enough horsing around time to be a productive member of society
my predictions for the future:
1) I don’t finish this paper and it keeps going in circles
2) the Iran deal falls through
3) France gets annexed by the United States so that Mistral can be nationalized
Rumour mill going crazy on this new mistral model
- Napoleon class model with >10T params
- smokes Mythos on VoltaireBench
- for safety reasons only outputs French language code
the infinite path of least resistance is spiritual death. a rot that has emptied you from the inside and left you as a husk.
at least act like blood still flows through your veins…
foam shoes make me so mad in like 99% of cases
YOU ARE NOT A GERIATRIC
WEAR SOMETHING REAL YOU COWARD
oh yeah let me wear my running shoes for leisure unless your hip or knees are giving out you need to wear something else there are no two ways about this
fundamentally I abhor this notion. a good partner in my eyes is absolutely not impotent in this way. a good partner and one that really cares about you for that matter is both willing and able to crucify you when the moment calls for it.
there seems to be this notion some people have in relationships that if someone loves you / cares about you they’ll treat you invariantly to how you act. eg even if you’re horrible to them or just in a bad way you want to be treated with the same degree of kindness.
there are even shit tests designed around this where people intentionally act in horrific ways to see if their partner still treats them well, and if they don’t continue to treat them in the same way that’s evidence that they in fact didn’t really love them.
I mean to be frank it’s quite clear at this point frontier models are considerably better than me or than I ever could’ve been at mathematics almost surely!
that’s how it goes brother, midwit down…
Claude Fable 5 scores very well on FrontierMath: Tiers 1–4 (v2), reaching 87% on Tiers 1–3 and 88% on Tier 4. This continues a streak of Anthropic models improving rapidly at math.
FrontierMath: Tiers 1–4 (v2) is live.
We concluded an audit that addressed errors in 42% of problems. Rankings are similar but scores are higher across the board. The current leaders are GPT-5.5 (xhigh) with 85% on Tiers 1–3 and Google’s AI co-mathematician with 76% on Tier 4.
Most companies didn’t budget for the exorbitant cost of AI tokens, but no company can function without them, Box CEO Aaron Levie told Semafor’s @ReedAlbergotti at the Semafor Tech summit in San Francisco.
https://t.co/N3DC89Rh9x
Nobody warned me that if you start feeling your feelings all at once there’s a chance you go temporarily insane
Like “has an ambulance called for you at the airport” insane
there are a lot of people that behave in an almost deterministic way and they’re tremendously telegraphed in the way they operate. if you have even a remotely decent theory of mind for them then you rarely need recent information to know what they’re doing.