Hacker News Reader

I just think the tech has been out for so long it's not as big of a deal. Mini-Gpt4 has been out for 6 months! Of course the descriptions aren't exactly gpt-4 grade, but with mistral 7b being used as the language model instead of llama 7b, the reasoning ability will improve noticeably.

[1] https://github.com/Vision-CAIR/MiniGPT-4

View on HN

Sure, the tech was out there for quite some time but never before the quality of the output was so good, it's almost (not 100%, there still are mistakes and hallucinations ocassionally) on par with a human, which to me is really stunning.

I've tried these kind of queries with other models (including Mini-GPT4) and never got any meaningful results until now. It’s the same thing with GPT 3.5/4 - sure, transformer models existed for few years already but ChatGPT crossed some kind of threshold in the quality of its output where finally people took notice.

Reply on news.ycombinator.com