DEV Community

Discussion on: How to Evaluate LLM Applications

Collapse
 
matijasos profile image
Matija Sosic

This is an interesting topic, thanks for sharing! We've been playing a lot with GPT models (built our coding agent for React & Node.js apps - usemage.ai/) so we started to develop an intuition on which one is better for what (GPT4 - creative, more complex tasks, GPT3.5 - simpler tasks with a lof of context upfront), so it's useful to see a more "official" approach to it :).