Jeremie Mary (@JeremieMary) profile picture

Senior Researcher

Criteo AI Lab
Joined July 2011
Photo by Jeremie Mary (@JeremieMary)Photo by Jeremie Mary (@JeremieMary)Photo by Jeremie Mary (@JeremieMary)Photo by Jeremie Mary (@JeremieMary)Photo by Jeremie Mary (@JeremieMary)
Claim: gpt-5-pro can prove new interesting mathematics. Proof: I took a convex optimization paper with a clean open problem in it and asked gpt-5-pro to work on it. It proved a better bound than what is in the paper, and I checked the proof it's correct. Details below.
314
1,214
443
8,000
7,529,193
Bonjour @Interieur_Gouv, notre système de visa marche tellement bien qu'une scientifique en IA de renommée internationale, récemment primée par l'ERC, ne peut même pas venir 10 jours pour présenter ses travaux à une conférence prestigieuse. Une idée pour améliorer notre image ?
I still don’t have a visa for ICCV. let’s keep telling ourselves how inclusive we are. the summer I got 3 papers and an ERC and still wasn’t allowed in Europe. if you read this and think I deserve this because I want to live in the same country as my family, just unfollow me.
1
8
38
6,397
My biggest gripe with these tools is the possibility of having my work stolen. What if you work on a new big idea and then OpenAI or Microsoft employees see it and scoop you. The fact that we cannot run this on our own server is a huge privacy problem.
10x engineer is a myth. 100x AI-powered engineer is more real than ever. As OpenAI winds down Codex, Microsoft announces GitHub Copilot X. I think it's almost as exciting as GPT-4 itself: - Copilot Chat: any piece of text database will be "chattable", and codebase is no different. Don't read your code, talk to it. - Copilot for Pull Request: improves *human collaboration*. Now GPT will accelerate not just a single dev, but entire OSS communities. - Copilot CLI: bash is so unintuitive and awkward sometimes. No more bash, just English. - Copilot doc: thanks to GPT-4's much longer context (32K tokens), you can fit entire docs in one go. No need to memorize any doc - simply retrieve from the prompt. It's a bit annoying that there is a separate waitlist for each item ... I'll link them in 🧵:
2
7
1,337
Jeremie Mary retweeted
OK so @OpenAI's new #ChatGPT can basically just generate #AIart prompts. I asked a one-line question, and typed the answers verbatim straight into MidJourney and boom. Times are getting weird...🤯
412
3,412
808
20,437
How colliders are the best microscopes we can possibly build A short thread 🧵1/10
10
98
7
447
Update to our paper "EdiBERT: a generative model for image editing" with @TIssenhuth, Ugo Tanielian, @JeremieMary! We can do all sorts of image editing with a single model! 📜arxiv.org/abs/2111.15264 💻github.com/EdiBERT4ImageMani… 📓colab.research.google.com/gi… 🧵👇 1/7
2
4
1
15
Surprising extension for Michelangelo's work by DALLE2 reddit.com/r/interestingasfu…
1
We have 3 space dimensions but only one dimension of time. Why is there just one time? What would happen if there were more time dimensions? A short 🧵that requires some math. (You can probably follow w/o math as well) 1/16
106
1,110
170
7,024
Jeremie Mary retweeted
We are looking for a research intern (willing to continue for a PhD). The project will be co-supervised by @gulvarol (@ImagineENPC) and by Andrew Zisserman (@Oxford_VGG). More at: imagine.enpc.fr/~varolg/jobs…
2
30
1
82
Jeremie Mary retweeted
ASML is the most important company you've never heard of. The $300B+ Dutch firm makes the machines that make semiconductors. Each one costs $150m and access to them are a huge geopolitical flashpoint. Here's a breakdown 🧵
252
3,587
713
15,872
The Criteo/CAp21 Privacy-preserving AI Challenge is over!✅ Congratulations to our winners: Rémy Frenoy, Louise Harding and @goulagman 👏🏆 #MachineLearning #privacy
We are proud to propose this year at #AdKDD2021 a challenge to develop new ML tools able to work in the context of the Privacy Sandbox. To get more info and to join, see our article by @JeremieMary 👉 tinyurl.com/ya5arvx2 #KDD2021 #machinelearning
The Criteo/CAp21 Privacy-preserving AI Challenge is🔛! Check out our explanatory @Medium article 👉tinyurl.com/5t5ktrwn 👈 and join in!😊 #MachineLearning #privacy
1
11
1
17
Please RT! We are looking for a research engineer in machine learning to join our team @inria_grenoble, with a focus on open source software. More information here: https://xitter.cf/t.co/cKfi6CVNu1
4
117
1
108
#COVID19 pourrait reculer devant #Interferon inhalé, et avec quasi aucun effet indésirable contrairement à l'#hydroxychloroquine. Aidez-nous à le prouver! N'hésitez pas à transmettre! Objectif envoi du projet aux autorités sanitaires : mardi 21 avril. leetchi.com/c/covid-19-essai…
1
33
5
25
Mea culpa. Objectif beaucoup plus urgent mardi 14 avril. Entre soins, x dossiers, la gestion financière, pour respecter l éthique et la loi, à faire le boulot de x personnes, dsl... J espère que la cagnotte stimulera les autorités à repondre plus vite qu en 10j. Joyeuses paques!
1
1
2
He is my brother, new on Tweeter but as an hospital pharmacist and researcher he has an important opinion to share about on #covid19 possible treatments.
1/8 Pharmacien Hospitalier clinicien en réanimation et chercheur à Amiens, Je vois nos capacités de soins de réanimation en extrême tension et une prise en charge du #COVID-19 sans consensus.
3
Can you explain what is about machine learning in the work she presented?
Folks, have a winner. For the stochastic bandit, this gem from Odalric-Ambrym Maillard's thesis, called EwS, works incredibly well and gives explicit probabilities on the arms. I compare EwS with Thompson & Tsallis in gist.github.com/n17s/44a7ddc… @djhsu @neu_rips @CsabaSzepesvari
2
9
1
39