Line 113 seems redundant. `sparse_group.size == sparse_group_size` is always (will eventually be) true, following the `while` on line 108. https://github.com/harvard-acc/DeepRecSys/blob/1383176a463aeed25f85f1aeb440a23f5d423013/data_generator/dlrm_data_caffe2.py#L113