fritzo

15 ms·

241.
by fritzo 3y ago
Laws to mandate accessibility, laws to mandate provable correctness, laws to mandate fairness, and laws to break up the big, for only the big could satisfy all the other laws.
242.
by fritzo 3y ago
Just curious, who came up with Stigler's law?
243.
by fritzo 3y ago
I feel like this is an engineering cultural issue. Few organizations can produce safety-critical code, say organizations who focus on tail events and long term profitability. And an engineering organization cannot transition from a growth o
244.
by fritzo 3y ago
lack of flow, as if it were a concatenated collection of notes that were not worked together in a final pass
245.
by fritzo 3y ago
Yeah seems like a category error: desalination is a mode of production, tap is a mode of delivery
246.
by fritzo 3y ago
Multi-modal models will be exciting only when each modality supports both analysis and synthesis. What makes LLMs exciting is feedback and recursion and conditional sampling: natural language is a cartesian closed category. Text + Vision mo
247.
by fritzo 3y ago
The demonstration isn't classifying based on compressed data, rather it's classifying on the compressibility of data. The idea is that "7 7" should be more compressible than "7 3", and similarly raster images
248.
by fritzo 3y ago
More recently Tim Thompson has been creating interactive art installations. I got to work with him at some installation shows in the bay area https://timthompson.com
249.
by fritzo 3y ago
Paywalled paper: https://www.nature.com/articles/s41550-019-0878-9 Preprint: https://arxiv.org/abs/1909.05218 Figure 2 pp. 14 of the preprint shows much more plausible error bounds on the curve fi
250.
by fritzo 3y ago
Right, show a posterior distribution not some hallucinated point estimate.
251.
by fritzo 3y ago
Neat! reminds me of my old web app https://fritzo.org/keys
252.
by fritzo 3y ago
I guess it's a good idea to check your product name in https://www.urbandictionary.com early
253.
by fritzo 3y ago
> since our work touched many parts of the codebase and demanded collaboration with lots of different devs, we now have a strong distributed knowledge base about the existing system Great to see this cultural side-effect called out.
254.
by fritzo 3y ago
I've been following this flow where I write a few successive versions and perform refactoring steps to move one version "towards" another version, as if there is a conceptual force of attraction. It's often unclear which
255.
by fritzo 3y ago
Yet I find when rewriting from scratch there's a countering force of simplification where I find myself thinking "We spent way too much complexity on that part, and we can entirely do away with that other part, and we turned out n
256.
by fritzo 3y ago
What's your business model? Contract development of features? Paid support? Paid-only features?
257.
by fritzo 3y ago
Plug for the Funsor library, written by Eli Bingham and me for use in the Pyro and NumPyro probabilistic programming languages. We tried to take the "functions are tensors" perspective and make a numpy-like library for functions,
258.
by fritzo 3y ago
That D in DVCS is much appreciated when my power goes off, my internet goes down, etc.
259.
by fritzo 3y ago
I'd go farther and say the real central limit theorem is the generalized central limit theorem, and the finite variance central limit theorem is just a quirky special case.
260.
by fritzo 3y ago
> “I do think there was a time when perhaps legacies needed a boost” -Dean Emeritus of Admission Mike Steidel Is there any way of charitably interpreting Mike Steidel's words? I have a tough time reading anything here but classist
261.
by fritzo 3y ago
Sorry for the confusion, LLMs are great. But it seems like there's a surplus of platforms and dearth of applications, like too many people (myself included) have taken advice to "sell shovels in a goldrush", and Vercel's
262.
by fritzo 3y ago
AI services are the new tupperware, and this is the party.
263.
by fritzo 3y ago
Do you mean Python's lack of a labeled union type?
264.
by fritzo 3y ago
Great summary. Are you superhuman, or did you use an AI tool? What tool?
265.
by fritzo 3y ago
Ah what a clarifying perspective: lumpers vs splitters /s
266.
by fritzo 3y ago
Once back in 2012
267.
by fritzo 3y ago
How do hunters haul a deer over that little ladder?
268.
by fritzo 3y ago
Figure 9 of the original paper shows some side-by-side comparisons of kite representations vs satellite imagery https://journals.plos.org/plosone/article?id=10.1371/journal...
269.
by fritzo 3y ago
Even my own reading speed even varies by a factor of 5 day to day, depending on how much reading I've been doing, sleep I've gotten, etc.
270.
by fritzo 3y ago
Full video of testimony on CSPAN https://www.c-span.org/video/?528117-1/openai-ceo-testifies-...

More ›