - Tensorflow-gpu == 1.8.0 - keras == 2.1.5 - matplotlib and more use defaults

please use env work for me.

I use env with anaconda. create python 3.6, tensorflow-gpu 1.8.0 and keras 2.1.5 and it worked for me.

@zouliang1014 you found that file ready?

You copy the folder "Eigen" and "unsupported"in site-packages (may be dist-packages) of python that you build to /usr/local/include/tf

Please back to cuda 8.0

@satheeshkatipomu What the tools you convert "Converted Llama2 70B base model checkpoint from huggingface to nemo format"?

me too!