News for hackers. Stuff that matters.
Re:GPT-3.5 Turbo fine-tuning and API updates by msp26 on Tuesday August 22, 2023 @05:52PM (#37229014)
8x is a big gap. I feel that it might be better to just use few-shot prompting and supply some examples (not applicable to every use case ofc) every prompt to steer the output.

But it's worth experimenting with at least.

Edit: no function calling (until later this fall) too. That's most of my usage right now so I'll pass. Curious on what the verdict on the OAI finetunes are. I remember reading this thread which indicated it wasn't really necessary: https://news.ycombinator.com/item?id=37174850

Re:GPT-3.5 Turbo fine-tuning and API updates by jondwillis on Tuesday August 22, 2023 @08:16PM (#37230319)
That’ll add latency.
Re:GPT-3.5 Turbo fine-tuning and API updates by explaininjs on Tuesday August 22, 2023 @09:55PM (#37230969)
and reduce maximum output length
Re:GPT-3.5 Turbo fine-tuning and API updates by sheepscreek on Tuesday August 22, 2023 @08:17PM (#37230325)
There’s a place for this. Like fine tuning the model on advanced particle physics so that it can decipher cutting edge research more accurately, without teaching it particle physics every time in the prompt.

It’s not a simple math calculation though.