Anthony Di Franco

pataphysicist, leviathan scatologist, global weirding scholar/activist/archivist, erudition fan.
rt = perverse fascination

[Michel Bauwens @mbauwens]

Venkatesh Rao's closing analysis of the Khlongs and Subaks workshop held in Bangkok.

@vgr writes:

"A new cosmopolitanism and cosmolocalism brewed in SEA rather than Davos

Techno-animism in the "SBA" mode as a fresh and unexpected vision of the digital future for the world
"Protocol ethnography" of systems like subaks, tiger preserves, khlong systems, and the other systems we talked about

Technology and meaning-making

The dark protocol magick arts of making video with AI
While SEA is of course an amazing laboratory for experimenting with technology futures, novel models of governance, and real-world applications, I'm personally convinced the region can also give the world fresh new theories, abstract models, and entirely new technologies. ...

good ideas are more likely to come from the fragmented network-based regions of the world, rather than the big monolithic "civilizational state" cores. I hope you guys will help prove me right."

[Pablo Antonio @PabloAntonio]

Aaron Swartz would have turned 38 today, Nov 8th.

His marble statue unveiling is set for Feb 7th at the @internetarchive.

RSVP: https://t.co/SaI7yj3kY3

Grateful to all who helped honor a hero of Internet Values - open access, free speech & privacy. 🧵

[Jonathan Gorard @getjonwithit]

New paper alert!

Numerical general relativity is hard, complicated and computationally expensive. Here, we develop a radically new approach to doing it: cheat, and do special relativity instead...

arXiv link and explanation in thread... (1/11)

[Justine Tunney @jartine]

You can now use 1.58 bit LLMs with llamafile. I just uploaded TriLM. Its weights are -1 / 0 / +1 just like the Soviet Setun. It comes in 8 different sizes (78MB to 1.3GB) and has good benchmark scores on four CPUs. https://t.co/XUNhfHnZO2

[Sophia Sanborn @naturecomputes]

Beyond Euclid: An Illustrated Guide to Modern Machine Learning with Geometric, Topological, and Algebraic Structures

New review out with @ninamiolane, co-led by myself, @johmathe & @mathildepapillo

Check out our visual taxonomies of non-Euclidean structure in data & ML models

[Justine Tunney @jartine]

I've just uploaded llamafiles for Google's new Gemma2 language model. This is the 27B model that folks have been saying is better than 70B and 104B models like LLaMA3 and Command-R+. People are even saying it's better than the original GPT4! Now you can run it air-gapped on your own computer, using a single file.

Your ./gemma-2-27b-it.Q6_K.llamafile is 22gb and we put a lot of work into making sure it produces outputs consistent with Google's hosted version. It's great for coding, roleplaying, admiring the quality of Google engineering, and more.

The tradeoff is that license is a bit cray. It's not open source. Like the Cohere license, Gemma's license is about as bad as they come. Its list of unacceptable uses is unacceptably broad. The terms of the license can be changed by Google at any time. They also refer to this as a Gemma Service (which they don't define) and say they reserve the right to "remotely" disable it if you violate the agreement.

It's totally a good look for the one company on Earth that has a backdoor into literally everything I own to remind me they might execute Order 66 with Chrome and hack into my computer to remotely disable the floating point array file I just downloaded.

When Gemma v1 was released back in February, we saw how certain Googlers responsible for RLHF training brought great shame and embarrassment to everyone in AI who cares about safety. Perhaps Google's legal team deserves a closer look too.

In any case, the pros clearly outweigh the cons. I think Gemma2 is worth trying! You can download the 27B llamafiles here: https://t.co/S5Abayf9gJ I've also posted the tinier 9B llamafiles here: https://t.co/vDaSxyWigw Enjoy!

[Sophia Sanborn @naturecomputes]

In this new paper, led by @giovannimarchet, we present a unique theoretical result that provides guarantees for the concrete representational structure expected to emerge in a learning system

[St. Rev. Dr. Rev ⏭️☯️🏴😻 @St_Rev]

A whole bunch of Leading Scientists with Professional Headshots on Twitter Dot Com are extremely buttmad about this quote. Genius is a dime a dozen, they are saying. Science is about project management and filling out form!

Well, Science is about that now, anyway.

[Danielle Fong 🔆 @DanielleFong]

saddest face in the universe

[Danielle Fong 🔆 @DanielleFong]

taxes are dissipative
from a systems theory standpoint in the way that intellectual or cultural activity or, say, energy at a dance party, isn't

it's centralising.

UBI is a way to support the decentralized genius of a country, to support the stuff that falls through the cracks

[John Robb @johnrobb]

The permacrisis decision making loop

[Jonathan Edwards @jonathoda]

We could have built bicycles for the mind but we built aircraft carriers instead, because we are fascinated by performance and complexity, and our culture embraces self-indulgence.

[Justine Tunney @jartine]

Meta LLaMA3 70B going 65 tokens per second on AMD Ryzen Threadripper PRO 7995WX using llamafile v0.7.1

[Anthony Di Franco @dfko_0]

@pearlymuklg My guess is probably yes, because the “place” in “place cell” probably is place in sensory-motor sequence, of which place in space is a special case.

[madison @dearmadisonblue]

transformers probably can't do recursion (or anything with trees) properly without hyperbolic latent spaces

[R.Сам 🦋🐏 @Logo_Daedalus]

I think of every “serious philosopher” propounding their theories to a baby, & the baby stares intently, as it slowly raises its fingers toward the philosopher’s face, jams fingers up their nose, & then into mouth.

[Ben Reinhardt @Ben_Reinhardt]

Startups are very good at many valuable things! But research is not one of those things.

[Yann LeCun @ylecun]

An article about my vociferous support of open source AI platforms.
Demis Hassabis, Dario Amodei, and Sam Altman (among others) have scared governments about what they claim are risks of AI-fueled catastrophes.

I know that Demis, at least, is sincere in his claims, but I think those claims are incorrect and counterproductive.

The inevitable *effect*, intentional or not, if governments believe those claims would be a regulatory capture profiting their companies.

I do think that the short-term societal dangers of proprietary AI systems that will soon mediate everyone's digital diet are considerably higher than any imagined catastrophe caused by the misuse of open source AI systems.

And I believe that the benefits of open source AI platforms in terms of progress, safety, economic development, and cultural diversity are overwhelming.

https://t.co/UQImxIaTEj

[hardmaru @hardmaru]

Artificial lifeforms are super fascinating to watch.

These self-organizing, self-replicating, “lifeforms” emerged from a continuous time cellular automata system called Flow-Lenia.

Lenia is a family of CAs generalizing Conway’s Game of Life to continuous space, time and states.