Alessandro (@Skyhunter___) header
Alessandro (@Skyhunter___) profile picture

I do stuff, code stuff and talk too much.

Switzerland
Joined August 2010
Photo by Alessandro (@Skyhunter___)Photo by Alessandro (@Skyhunter___)Photo by Alessandro (@Skyhunter___)Photo by Alessandro (@Skyhunter___)Photo by Alessandro (@Skyhunter___)Photo by Alessandro (@Skyhunter___)Photo by Alessandro (@Skyhunter___)Photo by Alessandro (@Skyhunter___)Photo by Alessandro (@Skyhunter___)Photo by Alessandro (@Skyhunter___)
Gemini 2.5 Pro is taking off 🚀🚀🚀 The team is sprinting, TPUs are running hot, and we want to get our most intelligent model into more people’s hands asap. Which is why we decided to roll out Gemini 2.5 Pro (experimental) to all Gemini users, beginning today. Try it at no cost at gemini.google.com
📣 Today, we’re introducing Gemini 2.5, our most intelligent AI model. An experimental version of Gemini 2.5 Pro is available now in the Gemini app for Gemini Advanced users: gemini.google.com/app Let’s get into this update ⬇️🧵
223
614
164
4,963
1,740,717
When you hear about "human-level AI" being just "one LLM training run away", it's as if people building 3D printers (the kind that makes small, flimsy plastic objects) were telling you that they've nearly built an infinite matter replicator, capable of scanning and cloning any object -- be it steel, gold, or diamond, be it minuscule or building-scale. "We're just one iteration away! Just need to increase spatial resolution a bit!" As long as you have absolutely no idea what "matter" is and how 3D printers work, it might sound sort of plausible. But if you do, it's absolutely clown shoes. The way to navigate such claims is to have a proper, formal definition of intelligence.
58
257
30
1,533
214,904
A common misconception about overfitting is that you can only overfit to a dataset if you are directly training on it. In reality, you can overfit to anything, as long as you have a feedback loop (however small) between results on the target dataset and your model. For instance, if you're basing your model design choices on metrics computed on a validation set, you will gradually overfit to that validation set.
24
87
14
843
106,254
Alessandro retweeted
LLMs’ sensitivity to design decisions, such as template choice, can degrade performance & prevent their robust application. Introducing Batch Calibration, a simple method that mitigates this effect & improves on existing methods w/ negligible add’l cost → goo.gle/45wlYj0
22
115
9
597
158,250
Alessandro retweeted
AWS is ready to power AI agents that can handle busywork instead of just chatting trib.al/ZYtxzwV
2
10
38
45,635
Alessandro retweeted
Had a wonderful #CHI2023 experience with my colleagues from the PET Lab @AnnikaAebli @herebejames @Skyhunter___ @lahariG @poojaraosb @_S_Aliyu @martigan and Diana Korka
1
3
12
1,215
Wrapping up #CHI2023 with the best team ever! ✨🤖 #UNIL #PETLabbers @poojaraosb , @Skyhunter___ , @AnnikaAebli , @_S_Aliyu , @herebejames , Diana Korka , @kavous_nik and @martigan (in essence)
2
5
298
Alessandro retweeted
1/6 Thrilled to announce that our paper: “Changes in Research Ethics, Openness, and Transparency in Empirical Studies between CHI 2017 and CHI 2022” has won the best paper award at ACM #CHI2023 !
2
5
3
17
2,525
😶uhh
13
13
3
394
The recent invasion on Ukraine, our friends and neighbours, left us shocked and outraged. In solidarity with all victims of this act of aggression, the CD PROJEKT Group has decided to support humanitarian aid efforts by donating 1 million PLN to the @PAH_org (1/2)
421
3,336
338
35,432
Never have I ever messed up with ACM documentclass parameters to create new conditions in a LaTeX document with 4 days to the submission date 😬
So I think during all these years I underestimated my computer. Adobe Premiere uses 120 Go / 16 Go of my RAM. (Now I know it relies on the memory of the SSD, but please, optimization seems even worse than FIFA 22)
Je savais pas qu'Aliexpress vendait aussi des patinoires toutes équipées 😂😂 @officialGSHC vous avez pris l'option chrono et glace en carton avec ?
Trying to build an AI driven RC car . Still pretty early in the project ! #arduino #mechanicalengineering #3dprint #makersgonnamake #MachineLearning
58
679
63
3,374
Alessandro retweeted
tail -f /var/log/syslog
40
1,603
152
5,287
18
650
58
5,555
Alessandro retweeted
Drive To Survive Season 4.... CONFIRMED Coming to your screens in early 2022 🍿 #F1
860
8,175
3,587
73,449
«Kreislaufwirtschaft als Strategie der Zukunft» - Alfred Münger, professeur à la HEG-FR a rédigé son livre comme outil pratique à disposition des entrepreneurs et entrepreneuses motivé·e·s à développer une économie durable. heg-fr.ch/fr/medias-et-commu…
4
3
Alessandro retweeted
8
508
29
2,100