[BUG] Properly preparing NVTabular preprocessed data for HugeCTR inference

[vinhngx]

Describe the bug
NVTab encodes categorical data is not directly compatible with HugeCTR requirement and would require the value to be shifted - so that they occupy different integer ranges

Concrete conceptual example: provided there are 10 users and 10 items, NVtab encodes users in 0-9 range, and items in 0-9 range. HugeCTR expect users in 0-9 range and items in 10-19 range

Steps/Code to reproduce bug
A concrete example of shifting:

https://gitlab-master.nvidia.com/dl/hugectr/hugectr/-/blob/master/notebooks/ecommerce-example.ipynb

Expected behavior
A clear and concise description of what you expected to happen.

Environment details (please complete the following information):

• Environment location: [Bare-metal, Docker, Cloud(specify cloud provider)]
• Method of NVTabular install: [conda, Docker, or from source]
  • If method of install is [Docker], provide docker pull & docker run commands used

Additional context
Add any other context about the problem here.