From Memories to Maps: Mechanisms of In-Context Reinforcement Learning in Transformers

OffRLRe-assign community