Snowflake ML now supports distributed processing capabilities for training multiple models and processing data across partitions.
You can use Many Model Training (MMT) to train multiple machine learning models efficiently across data partitions. MMT partitions your Snowpark DataFrame by a column that you specify and trains separate models on each partition in parallel.
You can use the Distributed Partition Function (DPF) to process data in parallel across one or more nodes in a compute pool. DPF partitions your Snowpark DataFrame by a column that you specify and executes your Python function on each partition in parallel.
Both features help you handle infrastructure complexity and scale automatically.
For more information, see Train models across data partitions and Process data with custom logic across partitions.
Source: Snowflake
Latest Posts
- Dynamic video tile resizing based on occupancy count from Teams Rooms on Android [MC1148542]
- Microsoft Teams: Collaborative Notes in Chats [MC1148543]
- Simultaneous role-specific Teams Town hall and Webinar views in Teams Rooms on Windows [MC1148546]
- Microsoft 365 admin center – Usage reports: Agent usage (preview) [MC1148545]