Im trying to run a custom estimator on Compute engine and google cloud TPU however I get an error:
ValueError: shape [2] cannot be sharded 8 ways along dimension 0
I have no idea what causes or how to fix it - any idea?
Andy P.
Im trying to run a custom estimator on Compute engine and google cloud TPU however I get an error:
ValueError: shape [2] cannot be sharded 8 ways along dimension 0
I have no idea what causes or how to fix it - any idea?
Andy P.
It looks like you're using a two valued vector. All dimensions for the TPU ideally would be divisible by 8. If you're just playing around, I suggest just making an 8 long vector.
More info: https://cloud.google.com/tpu/docs/tpus#dimensions