Found 1 document(s)
Page 1 of 1
-
Outcome-predictive signals are often treated as promising optimization targets in reinforcement learning, but predictive validity and optimization utility are different claims. We expose this distinction in GuanDan, a mu...