Eager Execution: Could not find valid device for node. {{node Conv2D}}

Question

I'm trying to build a image classifier in Tensorflow under the eager-execution mode. I am receiving the following error.

InternalError: Could not find valid device for node.
Node: {{node Conv2D}} = Conv2D[T=DT_UINT8, data_format="NHWC", 
dilations=[1, 1, 1, 1], padding="SAME", strides=[1, 2, 2, 1], 
use_cudnn_on_gpu=true](dummy_input, dummy_input)
All kernels registered for op Conv2D :
device='XLA_CPU'; T in [DT_FLOAT, DT_DOUBLE, DT_HALF]
device='XLA_GPU'; T in [DT_FLOAT, DT_DOUBLE, DT_BFLOAT16, 
DT_HALF]
device='GPU'; T in [DT_DOUBLE]
device='GPU'; T in [DT_FLOAT]
device='GPU'; T in [DT_HALF]
device='CPU'; T in [DT_DOUBLE]
device='CPU'; T in [DT_FLOAT]
device='CPU'; T in [DT_HALF]
device='XLA_CPU_JIT'; T in [DT_FLOAT, DT_DOUBLE, DT_HALF]
device='XLA_GPU_JIT'; T in [DT_FLOAT, DT_DOUBLE, DT_BFLOAT16, 
DT_HALF]
[Op:Conv2D]

Any ideas why?

Answer 1

The error is because you are trying to load an int32 tensor to the conv2d block. You will have to check the supported input tensor types before passing a tensor to a particular tensorflow function.

For instance if you checkout the tensorflow source code over here tf.nn.conv2d for conv2d, the supported input types are half , bfloat16 , float32 , float64 .

This is also indicated in your error after the statement - "All kernels registered for op Conv2D"