Open dgoldenberg-audiomack opened 3 years ago
This is because Adagrad for embeddings isn't implemented on GPUs.
I am talking to the TensorFlow folks to address this, but unfortunately there are a lot of competing priorities and it's not clear when this will be fixed.
Hi @maciejkula , appreciate you looking into this. Understood re: priorities. This is not a blocker here, can work around this for now, thanks.
@dgoldenberg-audiomack Does this problem reproduce in tf-nightly? NVIDIA added a GPU kernel for ResourceSparseApplyAdagradV2
in January, so it is possible that this just works now.
Any news on this bug?
I'm running the below type of code on a GPU enabled machine and I get the below error. Any idea as to what might be causing it and/or how to get around it?
Versions: TF = v.2.3.1 TFRS = v.0.3.2 TF IO = v.0.16.0
The exception + more of the log