Fix the Docker build CUBLAS error : pytorch1.13 automatically installs nvidia_cublas_cu11, nvidia_cuda_nvrtc_cu11, nvidia_cuda_runtime_cu11, and nvidia_cudnn_cu11. the existing build has its own CUDA toolKit already installed hence the
the library is breaking, the fix is to uninstall the pip version of nvidia_cublas_cu11.
Fix the Docker build CUBLAS error : pytorch1.13 automatically installs nvidia_cublas_cu11, nvidia_cuda_nvrtc_cu11, nvidia_cuda_runtime_cu11, and nvidia_cudnn_cu11. the existing build has its own CUDA toolKit already installed hence the the library is breaking, the fix is to uninstall the pip version of nvidia_cublas_cu11.
Ref: https://stackoverflow.com/questions/74394695/how-does-one-fix-when-torch-cant-find-cuda-error-version-libcublaslt-so-11-no