Natural Language Processing in Finance: Shakespeare Without the Monkeys man.com/maninstitute/s…

The human capacity to abstract discrete categories over continuous phenomenon constitutes perhaps our greatest blessing and curse. It at once blesses us with the combinatoric potential we call creativity, and curses us to forever renegotiate the boundaries of the discrete.

"Specific knowledge is found by pursuing your genuine curiosity and passion rather than whatever is hot right now." - Naval

"Unnecessary meetings (and most are) are a mutually-assured-destruction of time. Learning how to avoid them is a prerequisite of doing anything great." - Naval

"You get paid for being right first, and to be first, you can’t wait for consensus." - Naval

Supply Chain 4.0 – the next-generation digital supply chain mck.co/2pqVpKk 来自 McKinsey & Company

Check Greenhouse Gas Emissions before Booking Freight e2open.com/check-greenhou…

Subscribe to Technically, by Technically technically.substack.com/?r=1mn9oy&utm_…

Please ignore the deluge of complete nonsense about Q*. One of the main challenges to improve LLM reliability is to replace Auto-Regressive token prediction with planning. Pretty much every top lab (FAIR, DeepMind, OpenAI etc) is working on that and some have already published

Yann LeCun SiBarrick Is the challenge really “planning within a learned world model”, or the process of learning the world model itself? In transitional Model Predictive Control, we use learned models (neural and Bayesian approaches) and the challenge is always learning a *causal* model such that

If you care too much about what others think, you cannot remain happy. Don't be afraid. Do what you get the most pleasure from. Develop your talents wherever they may lead. Fall in love with some activity and do it!

Lots of confusion about what a world model is. Here is my definition: Given: - an observation x(t) - a previous estimate of the state of the world s(t) - an action proposal a(t) - a latent variable proposal z(t) A world model computes: - representation: h(t) = Enc(x(t)) -

Like any great idea it was discovered in a few places. But these other discoveries were less general. For example, one was in statistics: Given (x , y) with densities p(x), q(y) find a function y = f(x) that maximizes 𝔼(xy) The solution is f =∇ϕ for some convex ϕ 4/5

Brenier's Thm has had deep impact in Statistics and Machine Learning It has been used, for instance, in diffusion models particularly in potential and probability flows. 5/5

Bayesian methods broke the Enigma codes, the first breakthrough of speech recognition, and Stein estimator is the foundation of ridge / lasso regression

I believe multi-agent game theory is heavily used in its LLM training.

Accounting and finance folks: 1. Who is the best Twitter follow re: automating bookkeeping/finance/accounting/legal with AI? 2. What are the top 5 best tools/software right now?

Introduction youtu.be/EaYrAZkWVbs?si… via YouTube

17/ IMO, the fact information on the Chinese internet is scattered across myriad walled gardens without a reliable means of search seems to be… pretty clearly bad for consumers? This feels like a key area where China tech has actually regressed rather than leapfrogged the US

I'm excited to be using the Fellou browser to publish this post! This next-gen AI browser makes browsing the web smarter and more efficient. #AI #AIBrowser #FellouAI #ProductivityTools