Built with Curiosity Over Compute

The tagline sounds nice. What it really means is we could not afford the compute so we got curious instead.

When you have a modest GPU, you start asking different questions. Not "how do we scale to trillion parameters?" but "how do we make what we have work better?" Not "what is the biggest model we can train?" but "what is the smartest model we can build with what we have?"

Curiosity over compute is not a philosophy. It is a survival strategy. It is what happens when you look at your electricity bill and then look at your bank account and then decide to get creative.