This repository explores various techniques and use-cases for embedding in Machine Learning, with a particular focus on text embeddings and their applications.

Notebooks

Vertex AI Vector Search Introduction: An introduction to setting up and using Vertex AI Vector Search 2.0.
Import from BigQuery into Vector Search: Learn how to import vector embedding data from a BigQuery data source into a Vector Search index.
Use Gemini and OSS Text-Embedding Models Against Your BigQuery Data: Learn how to generate text embeddings using BigQuery in conjunction with both Gemini and OSS text embedding models.
Vertex AI Vector Search Quickstart: A quickstart guide to setting up and using Vertex AI Vector Search.
Introduction to Text Embeddings and Vector Search: Provides an introduction to text embeddings and their application in building vector search engines.
Hybrid Search: Demonstrates building a hybrid search system leveraging both keyword-based search and semantic similarity search with embeddings.
Embedding Similarity Visualization: Visualizes similarity relationships between embeddings using dimensionality reduction techniques like PCA and t-SNE.
Introduction to Multimodal Embeddings: Introduces the concept of multimodal embeddings, which combine information from different modalities like text and images.
Introduction to Embeddings Tuning: Explores techniques for fine-tuning pre-trained embedding models to specific domains and tasks.
Task-specific Embeddings: Explores techniques for creating embeddings specialized for different tasks.
Large-scale Embeddings Generation for Vector Search: Demonstrates large-scale embeddings generation for Vertex AI Vector Search.

Use Cases

Outlier Detection

Outlier Detection with BigQuery Vector Search (Audit Logs): Shows how to detect and investigate anomalies in audit logs using BigQuery vector search and Cloud Audit logs as an example dataset.
Outlier Detection with BigQuery Vector Search (Infra Logs): Demonstrates building a real-world outlier detection using Gemini and BigQuery vector search.