Disable pruning on embedding look up operators when compressed_indices_mapping = {0} (#48672)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/48672
When a user specifies `pruned_weights = True`, compressed_indices_mapping = {0}, it means they have not pruned the weights. In this case, we need to go through non-sparse kernels for embedding bag lookup.
Test Plan:
buck test //caffe2/test:quantization
https://www.internalfb.com/intern/testinfra/testconsole/testrun/3377699760676256/
Reviewed By: radkris-git
Differential Revision: D25252904
fbshipit-source-id: 3a97dfd41ec8113d61135f02d9f534df3419e81f