- H2O AI became a launch partner for Snowflake's Snowpark Container Services, letting joint customers train, deploy, and score models inside their Snowflake account.
- H2O-3 and Driverless AI model training moved directly inside Snowflake through the Snowpark Container Services integration, removing the need for external data movement.
- Users could deploy H2O Driverless AI, H2O-3, eScorer, and Notebooks by starting and stopping instances within the Snowpark Container Services environment, according to the post.
- The Snowpark Container Services integration let Driverless AI run on instances with different CPU, memory, and GPU configurations.
H2O.ai today announced its participation as a launch partner for Snowflake’s Snowpark Container Services (available in private preview), which provides our joint customers with the flexibility to train, deploy, and score models all within their Snowflake account. This further expands the ease of use for data science teams to create machine learning models and deployment teams to efficiently score models by leveraging the performance, scalability, and unified governance of the Snowflake Data Cloud.
We are excited to announce that model training with H2O-3 and Driverless AI can now happen directly within Snowflake, eliminating the need for external data movement! With this deepened integration, we’re building on our existing ability to score models in Snowflake and bringing the full H2O.ai machine learning experience to the Data Cloud. The result is a fully-interactive, feature-rich experience, including distributed training and improved scoring performance powered by a graphical interface.
Key Benefits of using H2O.ai with Snowpark Container Services:
- Eliminate data movement from Snowflake
- Quickly and easily self-manage the machine learning environment
- Seamlessly collaborate with the Snowflake Data Cloud’s ecosystem of customers, partners, and solutions
- Run multiple product versions concurrently
How it works:
Using the H2O.ai application, a user can deploy H2O Driverless AI, H2O-3, eScorer and Notebooks by starting and stopping instances within the Snowpark Container Services environment.
The ability to use H2O-3 for distributed training within the Snowpark Container Services environment enables data scientists to leverage the data in Snowflake without data movement between environments, enabling large amounts of data to be used for model training.
Driverless AI can be used on a variety of different Snowpark Container Services instances, to leverage different CPU, memory configurations, and even GPUs.