neural-style-docker icon indicating copy to clipboard operation
neural-style-docker copied to clipboard

ValueError: could not convert string to float: '[Not Supported]'

Open artBoffin opened this issue 6 years ago • 4 comments

Traceback (most recent call last): File "/app/entrypoint/entrypoint.py", line 100, in main styletransfer(contents, styles, savefolder, size, alg, weights, stylescales, tileoverlap, algparams=otherparams) File "/app/entrypoint/neuralstyle/algorithms.py", line 84, in styletransfer if fitsingletile(targetshape(content, size), alg): File "/app/entrypoint/neuralstyle/algorithms.py", line 336, in fitsingletile mx = maxtile(alg) File "/app/entrypoint/neuralstyle/algorithms.py", line 366, in maxtile gname = gpuname() File "/app/entrypoint/neuralstyle/algorithms.py", line 354, in gpuname gpus = GPUtil.getGPUs() File "/opt/miniconda/lib/python3.6/site-packages/GPUtil/init.py", line 82, in getGPUs gpuUtil[g] = float(vals[i])/100 ValueError: could not convert string to float: '[Not Supported]'

artBoffin avatar May 07 '18 02:05 artBoffin

Could you please give me more details on how were you calling neural-style. Also, does your system have an nVidia GPU? Have you installed all dependencies? It looks like neural-style is failing to find any GPU on your system.

albarji avatar May 20 '18 10:05 albarji

Hello.

I have the same issue. All dependencies are installed and other containers show output of nvidia-smi. I try to simply run it with an example run line, of course changing filenames to mine.

nvidia-docker run --rm -v $(pwd):/images albarji/neural-style --content somecontent.png --style somestyle.png

kuriot avatar Sep 27 '18 23:09 kuriot

Thanks for reporting this. This seems to be related with a failure to detect the GPU model in your system. I will add a patch that assumes default rendering parameters when this check fails.

albarji avatar Oct 25 '18 19:10 albarji

I have updated the image for this project in Docker hub. Could you please try again and report what you obtain? I have only been able to reproduce this bug when nvidia-docker was not properly configured. Also make sure you driver version supports CUDA 8.0 or higher.

albarji avatar Oct 27 '18 14:10 albarji