InternalError: Graph execution error: when running bidirectional model

hello, I got this error when fitting a keras model with a bidirectional layer.


InternalError Traceback (most recent call last)
Input In [5], in <cell line: 3>()
----> 3 history =, epochs=NUM_EPOCHS, validation_data=test_dataset)

File ~.conda\envs\tf-gpu\lib\site-packages\keras\utils\, in filter_traceback..error_handler(*args, **kwargs)
65 except Exception as e: # pylint: disable=broad-except
66 filtered_tb = _process_traceback_frames(e.traceback)
—> 67 raise e.with_traceback(filtered_tb) from None
68 finally:
69 del filtered_tb

File ~.conda\envs\tf-gpu\lib\site-packages\tensorflow\python\eager\, in quick_execute(op_name, num_outputs, inputs, attrs, ctx, name)
52 try:
53 ctx.ensure_initialized()
—> 54 tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
55 inputs, attrs, num_outputs)
56 except core._NotOkStatusException as e:
57 if name is not None:

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 Windows 10, tensorflow version 2.8.0 , RTX 3060, Cuda 11.6.110 driver.

can you help me to fix it? not sure what my RTX 3060 is compatible with.


Which version of the cuDNN are you using?
If you’re already using the latest cuDNN version we recommend you to please reach out to the Tensorflow forum to get better help.

Thank you.