Pinned post
Kitten 馃悎
Verified account This account is verified. Learn more@kitten_beloved
Adorable and harmless
645 Following 27K Followers 69K Posts
Latest posts 5
If AI progress has gone super exponential it's not showing up in the expense reports
Every benchmark improvement costs more than the one before
Getting exponential improvement via exponential spending increases is not RSI
Every benchmark improvement costs more than the one before
Getting exponential improvement via exponential spending increases is not RSI
This slide from the Anthropic IPO roadshow was a big factor in my decision to leave. It鈥檚 an insane plot and I鈥檓 surprised they made it public.
It shows the compute efficiency of pretraining over time. The y-axis is a **log scale**: they are seeing continued superexponential algorithmic progress.
Mythos was not a one-time deviation, but a point on a superexponential. They could already retrain Mythos with a fraction of the compute originally needed.
It doesn鈥檛 strictly imply a software-only intelligence explosion- they have been spending more on R+D. But algorithmic improvements do not require ridiculous creative genius and are on the verge of being automated. You do the math.View quoted post on X
Eric Carle has to be the most over glazed children's author of all time, his illustrations are ugly and the writing is mostly bad
He had one banger, the very hungry caterpillar, and then dozens of total stinkers (honorable mention for brown bear brown bear)
He had one banger, the very hungry caterpillar, and then dozens of total stinkers (honorable mention for brown bear brown bear)
It's actually astounding, when you stop and think about it, how many people are completely bought in on these semantic word games
When I tried to explain the concept of population replacement to @Noahpinion, and he blocked me after.
Population replacement has happened many times in human demographic history. But for some reason Noah Smith seems to think that it's a concept that cannot be meaningful, regardless of the amount of immigration.View quoted post on X
Spatial reasoning seems to be one of the remaining weak points for current LLMs
Having solved some of the more difficult ones after only a lot of thought I'm pretty confident this is one area I can still beat the machines
But I want someone to try
Having solved some of the more difficult ones after only a lot of thought I'm pretty confident this is one area I can still beat the machines
But I want someone to try
Has anybody tried getting Fable 5.2 to solve the harder puzzles in order of the sinking starView quoted post on X