2 ms·
"davinci" is the original GPT-3 (175B) which had too many parameters per Chinchilla scaling law. And parameter count is strongly correlated with inference cost.
by cubefox 4y ago
"davinci" is the original GPT-3 (175B) which had too many parameters per Chinchilla scaling law. And parameter count is strongly correlated with inference cost. GPT-3.5 is likely Chinchilla optimal and much smaller than davinci.
Though this theory has the defect that GPT-4 is, I think, more expensive than GPT-3, but as I recall it was considered unlikely that GPT-4 is larger than 175 billion parameters. Not sure.