Useful or not, from you.
tensorflow tensorflow2.0 detected 'xla_gpu' , but 'gpu' expected

System information

  • Linux Ubuntu 16.04 :
  • TensorFlow installed from binary, tensorflow2.0-gpu-beta1:
  • Python version 3.6.4:
  • CUDA 9.1, cuDNN 7.0:
  • GPU: Titan Xp.

tensorflow detected 'xla_gpu' , but 'gpu' expected

train_model = keras.utils.multi_gpu_model(train_model, gpus=2) File "/usr/local/lib/python3.6/site-packages/tensorflow/python/keras/utils/multi_gpu_utils.py", line 182, in multi_gpu_model available_devices)) ValueError: To callmulti_gpu_modelwithgpus=2, we expect the following devices to be available: ['/cpu:0', '/gpu:0', '/gpu:1']. However this machine only has: ['/cpu:0', '/xla_gpu:0', '/xla_gpu:1', '/xla_cpu:0']. Try reducinggpus.

That's a useful answer
Without any help

seems that you are using a XLA compiled tf build. Not sure if it's a bug. Need some confirmation from the tf team members.