While there is some mention of chain-of-thought prompting, the idea of chain-of-thought can go further, and can actually solve the problem exactly.
The LLM can produce working code (or broken code that it can fix from reading the errors), which, when paired with a sandbox, can give the result of the computation back to the LLM.
The direction things are going is one where some tokens output by the LLM are hidden, just like human thoughts. That is what Bing Chat is doing, for instance.
The LLM can produce working code (or broken code that it can fix from reading the errors), which, when paired with a sandbox, can give the result of the computation back to the LLM.
The direction things are going is one where some tokens output by the LLM are hidden, just like human thoughts. That is what Bing Chat is doing, for instance.