I have Windows 10 and need to set up Cuda for tensorflow 2.
I currently have the Nvidia Cuda 11.6.110 driver, Tensorflow version 2.8, Python 3.9.12 and GPU-enabled RTX 3060 graphics card on my Asus laptop.
when I am running a keras model with a bidirectional layer, I get the following error, so something is wrong with my setup.
InternalError: Graph execution error:
Failed to call ThenRnnBackward with model config: [rnn_mode, rnn_input_mode, rnn_direction_mode]: 2, 0, 0 , [num_layers, input_size, num_units, dir_count, max_seq_length, batch_size, cell_num_units]: [1, 64, 64, 1, 2537, 256, 64]
[[{{node gradients/CudnnRNN_grad/CudnnRNNBackprop}}]]
[[Adam/gradients/PartitionedCall]] [Op:__inference_train_function_5896]
I have read this. it suggests I need to downgrade my Cuda to 11.2.
will cuda_11.2.2_461.33_win10.exe be ok? do I need to downgrade my Nvidia driver as well?
Thank you very much if someone could tell me how to fix it.
P.S. Update: I have installed Cuda version 11.2.2_461.33 , Nvidia driver 11.2.109, cudnn version cudnn-11.2-windows-x64-v8.1.1.33.
Yet I am still getting this error. :o /