Re:The bigger-is-better approach to AI is running out of road
by LASR on Saturday June 24, 2023 @05:48PM
(#36462945)
This is what I've been doing. GPT-4 to generate some data from some input, followed up by a 3.5T call to verify the output against the input to verify the content. You can feed the 3.5T output straight back into GPT-4 again and it will self-correct.
Doing this a couple of times gives me 100% accuracy for my use case that involves some level of summarization and reasoning.
Hallucinations are not as big of a deal at all IMO. Not enough that I'll just sit there and wait for models that don't hallucinate.
Doing this a couple of times gives me 100% accuracy for my use case that involves some level of summarization and reasoning.
Hallucinations are not as big of a deal at all IMO. Not enough that I'll just sit there and wait for models that don't hallucinate.