Yuzhang Luo
Titles as the sources gave them. In 1 entry, with 0 checked quotations.
In the Base
Every entry they appear in.
Represented but Not Used: Value-Policy Dissociation in LLM Agents
According to the abstract, Yuzhang Luo and colleagues study three open models and find that the relative value of candidate actions can be read from hidden states even where the model prefers a worse action, which they call value-policy dissociation.