Vigile Hoareau

604 posts

Vigile Hoareau banner
Vigile Hoareau

Vigile Hoareau

@Vigile_480

@_crowdaa #Cognitive_Psychology_PhD #TextMiner #entrepreneur #Programmer #PatentAuthor #HipHopOldSchooler #CreoleWriter

Saint-Gilles, Reunion 🇷🇪 Katılım Ocak 2011
631 Takip Edilen416 Takipçiler
Vigile Hoareau retweetledi
DAIR.AI
DAIR.AI@dair_ai·
Top AI Papers of The Week (September 1-7): - rStar2-Agent - Self-Evolving Agents - Adaptive LLM Routing - Universal Deep Research - Implicit Reasoning in LLMs - Why Language Models Hallucinate - Limitations of Embedding-based Retrieval Read on for more:
English
8
125
746
81.6K
Vigile Hoareau
Vigile Hoareau@Vigile_480·
@svlevine @ylecun Another way of saying it would be that the syntagmatic structure of language embodied sort of paradigmatic information ( one token can make a shift in the meaning of the current context) and the equivalent for video would not be true. Video would be a different type of language…
English
0
0
0
118
Sergey Levine
Sergey Levine@svlevine·
I always found it puzzling how language models learn so much from next-token prediction, while video models learn so little from next frame prediction. Maybe it's because LLMs are actually brain scanners in disguise. Idle musings in my new blog post: sergeylevine.substack.com/p/language-mod…
English
51
171
1.3K
314.7K
Vigile Hoareau
Vigile Hoareau@Vigile_480·
#LATechWeek is just killing it @KatiaAmeri!! This is the application we’ve crafted in the last minute using @crowdaa_! We gather +190 events in it and +150 speakers! I hope it helps!
English
1
0
1
62
Vigile Hoareau retweetledi
Jon Erlichman
Jon Erlichman@JonErlichman·
Average cost for 1 gigabyte of storage: 40 years ago: $238,000 35 years ago: $48,720 30 years ago: $5,152 25 years ago: $455 20 years ago: $5 15 years ago: $0.55 10 years ago: $0.05 5 years ago:   $0.03 Today:             $0.01
English
587
3.5K
48.7K
7.7M
Vigile Hoareau
Vigile Hoareau@Vigile_480·
« Vos e-mails ont été considérés comme spam par erreur. Le déblocage de votre compte est en cours et sera effectif sous 30 minutes. » X 3 🥳🎈
Français
1
0
0
105
Vigile Hoareau
Vigile Hoareau@Vigile_480·
@OVHcloud_FR @olesovhcom difficile de faire du business avec vous! Déjà deux fois dans la semaine que vous flaguer nos e-mails de service en spam. Deux fois qu’on résout l’incident avec vos équipes en nous disant que c’est ok… ça vous dirait d’arrêter de casser notre biz?
Français
2
0
0
126
Yann LeCun
Yann LeCun@ylecun·
I don't mean to make a bad taste joke, but pronouncing GPT-4 in quasi-French ("gé pé té for") sounds *very* awkward.
HT
119
152
1.6K
316.6K
Lucas Beyer (bl16)
Lucas Beyer (bl16)@giffmana·
@ylecun I mean, was anyone even working on AI before OpenAI started to work on it? I can't remember anything or anyone...
English
11
0
74
15.2K
Yann LeCun
Yann LeCun@ylecun·
This critique of AI gets absolutely everything wrong. Quote: "There have been no major breakthroughs in the academic discipline of artificial intelligence for a couple of decades" What ??? Seriously ???
James Bridle@jamesbridle

"It has read most of the internet, and it knows what human language is supposed to sound like, but it has no relation to reality whatsoever." I'm in the @guardian today, writing about image generators, ChatGPT, and the stupidity of AI: theguardian.com/technology/202…

English
108
94
1.5K
582.6K
Vigile Hoareau
Vigile Hoareau@Vigile_480·
@ylecun @pmarca You can also consider that most of the “non-super-valuable” papers are, at least, contributing to perpetuate the methodology of practicing science… it’s part of the process.
English
0
0
0
13
Yann LeCun
Yann LeCun@ylecun·
@pmarca And you rarely know right away which ones are going to spark a new avenue. It often takes several years before an idea comes to fruition in practice. A good recent example is diffusion models.
English
4
4
90
7.4K
Marc Andreessen 🇺🇸
Marc Andreessen 🇺🇸@pmarca·
What % of published academic research papers are worth anything? Roughly: Half of the fields are useless. 90% of the research from real fields is marginal at best. Half or more of the remainder doesn’t replicate. = ~2.5%? AI text generation can’t make this much worse, can it?
English
333
474
4.2K
2M
Vigile Hoareau retweetledi
Yann LeCun
Yann LeCun@ylecun·
My unwavering opinion on current (auto-regressive) LLMs 1. They are useful as writing aids. 2. They are "reactive" & don't plan nor reason. 3. They make stuff up or retrieve stuff approximately. 4. That can be mitigated but not fixed by human feedback. 5. Better systems will come
English
97
554
3.3K
791K

Keşfet

@svlevine @ylecun @KatiaAmeri @crowdaa_ @H974Eric @saintdenis974 @zamela9748 @Rodpaname