ang42.bsky.social
Interested in applied machine learning.
Volunteer @neuromatch.bsky.social & Climatematch
About science, more or less.
"È tutto sbagliato, è tutto da rifare" (Bartali)
(... someday I may start to write something on substack.com/@drawanowl)
118 posts
309 followers
1,859 following
Prolific Poster
Conversation Starter
comment in response to
post
Del resto perché consentire alle pericolosissime ngo di promuovere regole più stringenti su ambiente e sanità?
www.politico.eu/article/eu-p...
comment in response to
post
Amazing! I literally just left a careers session at HPS in Cambridge where I explained to a group of PhD students that their ability to process and reflect on large nebulous quantities of information at speed was a skill so desirable the tech-bros are trying to mimic it with LLMs and AI.
comment in response to
post
ChatGPT to write papers and ChatGPT to read papers. What can go wrong?
comment in response to
post
... I wonder if there's the same trend across Europe, and if this could result from a general perception that a university degree doesn't always mean a better income, despite years of training and personal investments.
comment in response to
post
Pragmatism, what else?
comment in response to
post
And some people like to drink it mixed with alcohol, grappa, anise, or sambuca
comment in response to
post
I usually drink straight charcoal, no sugar 😂
comment in response to
post
È incredibile come gli Usa possano cambiare posizione da un giorno all'altro. In passato la loro coerenza, soprattutto in politica estera, era una certezza (nel bene e nel male).
comment in response to
post
The end game would be "Pineapple on pizza is a government plot". This could be the end of the Italian Republic as we know it.
comment in response to
post
... come un singolo articolo o PDF, cercando di ottenere risposte che includano i riferimenti al testo.
Quando chatgpt non sa qualcosa semplicemente la inventa di sana pianta.
comment in response to
post
A me sembra che fare domande generiche ai bot non porti mai a nulla di buono, sono pappagalli stocastici, inventano sempre qualcosa di nuovo. Da un annetto a questa parte probabilmente la cosa migliore da fare è usare qualche forma di RAG e far "ragionare" il bot solo su contenuti limitati ...
comment in response to
post
Did they go all in with AI?
comment in response to
post
i've already read something similar last week
comment in response to
post
Hello! Is there any apk available? Can you upload it on F-Droid?
comment in response to
post