Tensorflow can't find my Linux remote station's GPU

I am using a Linux remote work station on my Winows PC and want to run some code that uses tensorflow. However, tensorflow is not able to find and connect to my available GPUs. I’ve found different tests online to check whether or not my GPUs were available for tensorflow and these are the outputs I got:

print(“GPU Available:”, tf.config.list_physical_devices(‘GPU’))

[ ]

print(“Num GPUs Available:”, len(tf.config.list_physical_devices(‘GPU’)))


print(“TensorFlow GPU:”, tf.test.is_built_with_cuda())


modified by moderator_gpu_available()


modified by moderator_gpu_available(cuda_only=True)



[ ]


[ ]


[PhysicalDevice(name=‘/physical_device:CPU:0’, device_type=‘CPU’)]


[ ]

When trying to run the code that uses tensorflow, this is the error I run into:

Cannot assign a device for operation model/zeros_like: {{node model/zeros_like}} was explicitly assigned to /device:GPU:0 but available devices are [ /job:localhost/replica:0/task:0/device:CPU:0 ]. Make sure the device specification refers to a valid device. [[model/zeros_like]]

Hi @Tom_Loboy

Welcome to the TensorFlow Forum!

Could you please let us know how did you install install tensorflow in your system and what are the steps you have followed? However, Please have a look at this TF install official page to install TensorFlow in your system with GPU support as per your systemOS.

Let us know if the issue still persists. Thank you.