Pierluca D'Oro

@proceduralia

Making the world verifiable. Previously @meta, @Mila_Quebec

proceduralia.com

Joined September 2012

611 Following

2,561 Followers

Replies Reposts Media Articles

Pierluca D'Oro
@proceduralia
Aug 19
I have an answer for this, but you guys might not like it
Joe Weisenthal
@TheStalwart
Aug 18
To what degree would people say that so-called “reward hacking” by AI models is a conceptual descendant of what would have been called “overfitting” in the ML era?
2
Pierluca D'Oro
@proceduralia
Jul 26
Euro: Sicilian edition
1
Pierluca D'Oro
@proceduralia
Jun 28
I agree, and there's more to it. There is a entire class of problems that I call deployment-complete"reinforcement learning problems. Computer use belongs to this class. RL works well when one can easily reset the environment to diverse enough situations, when there is a
Dwarkesh Patel
@dwarkesh_sp
Jun 27
Here's a question I find confusing and interesting and which actually tells us a lot about the nature of current AI progress: Why has progress on computer use been so slow? Computer use is so clearly verifiable. I think the answer is that it is not enough for a domain to be
5
Pierluca D'Oro
@proceduralia
Jun 15
Article
The founder-mode researcher
There is something I learned during my PhD. AI research is far from being the discipline in which the only thing that matters for a scientist is to work on a specific problem, come up with a solution,
Pierluca D'Oro
@proceduralia
Jun 12
I myself didn't fully understand what the deepseek r1 hype was about, but at some point I realized where the mismatch in hype was coming from. For many people with reinforcement learning knowledge, the o1 announcement was pretty clear. o1 was a model trained with reinforcement
Elvis Nava
@elvisnavah
Jun 11
Looking back, I never understood the hype of the "deepseek moment" from last year. Distilling others' models is possible and easier than pushing the frontier. Like, nobody remembers Alpaca?
2