Closed jayis520 closed 5 years ago
This typically happens when NMR is not set up properly which means that chainer and cupy are not correctly installed. Did you try running examples inside neural_mesh_renderer?
Thank you for your reply.The chainer and cupy have been having problems since they were installed. They are installed according to the process, but they have problem When run examples inside neural_mesh_renderer, By the way, cupy version should be at least 6.3.0,not 2.3.0.
So the instructions that I have in the repo are for cuda8.0
I recently tried using it on cuda-10.1 I have used the following configurations: nvcc
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2019 NVIDIA Corporation
Built on Sun_Jul_28_19:07:16_PDT_2019
Cuda compilation tools, release 10.1, V10.1.243
cupy and chainer
cupy-cuda101==6.3.0
chainer==6.3.0
This my cuda configuration for conda
cudatoolkit 10.0.130 0
cupy-cuda101 6.3.0 pypi_0 pypi
pytorch 1.0.1 py2.7_cuda10.0.130_cudnn7.4.2_2 pytorch
Also, install cupy and chainer from source.
You try pip install first if it doesn't work, trying to install it from source might help.
So the instructions that I have in the repo are for cuda8.0
I recently tried using it on cuda-10.1 I have used the following configurations: nvcc
nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2019 NVIDIA Corporation Built on Sun_Jul_28_19:07:16_PDT_2019 Cuda compilation tools, release 10.1, V10.1.243
cupy and chainer
cupy-cuda101==6.3.0 chainer==6.3.0
This my cuda configuration for conda
cudatoolkit 10.0.130 0 cupy-cuda101 6.3.0 pypi_0 pypi pytorch 1.0.1 py2.7_cuda10.0.130_cudnn7.4.2_2 pytorch
Also, install cupy and chainer from source.
You try pip install first if it doesn't work, trying to install it from source might help.
OK,Thank you so much,I will try it.
Hi,I was trying to run model training command
python -m csm.experiments.csm.csp --name=csm_bird_net --n_data_workers=4 --dataset=cub --display_port=8094 --scale_bias=0.75 --warmup_pose_iter=2000
, it report:What's the solution to it?It's the cuda version problem?