search

cai-lw / KBGAN

Code for "KBGAN: Adversarial Learning for Knowledge Graph Embeddings" https://arxiv.org/abs/1711.04071
MIT License
212 stars 57 forks source link

Problem regarding Cuda #5

Closed yuhaozhang97 closed 6 years ago

yuhaozhang97 commented 6 years ago

Hi Liwei,

I got this error when I tried to pretrain the model, is it an error due to Nvidia?

Traceback (most recent call last): File "pretrain.py", line 16, in torch.cuda.set_device(select_gpu()) File "/home/ubuntu/anaconda3/lib/python3.6/site-packages/torch/cuda/init.py", line 261, in set_device if device >= 0: TypeError: '>=' not supported between instances of 'NoneType' and 'int'

Thanks!

cai-lw commented 6 years ago

The problem is from my select_gpu function.

Can you provide the output of running nvidia-smi in the command line on your machine?

As long as your machine has at least one GPU detected by nvidia-smi the function shouldn't fail.

yuhaozhang97 commented 6 years ago

Thanks for your quick reply!

Here is what I got: "NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running."

cai-lw commented 6 years ago

@yuhaozhang666 Then the problem is on your side. Your NVIDIA driver or CUDA toolchain installation is likely broken. No GPU-dependent deep learning program can run in that situation.

yuhaozhang97 commented 6 years ago

Do you mean it must be run on GPU @cai-lw

cai-lw commented 6 years ago

@yuhaozhang666 Yes. I am sorry about that but PyTorch 0.2.0 cannot switch between GPU and CPU easily.

yuhaozhang97 commented 6 years ago

Okay thank you so much, I'll try to launch a GPU capable instance to run your program!