Teradata supports a range of text analytics functions using large language models available on various cloud and on-premise platforms, utilizing data stored in Vantage. This has been done by building new SQL functions into the core Vantage analytic database, which utilize the massively parallel processing power Teradata is known for.
You can customize these functions by manipulating their hyperparameters to efficiently process and analyze unstructured textual data from diverse sources such as emails, academic papers, social media posts, and product reviews. This capability allows users to derive insights with precision and depth that can match human analysis at much greater speed.
- AWS Bedrock
- Azure AI
- Google Gemini in the cloud
- NVIDIA NIM hosted models on premise using structured or unstructured textual data stored in Teradata database tables.
- Lite LLM Hosted Models
As these functions are built into the Vantage analytic database, they are callable using SQL. However, for those preferring a Pythonic user experience, Teradata Package for Generative AI, teradatagenai, can be used. This is a companion package to Teradata Package for Python, teradataml, that generates and executes SQL queries for the Text Analytics AI functions for you automatically based on a Python function signature.
Cloud Platform Syntax
Each function uses cloud platform-specific arguments and syntax. See Common Functionality or Rules.
Examples Input
The examples provided with each function requires the same input. See Text Analytics AI Function Example Input.
Message Column in Example Output
The Message column in the example output is used to report any error messages when the query fails or there is no data available from the external model. In such cases, the output column is NULL.
Supported Functions
You can access Text Analytics AI functions using the Teradata Package for Generative AI in 'In-DB' mode to provide a user-friendly interface for utilizing these functions.