Use GPT-3 incorrectly: reduce costs 40x and increase speed by 5x

Re:Use GPT-3 incorrectly: reduce costs 40x and increase spee... by underanalyzer on Tuesday February 07, 2023 @11:50PM (#34704267)

Wow some of these comments seem harsh. The author isn't claiming to have invented these techniques and it's cool to see them being applied in the "real world" to save money.

[ Parent | on HN ]

Re:Use GPT-3 incorrectly: reduce costs 40x and increase spee... by Buoy on Wednesday February 08, 2023 @12:06AM (#34704362)

Yep, that was the aim, I'm just trying to put some stuff out there that helped us out!

[ Parent | on HN ]

Re:Use GPT-3 incorrectly: reduce costs 40x and increase spee... by JeremyHerrman on Wednesday February 08, 2023 @12:16AM (#34704409)

Agreed. Fine-tuning GPT-3 is often called out as expensive and sometimes unnecessary given K-shot & embeddings. I'm glad they shared their success story even if it's not ground-breaking.

[ Parent | on HN ]

Re:Use GPT-3 incorrectly: reduce costs 40x and increase spee... by Buoy on Wednesday February 08, 2023 @12:26AM (#34704474)

Thanks - we used to use a bunch of k-shot prompts (particularly with our previous idea before we pivoted when we got into YC), but with the davinci model we were sending ~3.5k tokens per invocation, which in the long term was costing far more than finetuning!

[ Parent | on HN ]