Probably the first fully analytic #NURBS based path tracer:
no triangles
no subdivisions or bisection
no micro-meshes
no hardware tessellation
no hardware ray tracing
no hardware rasterizer either
millions instances animated in real time
Anthropic themselves found that vibecoding hinders SWEs ability to read, write, debug, and understand code.
not only that, but AI generated code doesn’t result in a statistically significant increase in speed
don’t let your managers scare you into increased productivity. show them this paper straight from Anthropic.
Pi is maximally complex by one measure. Trivially simple by another. That gap explains what AI can and cannot do.
New post - previewing today's conversation with @martin_casado for @a16z, out next week.
https://t.co/tabFRNO01b
@tmikov The AI is a probabilistic tool. Engineering is about precision. There is no room for errors. Vibe coding increases the output but the quality is low. You need countless iterations and in depth analysis to get it to the point where it is usable
@hamptonism The AI is just a probabilistic tool which helps you to iterate faster. Engineering is about precision work, no room for errors. So far the AI revolution created more startups and unlocked very old projects which were too expensive to rewrite. This means more work for engineers
@satya164 What is this nonsense? 7 convoluted ways to do something very simple.. The only thing you need is to make all navigation actions async by default and add a global async redirect interceptor to allow you to switch to a different route based on business logic
That's right, we released our first iteration of JEPAs for LLMs: https://t.co/aNMNC9EgFI
And yes, the code is public! Learning by latent space prediction has revolutionized vision models, and it will revolutionize LLMs!
🌜One small step for JEPAs, one giant leap for LLMs🌛
@TivadarDanka The real power and efficiency in ML comes from the Johnson–Lindenstrauss Lemma which allows high to lower dimensional reduction without loosing accuracy
@tmikov The main reason why is unknown is because you stopped publishing new releases via GitHub /NPM. The last version for hermes-engine on npm is 0.11 from three years ago. People are taking this seriously
Advent of Numba: I'm using @numba_jit + CUDA to solve Advent of Code (https://t.co/peypadSOhe) this year. Trying to showcase something interesting about Numba CUDA in each solution. Solutions on Github: https://t.co/mC8fUyj4Pw