logoalt Hacker News

ilaksh • today at 6:34 PM • 4 replies • view on HN

Something like 2/3 of the comments in this post when I started reading were reviews of previous GPT models stated almost as if they were important information for evaluating the new models. I feel that when a new model comes out, it would be more helpful to refrain from leaving a review until you have actually done some minimal testing of the model.


Replies

seizethecheese • today at 8:59 PM

Disagree, you’re guaranteeing that comments are either shallow or seeded by labs by people given early access.

For 6.2, it’s fair game to mention 6.1. OpenAI is signaling it’s an update to the same model with the naming convention. Maybe for 7 it won’t be.

mirsadm • today at 7:02 PM

Most of the comments are bots. I wouldn't take them seriously.

antupis • today at 6:42 PM

But it usually makes sense new models probably use somewhat same training pipelines and data.

epolanski • today at 8:00 PM

They complain about vibe coding, just to shove vibe reviewing and vibe benchmarking.