Did you find it? I'm interested to see if it's testable. I think ChatGPT just drops context that doesn't fit in the context window anymore - and then based on c

Did you find it? I'm interested to see if it's testable. I think ChatGPT just drops context that doesn't fit in the context window anymore - and then based on context it continues the conversation. within the 8k context window, it works really well because ppl keep revisiting important bits, so there isn't much lost there. I haven't seen OpenAI talk about condensing prior context, and I don't think ChatGPT has any other knowledge about itself other than what is in it's system part of the prompt. It doesn't really know what happens to the prior context, it just wants to finish the document that is being created.