Baseball Data Scientist. Opinions my own.
Wrigleyville
- Tweets 516
- Following 666
- Followers 516
- Likes 4,122
OpenAI made a huge mistake not renaming GPT-6 Sol to GPT-6 Terra once it became clear it was mogged by Opus 5.5
You had a whole tier sitting there to bail you out! Use it!
1
1
1,138
I’d like to ban everyone on Twitter from ever discussing the “max” reasoning mode ever again. The only reason it exists is to inflate benchmarks by pushing models to solve overly complex tasks in one shot that no reasonable person would ever ask a model to do all at once
1
1
138
Cal Aldred retweeted
I literally generated the first hundred or so, if anyone is curious. This is for both even and odd fiber degree: jacobianfun.org/counterexamp…
1
1
219
Coming soon from the machine gods: the infinite ladder is just one row of a two-parameter factory of Jacobian counterexamples. The original map sits at (1,1), and every mapping degree from 3 onward appears.
1
1
2,559
Came up with a way to generate an infinite family of counterexamples, via Fable + GPT 5.6 Sol
hello there the jacobian conjecture is false thanx to my close friend akhil for asking about it and my other close friend fable for working during the world cup final
((1+xy)^3 z + y^2 (1+xy) (4+3xy), y + 3 x (1+xy)^2 z + 3 x y^2 (4+3xy), 2 x - 3 x^2 y - x^3 z): \C^3\to \C^3, has jacobian determinant -2, and sends (0, 0, -1/4), (1, -3/2, 13/2), and (-1, 3/2, 13/2) to (-1/4, 0, 0)
46
179
3,273
398,749
There's an infinite ladder of this geometric construction too, per GPT 5.6 Sol and Fable 5! Integrate D^{m−1} for a quadratic D, mark a root of the primitive, forget the mark. Generically 2m−1 sheets, so no two rungs are polynomially equivalent.
GPT:
Take π: P¹ × Sym²(P¹) → Sym³(P¹), (p, {q,r}) ↦ {p,q,r}.
R be its ramification divisor;
H ⊂ Sym³(P¹) ≅ P³ be hyperplane tangent but not osculating to the small diagonal;
X := (P¹ × Sym²(P¹)) \ (R ∪ π⁻¹(H)) ≅ A³;
Y := Sym³(P¹) \ H ≅ A³.
π|X: X → Y is counterexample
10
148
72,010
Someone needs to make a benchmark for the benchmarks that benchmarks how hard it is for an AI agent to tell that the benchmark is in fact a benchmark
1
623
Cal Aldred retweeted
Our team is hiring a machine learning engineer. Come work with me and the rest of our baseball analytics team on building models touching all areas of Cubs baseball: my1060wd.wd5.myworkdayjobs.c…
4
7
55
67,826
Urbanism fun fact of the day: 11 of the 15 most densely populated urban areas in the US with over 300k population are in California
2
562
Cal Aldred retweeted
I’m a little late to the party but:
GC Kuss!
GC Kuss!
GC Kuss!
🦅🇺🇸🦅🇺🇸🦅🇺🇸🦅🇺🇸🦅🇺🇸🦅🇺🇸🦅🇺🇸
@LanterneRougeCP
1
4
514