The Amazon Bedrock platform offers multiple APIs to interact with various large language models. The Embedding model APIs, such as that of Amazon Titan Multimodal G1, is used to measure the relatedness of text strings. These floating-point measurements, or embeddings, can then be used as input to other machine learning models for further data analysis.
The Teradata API Integration enables Teradata customers to obtain Bedrock text embeddings from Teradata Vantage data, using a model provided directly from Amazon Bedrock. Users can request text embeddings from a Bedrock model using Teradata input tables or queries, and benefit from the integration of embeddings with Teradata in-database analytics functions.
This integration enables end users to solve and address text analysis problems including the following and more:
- Text classification
- Recommendation
- Information retrieval
- Clustering
- Anomaly detection
- Text search
- Translation
Users can choose from several embedding models provided by Amazon Bedrock that are tailored for the computation of embeddings.
The Teradata API Integration with Amazon Bedrock includes the following:
- API_Request In-database Function
This in-database function enables Vantage users to connect with an endpoint for scoring using a Vantage query (SQL).
- teradataml Extension Library (tdapiclient)
This client package provides Python interface to execute API_Request in-database function.
API_Request In-database Function
Typical steps to use API_Request in-database function with Amazon Bedrock are:
- Data cleaning, feature engineering inside Teradata database.
- Exporting data for requesting embeddings to Bedrock.
- Performing in-database analytics on the embeddings obtained.
teradataml Extension Library
The tdapiclient package provides Python interface to execute API_Request in-database function.