balefulboy

5 ms·

1.
by balefulboy 13d ago
Well it's not exactly saturated when OAI refused to use the harness explicitly provided by ARC-AGI. I'm not really familiar enough with the benchmark to declare whether it's a perfect measure for AGI, but I kind of doubt it i
2.
by balefulboy 13d ago
Because they saw how much hype Glasswing was getting in April
3.
by balefulboy 13d ago
I'm always tired of seeing at the top of every new model release post on here. I say Simon should just keep it to Twitter.
4.
by balefulboy 13d ago
Don't forget the 3D demos. My favorite is in the house tour where the sink and stovetop(?) are obviously very misaligned from the counters
5.
by balefulboy 13d ago
72 to 74 on DeepSWE is AGI
6.
by balefulboy 2mo ago
it seems the older models were capped at 10kusd for the runs though?
7.
by balefulboy 3mo ago
Damn, that beam of light was a flashbang. I wouldn't call this tasteful UI design, but maybe I just need to go to sleep.
8.
by balefulboy 3mo ago
Greentext is eh. Very formulaic, in fact very similar to the bottomless pit one, which I'd argue is better because of it's absurdity. I have to ask, did you mention the older GPT version to fable in the prompt?
9.
by balefulboy 3mo ago
i got a 502
10.
by balefulboy 3mo ago
METR's time horizon is not a reliable metric of LLM capability growth: https://www.transformernews.ai/p/against-the-metr-graph-codi...
11.
by balefulboy 3mo ago
yeah man this sucks. i genuinely do not know how people find this stuff appealing
12.
by balefulboy 3mo ago
I still don't know why people are saying this. I don't really code but from what I've heard on here the models haven't improved since Opus 4.5