Effective skills can get state-of-the-art performance, even when using weaker models.

· Bits and Bobs 6/1/26
  • Effective skills can get state-of-the-art performance, even when using weaker models.
    • For example, see this paper: GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning.
    • There's a massive amount of capital and effort going into improving frontier models.
    • There's less going into improving the structures we use to extract value from models.

More on this topic

From other episodes