used this to build
SmythOS(394 points)
Llama models + fine-tuning = Fast, Performance, Affordable. We use a blend of OpenAI, Anthropic and Llama for our SmythOS product. Usually when we start we use either OAI or Anthropic, and when something is popular/in scale demand, we then train a Llama model to speed things up and reduce costs. Synthetic data is where it's at. To do that, we build classifier agents in SmythOS, and we also build agents to evaluate and compare model providers and identify gaps in our data set.
