News for hackers. Stuff that matters.

A comment on Sail 7B: New Fine Tuned LLM Outperforms ChatGPT and Vicuna with Search.

Re:Sail 7B: New Fine Tuned LLM Outperforms ChatGPT and Vicun... by phillipcarter on Monday June 05, 2023 @01:35PM (#36199865)
> outperforms ChatGPT

Doubt.

I mean, maybe in a while, but it seems like the general trend is that a month or two after these announcements, independent researchers end up finding that the opposite is actually true.

[ Parent | on HN ]
Re:Sail 7B: New Fine Tuned LLM Outperforms ChatGPT and Vicun... by hyluo on Tuesday June 06, 2023 @01:44PM (#36216462)
The model outputs ChatGPT by connecting to a search engine - it's expected because it is fed with search results. Without search, SAIL-7B does not outperform ChatGPT.

The interesting aspect is that the ChatGPT response quality is not improved with search results, and it can be misled by noisy search.

[ Parent | on HN ]