Localized reading
Penso sempre più che il software del futuro unirà il codice per il controllo del flusso a piccoli programmi neurali per i giudizi «sfumati». È ciò che sto esplorando con ProgramAsWeights, che risponde alla domanda da dove provengano questi programmi neurali: vengono «compilati» da descrizioni in inglese. Per esempio, ho combinato 30 programmi neurali con un albero decisionale per costruire un assistente per un sito didattico: sembra un chatbot, ma gira localmente. Ogni piccolo programma gestisce domande come «a quale specialista va inoltrata questa richiesta?», mentre il codice controlla l’intero flusso. Qui si può provare...
Original post / EN
I increasingly think future software will combine code for control flow with small neural programs for "fuzzy" judgments. That's what I've been exploring with ProgramAsWeights, which answers the question of where those neural programs come from: they are "compiled" from English descriptions. For example, I combined 30 neural programs with a decision tree to build a course website helper that looks like a chatbot but runs locally. Each small program handles a question like "Which specialist answerer should this be routed to", and code controls the overall flow. You can try building with it here: https://t.co/Bwgcn362bc
- views
- 13,5K
- likes
- 171
- saves
- 140
- reposts
- 21
Links and projects
Quoted post
Diogo Almeida @CompleteSkeptic
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI? I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev • 20-200x faster • 40-400x https://t.co/JSybNG2BKJ
Vedi originale ↗Fonte
This independent archive preserves a public post with attribution. Text, media, account details and trademarks belong to their respective owners.
