https://github.com/GoogleCloudPlatform/vertex-ai-samples

Notebooks, code samples, sample apps, and other resources that demonstrate how to use, develop and manage machine learning and generative AI workflows using Google Cloud Vertex AI.
https://github.com/GoogleCloudPlatform/vertex-ai-samples

automl colab colab-enterprise gemini gemini-api genai generative-ai google-cloud-platform ml mlops model model-garden notebook pipeline predictions samples vertex-ai vertexai workbench

Last synced: 20 days ago
JSON representation

Notebooks, code samples, sample apps, and other resources that demonstrate how to use, develop and manage machine learning and generative AI workflows using Google Cloud Vertex AI.

• Host: GitHub
• URL: https://github.com/GoogleCloudPlatform/vertex-ai-samples
• Owner: GoogleCloudPlatform
• License: apache-2.0
• Created: 2021-05-27T00:06:43.000Z (almost 4 years ago)
• Default Branch: main
• Last Pushed: 2024-10-29T11:59:58.000Z (6 months ago)
• Last Synced: 2024-10-29T14:32:39.388Z (6 months ago)
• Topics: automl, colab, colab-enterprise, gemini, gemini-api, genai, generative-ai, google-cloud-platform, ml, mlops, model, model-garden, notebook, pipeline, predictions, samples, vertex-ai, vertexai, workbench
• Language: Jupyter Notebook
• Homepage: https://cloud.google.com/vertex-ai
• Size: 94.4 MB
• Stars: 54
• Watchers: 35
• Forks: 4
• Open Issues: 21
• Metadata Files:
  • Readme: README.md
  • Contributing: CONTRIBUTING.md
  • License: LICENSE
  • Code of conduct: CODE_OF_CONDUCT.md
  • Codeowners: CODEOWNERS
  • Security: SECURITY.md

Awesome Lists containing this project

• awesome-generative-ai-data-scientist - GitHub
• awesome-generative-ai-data-scientist - Google Vertex AI Examples

README

        
# Step-by-Step Demo

This demo showcase how to use (1) custom training, (2) custom hyperparameter

tuning, and (3) custom prediction serving over endpoints with

[Vertex AI](https://cloud.google.com/vertex-ai) to build a contextual bandits

based movie recommendation system. We implement the RL training and prediction

logic using the [TF-Agents](https://www.tensorflow.org/agents) library. We also

illustrate how to use TensorBoard Profiler to track the training process and

resources, allowing speed and scalability analysis.



We use the

[MovieLens 100K dataset](https://www.kaggle.com/prajitdatta/movielens-100k-dataset)

to build a simulation environment that frames the recommendation problem:

1.  User vectors are the environment observations;

2.  Movie items to recommend are the agent actions applied on the environment;

3.  Approximate user ratings are the environment rewards generated as feedback

    to the observations and actions.

For custom training, we assume the system dynamically interacts with the

environment in real time so that the target policy is the same as the behavior

policy: At each time step, we interact with the envionment to obtain an

observation, query the current policy for an action given said observation, and

lastly obtain a reward from the environment given the aforementioned observation

and action; then we use these pieces of data to train the policy.

The demo contains a notebook that carries out the full workflow and user

instructions, and a `src/` directory for Python modules and unit tests.

Read more about problem framing, simulations, and adopting this demo in

production and to other use cases in the notebook.