Merlin Models Example Notebooks

The example notebooks demonstrate how to use Merlin Models with TensorFlow on a variety of datasets.

Inventory

Getting Started: Develop a Model for MovieLens: Get started with Merlin Models by training Facebook’s DLRM architecture with only 3 commands.
From ETL to Training RecSys models - NVTabular and Merlin Models integrated example: Learn how the Schema object connects the feature engineering and training steps. You’ll learn how to apply Merlin Models to your own dataset structures.
Iterating over Deep Learning Models: Explore different ranking model architectures, such as Neural Collaborative Filtering (NCF), MLP, DRLM, and Deep & Cross Network (DCN).
Exporting Ranking Models: Save a Workflow object and a Model object in preparation of deploying the model to production.
Two-Stage Recommender Systems: Build a two-tower model using synthetic data that mimics the Ali-CCP: Alibaba Click and Conversion Prediction dataset.
The Next Step: Define Your Own Architecture: See how to combine pre-existing blocks to define a custom model.

Running the Example Notebooks

You can run the examples with Docker containers. Docker containers are available from the NVIDIA GPU Cloud. Access the catalog of containers at http://ngc.nvidia.com/catalog/containers.

Depending on which example you want to run, you should use any one of these Docker containers:

Merlin-Tensorflow-Training (contains Merlin Core, Merlin Models, NVTabular and TensorFlow)
Merlin-Inference (contains Merlin Core, Merlin Models, Merlin Systems, NVTabular, TensorFlow and Triton Inference Server)

Alternatively, you can [install Merlin Models from source](…/README.md#installing-merlin-models-from-source() and other required libraries to run the notebooks on your host.

To run the example notebooks using Docker containers, perform the following steps:

Pull and start the container by running the following command:
```
docker run --gpus all --rm -it \
  -p 8888:8888 -p 8797:8787 -p 8796:8786 --ipc=host \
  <docker container> /bin/bash
```
The container opens a shell when the run command execution is completed. Your shell prompt should look similar to the following example:
```
root@2efa5b50b909:
```
Install JupyterLab with pip by running the following command:
```
pip install jupyterlab
```
For more information, see the JupyterLab Installation Guide.

Start the JupyterLab server by running the following command:

jupyter-lab --allow-root --ip='0.0.0.0'

View the messages in your terminal to identify the URL for JupyterLab. The messages in your terminal show similar lines to the following example:

Or copy and paste one of these URLs:
http://2efa5b50b909:8888/lab?token=9b537d1fda9e4e9cadc673ba2a472e247deee69a6229ff8d
or http://127.0.0.1:8888/lab?token=9b537d1fda9e4e9cadc673ba2a472e247deee69a6229ff8d

Open a browser and use the 127.0.0.1 URL provided in the messages by JupyterLab.
After you log in to JupyterLab, navigate to the /merlin/examples directory to try out the example notebooks.