On: About That OpenAI "Breakthrough"

[shwaj]

They’re not assuming that the Q refers to the Transformer’s “query”, they’re talking about the reward function in reinforcement learning, using chess as an example.