François Fleuret ✓

@francoisfleuret · Switzerland · joined 26 Dec 2009

Research Scientist @meta (FAIR), Prof. @Unige_en, co-founder @neural_concept_. I like reality.

54 513Followers

480Following

30 876Posts total

887.6KViews on collected posts

Derniers posts

François Fleuret ✓@francoisfleuret14 Sep

So doing prompt as minimax documentation says actually works very well. Still on a $500 4060Ti with 16GB VRAM. https://t.co/RSLnQvyLIy

0:13

1.3K views · 5 likes · 0 reposts · 0 replies Open on X →

François Fleuret ✓@francoisfleuret14 Sep

Let's put it another way: AI is cargo cult that works. 3.3K views · 37 likes · 0 reposts · 2 replies Open on X →

François Fleuret ✓@francoisfleuret14 Sep

@francoisfleuret Wait, 'biased stochastic search that eventually hits target' also describes a PhD student pretty well. The open question is why the bias transfers across tasks at all. 5.8K views · 21 likes · 0 reposts · 1 replies Open on X →

François Fleuret ✓@francoisfleuret14 Sep

IMO the most reasonable high level model one can have of the AI behemoths is still a brute-force stochastic search in token space where the sampling is biased to mimic mathematician writing that eventually hit target. 1/3 97.8K views · 632 likes · 47 reposts · 30 replies Open on X →

François Fleuret ✓@francoisfleuret14 Sep

Given how often an original idea suddenly appears in many places at the same time, simply because the said idea naturally derives from the collective knowledge at the time, I'd be very careful with assessing if an idea was "stolen" or not without tangible evidences. 4.6K views · 85 likes · 1 reposts · 3 replies Open on X →

François Fleuret ✓@francoisfleuret13 Sep

OpenAI says it’s “categorically” impossible that a mathematician’s user data over the last two months could have influenced its monumental solution to a Millennium Prize problem. Says Tristan Buckmaster: “I just don’t believe them.” https://t.co/5ZvJSYnAIp 189.4K views · 576 likes · 46 reposts · 36 replies Open on X →

François Fleuret ✓@francoisfleuret17 Mar 2024

@francoisfleuret @alfcnz @unige_en @PyTorch Hi Professor, I think the lectures would be even more popular if they were uploaded to YouTube. 1.7K views · 11 likes · 0 reposts · 1 replies Open on X →

François Fleuret ✓@francoisfleuret10 Oct 2023

Sorry for the marketing, that was to update the pinned tweet! 8.9K views · 21 likes · 0 reposts · 6 replies Open on X →

François Fleuret ✓@francoisfleuret10 Oct 2023

My deep learning course @unige_en is available on-line. 1000+ slides, ~20h of screen-casts. Full of examples in @PyTorch. https://t.co/6OVyjPdwrC And my "Little Book of Deep Learning" is available as a phone-formatted pdf (400k downloads!) https://t.co/qXni5GZOMT https://t.co

574.9K views · 3.6K likes · 709 reposts · 63 replies Open on X →

Face aux comptes de taille comparable

6 posts des 90 derniers jours, à côté de la tranche de 10K–100K abonnés. diffusé largement, mais peu de ces spectateurs réagissent.

Vues médianes5 174ce compte924médiane pour 10K–100K

Portée, %9.49%ce compte3.62%médiane pour 10K–100K

Engagement, %0.56%ce compte1.52%médiane pour 10K–100K

Indicateur	Ce compte	Médiane pour 10K–100K	Rapport
Vues médianes par post	5 174	924	5.60×
Portée (vues ÷ abonnés)	9.49%	3.62%	2.62×
Taux d'engagement	0.56%	1.52%	0.37×

Autres comptes de cette tranche → Comparer avec un autre compte → Comment ces repères sont établis →

Growth & engagement

How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.

Views per post

574.9K10 Oct

8.9K

1.7K17 Mar

189.4K13 Sep

4.6K14 Sep

97.8K

5.8K

3.3K

1.3K

Last 9 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

0.76%10 Oct

0.30%

0.69%17 Mar

0.35%13 Sep

1.95%14 Sep

0.73%

0.38%

1.18%

0.39%

Reactions — likes, reposts, replies and quotes — divided by views. Median for 10K–100K accounts is 1.52%.

What the audience does

Likes47.7%5 010 in total

Reposts7.6%803 in total

Replies1.4%142 in total

Bookmarks43.3%4 554 in total

Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.

The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.

Derniers posts

Face aux comptes de taille comparable

Growth & engagement

Views per post

Engagement rate per post

What the audience does

Comptes similaires