I have been trying to understand the performance benefits of OneDNN for CPU inference (reference).
To this end, I am trying benchmark a pre-trained model with and without OneDNN optimizations. Inference with the traced model seems to be breaking with:
RuntimeError: Expected tensor for argument #1 'indices' to have one of the following scalar types: Long, Int; but got torch.FloatTensor instead (while checking arguments for embedding)
Any suggestions to resolve it?
Here’s my Colab Notebook: Google Colab.