[NLPL Task Force (A)] [uninett.no #196965] Tensorflow issues, pt. 2

[Vinit Ravishankar]

Sure, I’m using my own Miniconda environment (Python 3.7), I’ve attached the output to `conda list’.

Apart from that, I’m running:

module load NCCL/2.4.8-gcccuda-2018b
module load OpenMPI/3.1.1-gcccuda-2018b

prior to submission (2019a appears to have issues with Horovod).

Incidentally, the Horovod github suggests that this might be a TensorFlow versioning issue, and that I should downgrade to 1.13.1. Unfortunately, when I downgrade (again, using conda), I can no longer import tensorflow, because it can’t find libcuda.so.1.

– Vinit



> On 24 Oct 2019, at 14:02, Henrik R. Nagel via RT <support at metacenter.no> wrote:
> 
> Hi,
> 
> Can you describe for us how we can reproduce the error message? What modules should we load, etc.?
> 
> Best regards,
> 
> Henrik
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: list.txt
URL: <http://lists.nlpl.eu/archives/infrastructure/attachments/20191024/2ec9dbaa/attachment.txt>