alvinjamur

What you need will come to you, if you do not ask for what you do not need. - Nisargadatta Maharaj

One theorem every ML engineer should know: The Johnson–Lindenstrauss Lemma. It states that high-dimensional data can be projected into a much lower-dimensional space while approximately preserving pairwise distances. Why it matters: • Explains why random projections work • Enables scalable learning in high dimensions • Used in embeddings, compressed learning, and ANN search • Helps fight the curse of dimensionality The surprising part: You can reduce dimensions dramatically without destroying the geometry of the data. That’s why many ML systems can operate efficiently even with massive feature spaces. Modern representation learning is deeply connected to this idea: Good embeddings preserve structure while compressing information. In ML, compression is often not loss of intelligence — it’s removal of redundancy.

Earlier today, a Geometer friend left me this voice mail : "Hey, Al! Please don't forget.... Data are the domain of facts, likelihoods are the domain of logical deduction, and the priors are the domain of theoretical prejudice. Call me."

😳🙌🏽

May the fourth be with you

@larpcapitalwc 😳

@alvinjamur Who is Timothy masters

If you’d rather not fall back on these generative approaches to Quant Finance (replete with intellectual masturbation & beginner errors for answers) AND are a beginner and really want to learn. Timothy Masters can come next…

Once again, I found ya a gem publicly available on arXiv, this time around, it is Bruno Martelli's 'An Introduction to Geometric Topology', a fantastic book on the topology and geometry of surfaces and three-manifolds, which would be useful to both mathematicians and physicists. If you are curious and you have a standard background in topology and analysis, you will be able to pick this one up quickly, including differential topology, Riemannian geometry, measure theory and homology. gl hf 👇👇👇👇

Today, MIT & the IMO released MathNet, the world’s largest dataset of International Math Olympiad problems & solutions 🌍 MathNet is 5x larger than previous datasets & is sourced from over 40 countries across 4 decades: bit.ly/4u1bhBC

As someone who started using Claude Code approx 2 weeks after its release, I appreciated reading this more than the source : arxiv.org/pdf/2604.14228

Milla Jovovich (actress from The Fifth Element) created a world-beating Claude memory system with @bensig?! - 100% on LongMemEval — first perfect score ever recorded. Free and 100% open source. Github link in the quoted post from Ben. I'm keen to hear how it works for you.

@HF_Trader now, u are a “mountain man”!

Backyard views this am…..late season nuke very welcomed. 3 seasons ago we had a record year. Alta had more snow than any place in the world (903in). Now we had another record year but opposite end of the spectrum. Worst year in history here at Alta. Last week it was 85 degrees all week in the valley and the snowpack got decimated causing several resorts to close early for the season. The resorts in the cottonwoods held on strong and this late storm brought nearly 3ft of snow. This storm is the best skiing we have had all year long. Pretty wild if you think about it.

Running OpenClaw with Gemma 4🦞 Free Open Source Local Model Device: MacBook Air M4 16Gb

Qwen releases Qwen3.6-Plus - 1M context window - swe-bench verified at 78.8 (opus at 80.9) - outperforms Claude opus 4.5 and comes close on select benchmarks - stronger coding model - understands images and screens like a real user - more reliable in real tasks

So astute.

Droids can now pursue goals autonomously over multi-day horizons. You describe what you want, approve the plan, and come back to finished work. We call these Missions.

Introducing Hyperagents: an AI system that not only improves at solving tasks, but also improves how it improves itself. The Darwin Gödel Machine (DGM) demonstrated that open-ended self-improvement is possible by iteratively generating and evaluating improved agents, yet it relies on a key assumption: that improvements in task performance (e.g., coding ability) translate into improvements in the self-improvement process itself. This alignment holds in coding, where both evaluation and modification are expressed in the same domain, but breaks down more generally. As a result, prior systems remain constrained by fixed, handcrafted meta-level procedures that do not themselves evolve. We introduce Hyperagents – self-referential agents that can modify both their task-solving behavior and the process that generates future improvements. This enables what we call metacognitive self-modification: learning not just to perform better, but to improve at improving. We instantiate this framework as DGM-Hyperagents (DGM-H), an extension of the DGM in which both task-solving behavior and the self-improvement procedure are editable and subject to evolution. Across diverse domains (coding, paper review, robotics reward design, and Olympiad-level math solution grading), hyperagents enable continuous performance improvements over time and outperform baselines without self-improvement or open-ended exploration, as well as prior self-improving systems (including DGM). DGM-H also improves the process by which new agents are generated (e.g. persistent memory, performance tracking), and these meta-level improvements transfer across domains and accumulate across runs. This work was done during my internship at Meta (@AIatMeta), in collaboration with Bingchen Zhao (@BingchenZhao), Wannan Yang (@winnieyangwn), Jakob Foerster (@j_foerst), Jeff Clune (@jeffclune), Minqi Jiang (@MinqiJiang), Sam Devlin (@smdvln), and Tatiana Shavrina (@rybolos).

Most people in machine learning still misunderstand probabilities. A model can be perfectly calibrated and still be completely useless. This was proven more than 40 years ago by DeGroot & Fienberg (1983). Yet many ML papers still miss this point. Here is the idea.

Happy birthday to Joseph Fourier!! That’s Mathematics from 1822 allows us to .....just add to the looooong list....