Databricks
457 Case Studies
A Databricks Case Study
Dynamo AI, founded in 2022, needed a secure and compliant way to build multilingual foundation models for enterprise GenAI, but faced challenges with model training efficiency, memory usage, and preparing large multilingual datasets. To accelerate development, Dynamo AI used Databricks Mosaic AI Training to support its AI evaluation, remediation, and deployment workloads.
Databricks helped Dynamo AI streamline training with out-of-the-box scripts and hands-on guidance from the Databricks AI Research team, improving efficiency and resolving a memory leak issue. As a result, Dynamo AI trained an 8-billion-parameter multilingual LLM in just 10 days, about 20% faster than competitors, while saving weeks of development time.
Eric Lin
Head of Applied AI