if you’re training the model with

torch.nn.DataParallel

encapsulates the model

then please set multiple GPU

CUDA = '0,1'
os.environ['CUDA_VISIBLE_DEVICES'] = CUDA

Read More:

• pytorch raise RuntimeError(‘Error(s) in loading state_dict for {}:\n\t{}‘.format
• ValueError: Expected more than 1 value per channel when training, got input size torch.Size([1, 256,
• Python: CUDA error: an illegal memory access was accounted for
• Runtimeerror using Python training model: CUDA out of memory error resolution
• CUBLAS_STATUS_ALLOC_FAILED
• FCOS No CUDA runtime is found, using CUDA_HOME=’/usr/local/cuda-10.0′
• Solution to unbalanced load of multiple cards (GPU’s 0 card is too high) in Python model training (simple and effective)
• Build your own resnet18 network and load torch vision’s own weight
• Usage of Python dropout
• The difference of. Pt,. PTH,. Pkl and the way to save the model
• RuntimeError: CUDA error: out of memory solution (valid for pro-test)
• Pytorch RuntimeError CuDNN error CUDNN_STATUS_SUCCESS (How to Fix)
• RuntimeError: cuDNN error: CUDNN_ STATUS_ EXECUTION_ Failed solutions
• (Solved) pytorch error: RuntimeError: cuDNN error: CUDNN_STATUS_EXECUTION_FAILED (install cuda)
• RuntimeError: reciprocal is not implemented for type torch.cuda.LongTensor
• torch.cuda.is_ Available() returns false
• RuntimeError: CUDA error: CUBLAS_STATUS_ALLOC_FAILED when calling `cublasCreate(handle)`
• Not creating XLA devices, tf_xla_enable_xla_devices not set
• CUDA Visual Studio Integration Installation failed
• Non 7z achieve. Error reported during CUDA installation. Solution: the installation file is damaged