Meme benchmarks like this and Strawberry are funny but very easy to game, I bet ... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		arnaudsm 6 months ago \| parent \| context \| favorite \| on: Gemini 2.5 Deep Think Meme benchmarks like this and Strawberry are funny but very easy to game, I bet they're all over training sets nowadays.

simonw 6 months ago [–]

If you train a model on the SVGs of pelicans on a bicycle that are out there already you're going to get a VERY weird looking pelican on a bicycle: https://simonwillison.net/tags/pelican-riding-a-bicycle/

nickandbro 6 months ago | [–]

Thanks for the mention on your blog. You’re the original GPT pelican artist

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact