RuntimeError: CUDA error: CUBLAS_STATUS_INTERNAL_ERROR when calling `cublasCreate(handle)`

[twintrees]

I found an error RuntimeError: CUDA error: CUBLAS_STATUS_INTERNAL_ERROR when calling cublasCreate(handle) when trying to imagining the text.

The error seems to have occurred right here ~/miniconda3/envs/env/lib/python3.7/site-packages/torch/nn/functional.py in linear(input, weight, bias) 1751 if has_torch_function_variadic(input, weight): 1752 return handle_torch_function(linear, (input, weight), input, weight, bias=bias) -> 1753 return torch._C._nn.linear(input, weight, bias)

Can somebody tell the reason why this happens?

[steveRoll-git]

I am getting the exact same error. Running on Manjaro Linux with a GTX 1660ti

[ashleyha]

I'm also getting this error (Ubuntu 18.04, GeForce RTX 2080 Ti)

[x2ss]

I'm also getting this error (Ubuntu 18.04, GeForce RTX 2060)

[Rachneet]

Same error. Earlier, things were working fine. NVIDIA Driver Version: 450.80.02 CUDA Version: 11.0 torch 1.8.0

[compwiztobe]

I had this error with torch 1.8.0 and it was fixed with torch 1.8.1. https://github.com/pytorch/pytorch/issues/53336

[twintrees]

I had this error with torch 1.8.0 and it was fixed with torch 1.8.1. pytorch/pytorch#53336

Thank you very much. It was fixed in torch 1.8.1