> "Before answering, quietly think about whether "
I thought generating text is the only way for GenAI/LLM models to "think".
How exactly does ChatGPT "quietly think"?
Is there text generation happening in layers where some of the generated text is filtered out / reprocessed and fed back into another layer of text generation model before a final output is shown to the user as a respose on UI? So a "thinking" layer separate from a "speaking" layer?
I thought generating text is the only way for GenAI/LLM models to "think".
How exactly does ChatGPT "quietly think"?
Is there text generation happening in layers where some of the generated text is filtered out / reprocessed and fed back into another layer of text generation model before a final output is shown to the user as a respose on UI? So a "thinking" layer separate from a "speaking" layer?