That’s a lot of rounds (and money), but we’re slowly getting to the point where they might exhaust the alphabet before going public (which I believe they should try doing just because it would be funny). Maybe Databricks is the company to push us towards different character systems that provide a higher base than the 26 uppercase letters? The sky is the limit!
I have nothing against databricks by the way, I used their “Dolly” dataset for my bachelor’s thesis (good times). Now MosaicML seems to be more focused on getting you to deploy your models on Databricks (or to be precise to use their Mosaic AI Gateway - why would you call this the same as your research lab?). Come on, just drop DBRX 2 for the nostalgia.
(Also, I know that this is a strange first blog post, but if I later publish other ones, I can change the order, and you would never know.)