I'm really happy someone said it.
Uncategorized
3
Posts
3
Posters
5
Views
-
RE: https://mastodon.bsd.cafe/@stefano/115547315643919873
I'm really happy someone said it. I have felt really excluded in this environment and I could never really put my finger on it. A decade ago, I contributed a number of patches for an open source database project under my real name. It was one of the only enjoyable experiences I ever had with open source, because the database was a niche project, designed for a very particular kind of dataset, and the author immediately thanked me for the patches. He was clearly overworked and grateful. This is a feeling I now work hard to reproduce in my little cloister, where we'll keep working until we have something ready to explain itself. It feels like a shame that it has to be that way, because the second you give something a face, you're expected to compete.
Monoculture is a death spiral. I want to create things that reflects specific virtues and use cases I care about. I take a really dim view of people telling me to do something marketable. I want it to satisfy specific criteria. It's driven by me, not a market.
-
undefined stefano@mastodon.bsd.cafe shared this topic on
-
RE: https://mastodon.bsd.cafe/@stefano/115547315643919873
I'm really happy someone said it. I have felt really excluded in this environment and I could never really put my finger on it. A decade ago, I contributed a number of patches for an open source database project under my real name. It was one of the only enjoyable experiences I ever had with open source, because the database was a niche project, designed for a very particular kind of dataset, and the author immediately thanked me for the patches. He was clearly overworked and grateful. This is a feeling I now work hard to reproduce in my little cloister, where we'll keep working until we have something ready to explain itself. It feels like a shame that it has to be that way, because the second you give something a face, you're expected to compete.
Monoculture is a death spiral. I want to create things that reflects specific virtues and use cases I care about. I take a really dim view of people telling me to do something marketable. I want it to satisfy specific criteria. It's driven by me, not a market.
@somebody back then when I joined a little project in the mid nineties we coded mainly on Linux (RH and so on and me on Slackware, iirc)... But portability was always important. (parisc, SPARC, iirc)
Among the people for implementing support for other Platforms and OS someone from FreeBSDland helped to add support and I was happy/surprised to find out when I started with BSD it's still maintained and updated in FreeBSD's ports and pkgs by some gold souls (kudos for that, mates!)
And another "Beastie boy" massively helped to test to improve things. I think that's key. Working together without building borders or artificial boundaries across platforms.
btw good blog post @stefano Thanks!
-
@somebody back then when I joined a little project in the mid nineties we coded mainly on Linux (RH and so on and me on Slackware, iirc)... But portability was always important. (parisc, SPARC, iirc)
Among the people for implementing support for other Platforms and OS someone from FreeBSDland helped to add support and I was happy/surprised to find out when I started with BSD it's still maintained and updated in FreeBSD's ports and pkgs by some gold souls (kudos for that, mates!)
And another "Beastie boy" massively helped to test to improve things. I think that's key. Working together without building borders or artificial boundaries across platforms.
btw good blog post @stefano Thanks!
@Tionisla thank you!
Gli ultimi otto messaggi ricevuti dalla Federazione
-
@stefano grazie gentilissimo
-
Come Pensa la Macchina: un LLM smontato pezzo per pezzo. Tokenizzazione, embeddings, attention, hallucinations. Ollama in locale, zero fuffa
Smonto la macchina pezzo per pezzo. Ho #Ollama sul Mac con una decina di modelli. Scelgo il più piccolo: llama3.1:8b, 8 miliardi di parametri, 4.9 gigabyte su disco. Il più facile da maneggiare senza sbatti, e tanto l'architettura è identica per tutti: che siano 8 miliardi o 405 miliardi, il meccanismo è lo stesso. Cambiano le dimensioni delle matrici, non come funziona la macchina. Lo apro dal terminale, guardo i byte, e seguo il percorso completo: dal testo che entra al testo che esce. Ogni passaggio, ogni formula, ogni decisione matematica. Niente metafore del cervello. Niente fuffa. Se vuoi capire come funziona una cosa, la smonti. Non leggi chi ne scrive.
pinperepette.github.io/signal.…
Per conoscere altre notizie sulla #IntelligenzaArtificiale, segui il gruppo @aitech
-
@n1xnx @keith_lawson @GossiTheDog @quixoticgeek Similarly, clanker stans don't seem to realize that when they're asking their spicy autocomplete pal for advice they're communing with every shitpost and ironic negation ever posted on reddit and twitter. "No, you CAN shave more efficiently by setting your beard on fire!"
-
@joeyh @GossiTheDog found it, it's under the achievment badges for whatever reason
-
Viaggiatori,
#diariosiviglia e #diarioestremadura continuano! Non perché sono ancora in viaggio, ma perché tra qualche giorno inizierà a pubblicare foto sull'account di @Perry77 , e video su @viaggiatore_tv
Non vi resta che seguire i 2 tag o i profili per continuare a seguire questa strana avventura.
Prima però, devo affrontare il ricordo più intenso che mi sono riportato a casa da questo viaggio: una bella influenza...
-
@joeyh @GossiTheDog where is the block button?
-
@votolibeguale ma chevve frega daa legge elettorale che tanto poi nun vota nessuno
Come se vivi a Bolzano e te inpicci daa pesca de i tonni
@politica @informapirata @fucinafibonacci @sabrinaweb71 @MajDen @carneade @martellum @Otttoz @Alfonso1 @pare @emama @maza
