DLRM Recommendation Model Training

**Overview**
This is a sample implementation of the [DLRM Recommendation Model](https://arxiv.org/abs/1906.00091) on Databricks with the following features:
1. [TorchRec](https://pytorch.org/torchrec/): for handling large datasets with many categorical features (where the embedding tables can't fit inside one GPU)
2. [TorchDistributor](https://docs.databricks.com/en/machine-learning/train-model/distributed-training/spark-pytorch-distributor.html): for doing distributed training on Databricks
3. [Mosaic StreamingDataset](https://docs.mosaicml.com/projects/streaming/en/stable/): for efficient data loading in a distributed environment

**Use case**
The DLRM model is an effective architecture for predicting click-through rates. The example provided uses a randomized synthetic dataset that has a similar layout to other publicly available datasets, such as the [Criteo](https://ailab.criteo.com/download-criteo-1tb-click-logs-dataset/) click logs dataset. The provided example notebooks can be updated to support those public datasets as well as your own recommendation use cases.

**Product details**
For more specific details, refer to the embedded notebook which contains a guide for how to train deep recommendation models on Databricks.

**Licenses**
- The implementation of DLRM on Databricks is based on this [repository](https://github.com/facebookresearch/dlrm/tree/main) by Facebook with the MIT License.