“jacobinnl” — Fediverse search results on home.social

C’è una generazione che è nata sotto le macerie della modernità: non si sa bene cosa sia, non si capisce quali maestri abbia, ma é chiaro cosa vuole: tutto.

Tutto quello che le è stato rubato, tutto quello che le è stato promesso ma non si è mai realizzato: una vita bella, fuori dalla precarietà, fuori dall'emergenza, fuori dalla catastrofe. Ma ora che ne siamo ancora dentro, sentiamo il bisogno impellente di organizzarci dal qui e ora per abitare un mondo diverso.

Dalle mobilitazioni per il clima alla solidarietà con la resistenza palestinese; dalle periferie alle zone sacrificate, c’è una generazione che dalla necessità ha imparato a lottare, a ricomporsi, a rialzare la testa.

Il futuro è nostro e ce lo riprendiamo.

Ci vediamo il 28 ottobre per parlare insieme di tutto questo.

https://roma.convoca.la/event/la-genz-vuole-tutto-1

#ecologismo #movimentisociali #genz #lottesociali #roma #iniziativa

Young Beard @[email protected]

2026-10-25

Books finished in 2026:

44. Second Foundation by Isaac Asimov (1953)
43. Men At Arms by Terry Pratchett (1993)
42. Full Circle: A Life in Rebellion by Ben Morea (2025)
41. Foundation and Empire by Isaac Asimov (1952)
40. Mother Night by Kurt Vonnegut (1962)
39. Wobblies! A Graphic History of the Industrial Workers of the World (2005) Edited by Paul Buhle and Nicole Schulman with various authors
38. Foundation by Isaac Asimov (1951)
37. Clay’s Ark by Octavia E Butler (1984)
36. We Fight: Three Decades of Rebellion Against The Police 2nd Ed. by various authors (2020; 2025)
35. Rabbit, Run by John Updike (1960)
34. Mind Of My Mind by Octavia E Butler (1977)
33. The Black Jacobins: Toussaint L’Ouverture and The San Domingo Revolution by CLR James (1938)
32. Lords and Ladies by Terry Pratchett (1992)
31. Escape From Freedom by Erich Fromm (1941)
30. Walking With The Comrades by Arundhati Roy (2011)
29. Wild Seed by Octavia E Butler (1980)
28. Ready Player Two by Ernest Cline (2020)
27. The Last Hero: A Discworld Fable by Terry Pratchett (2001)
26. Revolutionary Suicide by Huey P Newton with J Herman Blake (1973)
25. The Body Snatchers by Jack Finney (1955)
24. Granny Made Me An Anarchist by Stuart Christie (2004)
23. The Day of The Triffids by John Wyndham (1951)
22. Soledad Brother: The Prison Letters of George Jackson (1970)
21. Secolo Nuovo: Or the Times of Promise, A Novel by Fulvia Ferrari (2021)
20. Socialism: Utopian and Scientific by Friedrich Engels (1880)
19. Working Class History: Everyday Acts of Resistance and Rebellion by Working Class History (2020)
18. The Diary of a Drug Fiend by Aleister Crowley (1922)
17. Small Gods by Terry Pratchett (1992)
16. The Red and The Black by Stendhal (1830)
15. The Book of Sand (1975) and Shakespeare’s Memory (1983) by Jorge Luis Borges
14. Anarchist Feminists In The Philippines From Etniko Bandido (~2016)
13. The Unseen by Nanni Balestrini (1989)
12. Handbook of Revolutionary Warfare: A Guide to the Armed Phase of the African Revolution by Kwame Nkrumah (1968)
11. Rights of Man by Thomas Paine (1792)
10. Ready Player One by Ernest Cline (2011)
9. Lair of the White Worm by Bram Stoker (1911)
8. Blue Mars by Kim Stanley Robinson (1996)
7. Witches Abroad by Terry Pratchett (1991)
6. The Nation On No Map: Black Anarchism and Abolition by William C Anderson (2023)
5. baedan - journal of queer nihilism - issue one (2012)
4. Snow Crash by Neal Stephenson (1992)
3. They Will Beat The Memory Out Of Us: Forcing Nonviolence on Forgetful Movements by Peter Gelderloos (2024)
2. It Can’t Happen Here by Sinclair Lewis (1935)
1. Rainbows End by Vernor Vinge (2006)

Finished the original trilogy of the Foundation series. To continue or not to continue? This one wrapped up the story of The Mule and showed us the conclusion of the first four centuries after setting up the Foundation in preparation for a thousand years between the fall of the Galactic Empire and the rise of a Second Empire, if all goes according to the Seldon Plan. It explained who and what the Second Foundation was and how they operate.

#scifi #fiction #books #bookstodon #asimov #foundation #reading #book #nowreading

#scifi #fiction #books #bookstodon #asimov #foundation

Xamanismo Coletivo @eliasulrich

2026-10-25

"This “ #Brazilianization ” (or, in more general terms, #peripheralization ) thesis states, in a nutshell, that “the world’s horizon of expectation is no longer informed by the center, but by the peripheries.” That is, both by the peripherical areas that proliferate in the metropolises of a “planet of slums” and by the former colonies that have from birth been “perfect laboratories for new forms of capitalist exploitation integrating authoritarian control of surplus populations, the management of unresolvable social conflicts and a new police state compatible with parliamentary democracy.” ( #Gaza being, of course, the most extreme form yet experienced of such laboratories.)"

https://jacobin.com/2026/10/brazil-far-right-social-disintegration-exploitation

#brazilianization #peripheralization #gaza

Dev Community @[email protected]

2026-10-24

Choosing one option out of many should not cost a paragraph of generation, yet that is how most agents route today. A language model decodes a structured answer one token at a time, and on the project's own benchmark a 50-question batch takes about 11,500 ms.
ELMDE, the Equilibrium Latent-MoE Decision Engine, takes a different route. It reads the context once, lets a small recurrent core settle on a fixed point, and reads the decision directly off that point in a single pass. The same batch takes about 14 ms.
Why token-by-token decisions are the wrong tool
Constrained decoding forces an answer into a valid format, but it still pays for every token and it distorts the probabilities along the way. Splitting a choice into subword tokens makes early tokens that several options share carry more weight than they should, a bias the project calls myopic prefix-sharing distortion.
The fixed-depth classifiers that came before have their own limits. Convai Laya re-encodes the state for each question, so 50 questions mean 50 encodes, and it hits a 192-token header limit when there are many options. On Banking77, which has 77 intents, the README reports Laya falling to 0.425 accuracy for that reason.
The idea: encode once, settle, project
ELMDE sits between a fast classifier and a slow reasoning model, which the project calls System 1.5. It does three things, in order.

Encode once. The context is compressed into a small latent cache and reused for every option.
Settle. A recurrent core refines a handful of working-memory slots until they stop changing, at a fixed point called z*.
Project. The decision heads read z* directly, so the answer comes from one pass instead of a generated sequence. A decision this fast is only useful if it knows when it is unsure, so the last stage is a gate that escalates hard cases to a slower model or a person. How it works Four pieces do the work. • Latent cache. Multi-Head Latent Attention stores each context as a 128-dimension vector plus a 32-dimension rotary key. Because the up-projection matrices are absorbed into the query and output matrices, full keys and values are never built. The repo's check shows a maximum deviation of 5.68e-14 from the standard computation, and a KV cache about 93% smaller in the table, 98.5% in the theorem. • Equilibrium core. A mixture-of-experts layer is applied in a damped loop, z(r+1) = (1 - a) z(r) + a Phi(z(r)), for at most six ticks. It stops early once the change falls below 1e-3. With damping 0.45 and a measured Jacobian bound of 0.1358, the contraction factor is at most 0.6111, so the loop is guaranteed to converge to a single fixed point. • Three heads. Choice scores every option in parallel, with no cap on how many. CORN ordinal keeps ranked scores monotonic by construction. A Beta head returns a probability and a closed-form variance. • Conformal gate. Selective conformal risk control sets a confidence threshold, tau* = 0.9848, calibrated on held-out data. Decisions above it are accepted with a certified error of at most 1%. Everything else is escalated. Results, and what they do not show These are the figures reported in the repository, produced by its own benchmark scripts. Measure ELMDE Compared with 50-question batch latency ~14 ms 771 ms Laya, ~490 ms Jev (cloud API), ~11,500 ms autoregressive CDM Banking77 accuracy (77 intents) 0.912 0.870 Jev, 0.425 Laya Calibration error (ECE) 0.0434 0.144 Jev, 0.081 to 0.213 Laya Conformal risk gate 99.45% certified Error of at most 1% on accepted decisions Three limits matter when reading that table. • The speedup depends on the baseline. The 55x figure is against Laya (771 ms to 14 ms). Against the autoregressive decoder it is closer to 820x. • The benchmarks are the project's own. Nobody outside the project has reproduced them yet, and Jev is a hosted API whose setup I cannot control. Treat the comparison as a claim to test. • The certificate has a condition. Conformal guarantees hold when live data resembles the calibration data. If the distribution shifts, the 1% bound can stop holding, which is why the repo also includes an online adaptive variant. Try it, and tell me where it breaks The repo runs on Python 3.10+ and PyTorch 2.2+, and python elmde_engine.py runs the mathematical checks first. After that, four drop-in patterns in example_injection.py show it in use: • fast-pathing agent tool calls in under 35 ms with zero generated tokens • routing across 77 or more intents • monotonic triage tiers for clinical or legal queues • a decorator that escalates risky calls before they execute I am most interested in three kinds of feedback: someone reproducing the benchmarks on their own hardware, a dataset where the gate fails, and any flaw in the contraction argument. Open an issue on the repository or send a pull request. The code is Apache 2.0.

#ai #programming #opensource #llm #software #coding #development #engineering #inclusive #community
Building ELMDE: Deciding in One Pass

#opensource #ai #community #programming #software #coding

Search