In Short:
MosaicML was founded to make AI models more accessible by allowing users to train, improve, and monitor open-source models using their own data. The company was acquired by Databricks, releasing one of the highest performing open-source LLMs, DBRX. Databricks believes in the power of open source and hopes the industry embraces it for advancements in AI. Frankle emphasizes the importance of accessibility, customization, and openness in AI development.
The impact of artificial intelligence can be limited if only one company builds and controls the models and data. MosaicML aims to make AI models more accessible by allowing users to train, improve, and monitor open-source models using their own data. Founded by Jonathan Frankle PhD ’23 and MIT Associate Professor Michael Carbin, MosaicML developed a platform to democratize deep learning, particularly as interest in generative AI and large language models grew after the release of Chat GPT-3.5. MosaicML’s acquisition by Databricks, a global data storage, analytics, and AI company, led to the development of DBRX, an open-source LLM that has set new benchmarks in performance.
DBRX is recognized for its speed and utility, especially at large enterprises. The goal of empowering organizations to leverage AI without compromising data privacy has been a driving force behind MosaicML’s and Databricks’ collaboration.
Making algorithms efficient
Jonathan Frankle’s journey from Princeton University to MIT, focusing on deep learning, was shaped by the evolving landscape of AI. With the emergence of transformative deep-learning architectures such as the transformer, enthusiasm for AI grew, leading to the founding of MosaicML by Frankle, Michael Carbin, and Hanlin Tang. MosaicML’s innovative techniques in model training and efficiency paved the way for the development of an open-source LLM and a library of methods for data scientists.
The democratization of AI through open-source initiatives has been a core part of MosaicML’s mission, enabling a wider range of companies to engage in generative AI.
Open sourcing innovation
The acquisition of MosaicML by Databricks for $1.3 billion in 2023 marked a significant milestone in the collaboration between academia and industry. The release of DBRX by Databricks has expanded the capabilities of open-source LLMs, providing enterprises and the broader community with enhanced access to cutting-edge AI models.
Jonathan Frankle underscores the importance of openness and collaboration in advancing AI. As Databricks continues to innovate in the AI space, the emphasis on open-source contributions plays a key role in driving progress and scientific discovery.
