An irrepressible, uncontainable, uncontrolled stream of papers proposing the use of pretrained Large Language Models (LLMs) for sequential decision-making is manifesting itself. Armed with hand-crafted representations of the world and pages and pages of carefully-engineered prompts, LLMs seem to show unprecedented abilities that allow them
This is a great piece - I hope you keep writing. I am personally going to fight to keep the name RL because it's the only term that connects multiple things in my mind: time/feedback and reward. It's important.
This is a great piece - I hope you keep writing. I am personally going to fight to keep the name RL because it's the only term that connects multiple things in my mind: time/feedback and reward. It's important.
I wrote about the identity crisis from a different way, here, saying there are three metaphors for RL: https://www.interconnects.ai/p/rl-tool-or-framework-or-agi
Thank you, Nathan! I liked your post. And yes, I'll keep writing -- stay tuned!