801: Merged LLMs Are Smaller And More Capable — with Arcee AI's Mark McQuade and Charles Goddard

Enterprise Success

Arcee AI has demonstrated significant success in the enterprise sector by providing tools that enable companies to train their own language models efficiently. Charles Goddard highlights their dual approach of offering a tool suite for model training and releasing their own high-performing models like RC Spark, which outperformed larger models such as GPT 3.5 in specific evaluations 1. This strategy positions Arcee AI uniquely in the market, allowing enterprises to leverage cutting-edge AI without compromising on performance or size. Mark McQuade explains their innovative evolutionary model merging technique, which uses algorithms to optimize model parameters, enhancing flexibility and power in AI applications 2.

We deliver it as a product with a, you know, an SDK and a UI. What that is, is a set of tools to enable companies to train their own models, right?

--- Charles Goddard

   

Custom LLMs

Businesses are increasingly recognizing the value of customizing language models to fit their specific needs, offering a competitive edge in their markets. Charles Goddard emphasizes the importance of proprietary models trained on unique data, which act as a moat against competitors 3. This approach not only ensures data privacy but also enhances the scalability and effectiveness of AI solutions. Jon Krohn notes that data remains a critical asset, akin to gold, and maintaining control over it is essential for leveraging AI's full potential 4.

Data is the new gold. Everyone has data. Let's keep our data, let's do, we'll build these massive data warehouses.

--- Charles Goddard