Is Jev secretly learning a value function? Calibration in System 1/2 switching
The author places Jev in a System 1/2 decision framework, discussing when to answer fast versus switch to deeper reasoning, and maps predicting task success to the RL Q-function.
Is Jev secretly learning a Value function? A model can always choose “A” without being certain that “A” is correct. That distinction matters when an agent must decide whether to act fast or switch to deeper reasoning. We spent some time exploring: 1. How Jev fits into System 1/2 decision-making and why calibration matters for knowing when to switch 2. Why maximising expected reward in RL does not guarantee calibrated probabilities. 3. How predicting eventual task success in agentic tasks connects to a familiar RL quantity: the Q-function. Could value prediction help build a calibrated System 1—and could that be part of the story behind Jev? Link: https://souradip-chakraborty.github.io/blog/jev-rlcd/ #Jev #RLCD #System1