David Picard
@davidpicard.bsky.social
Professor of Computer Vision/Machine Learning at Imagine/LIGM, École nationale des Ponts et Chaussées @ecoledesponts.bsky.social Music & overall happiness 🌳🪻 Born well below 350ppm 😬 mostly silly personal views
📍Paris 🔗 https://davidpicard.github.io/
📍Paris 🔗 https://davidpicard.github.io/
Pinned
David Picard
@davidpicard.bsky.social
· Oct 8
🚨Updated: "How far can we go with ImageNet for Text-to-Image generation?"
TL;DR: train a text2image model from scratch on ImageNet only and beat SDXL.
Paper, code, data available! Reproducible science FTW!
🧵👇
📜 arxiv.org/abs/2502.21318
💻 github.com/lucasdegeorg...
💽 huggingface.co/arijitghosh/...
TL;DR: train a text2image model from scratch on ImageNet only and beat SDXL.
Paper, code, data available! Reproducible science FTW!
🧵👇
📜 arxiv.org/abs/2502.21318
💻 github.com/lucasdegeorg...
💽 huggingface.co/arijitghosh/...
Reposted by David Picard
Reposted by David Picard
I…
November 26, 2025 at 7:40 PM
I…
Nicolas ( @nicolasdufour.bsky.social ) is defending his PhD right now.
I was so in awe of the presentation that I even forgot to take pictures 😅
I was so in awe of the presentation that I even forgot to take pictures 😅
November 26, 2025 at 6:00 PM
Nicolas ( @nicolasdufour.bsky.social ) is defending his PhD right now.
I was so in awe of the presentation that I even forgot to take pictures 😅
I was so in awe of the presentation that I even forgot to take pictures 😅
Reposted by David Picard
@karpathy.bsky.social posted this on the other site (link in alt) and I wanted to talk about it for a second.
There's no doubt that Nano Banana is impressive. But there's errors in its solutions. Worse, the errors are subtle.
The final output is right, but steps aren't. Making it wrong!
There's no doubt that Nano Banana is impressive. But there's errors in its solutions. Worse, the errors are subtle.
The final output is right, but steps aren't. Making it wrong!
November 25, 2025 at 9:35 PM
@karpathy.bsky.social posted this on the other site (link in alt) and I wanted to talk about it for a second.
There's no doubt that Nano Banana is impressive. But there's errors in its solutions. Worse, the errors are subtle.
The final output is right, but steps aren't. Making it wrong!
There's no doubt that Nano Banana is impressive. But there's errors in its solutions. Worse, the errors are subtle.
The final output is right, but steps aren't. Making it wrong!
Reposted by David Picard
Check out NAF: an effective ViT feature upsampler to produce excellent (and eye-candy) pixel-level feature maps.
NAF outperform both VFM-specific upsamplers (FeatUp, JAFAR) and VFM-agnostic methods (JBU, AnyUp) over multiple downstream tasks 👇
NAF outperform both VFM-specific upsamplers (FeatUp, JAFAR) and VFM-agnostic methods (JBU, AnyUp) over multiple downstream tasks 👇
Need pixel-level features from your backbone (DINOv3, CLIP, RADIO, FRANCA...)?
🚀Introducing NAF: A universal, zero-shot feature upsampler.
It turns low-res ViT features into pixel-perfect maps.
-⚡ Model-agnostic
-🥇 SoTA results
-🚀 4× faster than SoTA
-📈 Scales up to 2K res
🚀Introducing NAF: A universal, zero-shot feature upsampler.
It turns low-res ViT features into pixel-perfect maps.
-⚡ Model-agnostic
-🥇 SoTA results
-🚀 4× faster than SoTA
-📈 Scales up to 2K res
November 25, 2025 at 6:36 PM
Check out NAF: an effective ViT feature upsampler to produce excellent (and eye-candy) pixel-level feature maps.
NAF outperform both VFM-specific upsamplers (FeatUp, JAFAR) and VFM-agnostic methods (JBU, AnyUp) over multiple downstream tasks 👇
NAF outperform both VFM-specific upsamplers (FeatUp, JAFAR) and VFM-agnostic methods (JBU, AnyUp) over multiple downstream tasks 👇
Reposted by David Picard
🧵 #Social | Les professionnelles du social perçoivent de faibles salaires (14 640 euros de revenu salarial annuel moyen) et sont confrontées à des conditions de travail difficiles : temps partiel subi, horaires de travail irréguliers, pénibilité...
👉 www.insee.fr/fr/statistiq...
👉 www.insee.fr/fr/statistiq...
November 24, 2025 at 3:14 PM
🧵 #Social | Les professionnelles du social perçoivent de faibles salaires (14 640 euros de revenu salarial annuel moyen) et sont confrontées à des conditions de travail difficiles : temps partiel subi, horaires de travail irréguliers, pénibilité...
👉 www.insee.fr/fr/statistiq...
👉 www.insee.fr/fr/statistiq...
Reposted by David Picard
Applications now open for our PhD program at ESSEC:
essec.edu/en/program/p...
essec.edu/en/program/p...
November 22, 2025 at 11:38 AM
Applications now open for our PhD program at ESSEC:
essec.edu/en/program/p...
essec.edu/en/program/p...
If everything is frozen, nothing feels the same
November 22, 2025 at 7:55 AM
If everything is frozen, nothing feels the same
Reposted by David Picard
Une raison de plus de vous abonner à la chaîne YouTube du colloque #GRETSI :
"Transport optimal, de Monge à l’apprentissage profond", conférence plénière de Julie Delon au colloque #GRETSI2025
www.youtube.com/watch?v=ujYS...
"Transport optimal, de Monge à l’apprentissage profond", conférence plénière de Julie Delon au colloque #GRETSI2025
www.youtube.com/watch?v=ujYS...
Plénière de Julie Delon
YouTube video by GRETSI
www.youtube.com
November 21, 2025 at 10:28 AM
Une raison de plus de vous abonner à la chaîne YouTube du colloque #GRETSI :
"Transport optimal, de Monge à l’apprentissage profond", conférence plénière de Julie Delon au colloque #GRETSI2025
www.youtube.com/watch?v=ujYS...
"Transport optimal, de Monge à l’apprentissage profond", conférence plénière de Julie Delon au colloque #GRETSI2025
www.youtube.com/watch?v=ujYS...
Reposted by David Picard
Check out the amazing work by my collaborators Lukas and René, who trained a ViT-based masked autoencoder to learn patterns in mel-spectrograms of bird vocalizations without annotations, and then cleverly aggregated the learned features to solve downstream bird species classification tasks. 🐦
Can Masked Autoencoders Also Listen to Birds?
Lukas Rauch, René Heinrich, Ilyass Moummad, Alexis Joly, Bernhard Sick, Christoph Scholz
Action editor: Chuan Sheng Foo
https://openreview.net/forum?id=GIBWR0Xo2J
#autoencoders #bioacoustic #audio
Lukas Rauch, René Heinrich, Ilyass Moummad, Alexis Joly, Bernhard Sick, Christoph Scholz
Action editor: Chuan Sheng Foo
https://openreview.net/forum?id=GIBWR0Xo2J
#autoencoders #bioacoustic #audio
November 21, 2025 at 9:28 AM
Check out the amazing work by my collaborators Lukas and René, who trained a ViT-based masked autoencoder to learn patterns in mel-spectrograms of bird vocalizations without annotations, and then cleverly aggregated the learned features to solve downstream bird species classification tasks. 🐦
Reposted by David Picard
Nice update from #ICLR2026 PCs 🫡
Curious what "LLM detection tools" PCs are using
I am also wondering if NeuRep workshop is taking some action for papers (presumably in question and then) "withdrawn" from ICLR 2026 by the authors
blog.iclr.cc/2025/11/19/i...
Curious what "LLM detection tools" PCs are using
I am also wondering if NeuRep workshop is taking some action for papers (presumably in question and then) "withdrawn" from ICLR 2026 by the authors
blog.iclr.cc/2025/11/19/i...
November 21, 2025 at 7:13 AM
Nice update from #ICLR2026 PCs 🫡
Curious what "LLM detection tools" PCs are using
I am also wondering if NeuRep workshop is taking some action for papers (presumably in question and then) "withdrawn" from ICLR 2026 by the authors
blog.iclr.cc/2025/11/19/i...
Curious what "LLM detection tools" PCs are using
I am also wondering if NeuRep workshop is taking some action for papers (presumably in question and then) "withdrawn" from ICLR 2026 by the authors
blog.iclr.cc/2025/11/19/i...
November 20, 2025 at 4:12 PM
Reposted by David Picard
#CNRSnews 📰 Energy transition is a myth, based on the misconception that energy sources were replaced by “cleaner” alternatives over the past 200 years. Carbon emissions have actually increased, hence the need for better informed strategies to fight global warming
news.cnrs.fr/articles/the...
news.cnrs.fr/articles/the...
The myth of energy transition
The concept of an “energy transition” is misleading, states the CNRS science historian Jean-Baptiste Fressoz. He explains why coal and oil never replaced wood, and that the fight against climate chang...
news.cnrs.fr
November 20, 2025 at 8:02 AM
#CNRSnews 📰 Energy transition is a myth, based on the misconception that energy sources were replaced by “cleaner” alternatives over the past 200 years. Carbon emissions have actually increased, hence the need for better informed strategies to fight global warming
news.cnrs.fr/articles/the...
news.cnrs.fr/articles/the...
Reposted by David Picard
Grok n'est pas une source d'information. Que faut-il de plus – après une semaine de désinformation par Grok concernant les attentats du 13 novembre – voyez ce cas gravissime de #négationnisme et d' #antisémitisme.
En France, c'est un délit. Pourquoi puis-je voir ce post alors ?
En France, c'est un délit. Pourquoi puis-je voir ce post alors ?
November 19, 2025 at 11:48 AM
Grok n'est pas une source d'information. Que faut-il de plus – après une semaine de désinformation par Grok concernant les attentats du 13 novembre – voyez ce cas gravissime de #négationnisme et d' #antisémitisme.
En France, c'est un délit. Pourquoi puis-je voir ce post alors ?
En France, c'est un délit. Pourquoi puis-je voir ce post alors ?
Reposted by David Picard
Autant de *factchecking* pour en arriver là. Étonnant, non?
—-> Spoiler : NON. Le négationnisme n’est pas un problème de rapport aux faits. C’est un problème de vision politique. Quand les fafs prennent le pouvoir, ils imposent leur vision politique. Pour s’y opposer, nous devons imposer la nôtre.
—-> Spoiler : NON. Le négationnisme n’est pas un problème de rapport aux faits. C’est un problème de vision politique. Quand les fafs prennent le pouvoir, ils imposent leur vision politique. Pour s’y opposer, nous devons imposer la nôtre.
November 19, 2025 at 11:32 AM
Autant de *factchecking* pour en arriver là. Étonnant, non?
—-> Spoiler : NON. Le négationnisme n’est pas un problème de rapport aux faits. C’est un problème de vision politique. Quand les fafs prennent le pouvoir, ils imposent leur vision politique. Pour s’y opposer, nous devons imposer la nôtre.
—-> Spoiler : NON. Le négationnisme n’est pas un problème de rapport aux faits. C’est un problème de vision politique. Quand les fafs prennent le pouvoir, ils imposent leur vision politique. Pour s’y opposer, nous devons imposer la nôtre.
Pro-tip: if you send me an email asking for an internship, a PhD or a postdoc position, don't copy/paste an over-hyped summary of one of my papers you just asked ChatGPT to spit out.
Pro-tip#2: don't do that with other professors as well. It's not just me.
Pro-tip#2: don't do that with other professors as well. It's not just me.
November 19, 2025 at 8:24 AM
Pro-tip: if you send me an email asking for an internship, a PhD or a postdoc position, don't copy/paste an over-hyped summary of one of my papers you just asked ChatGPT to spit out.
Pro-tip#2: don't do that with other professors as well. It's not just me.
Pro-tip#2: don't do that with other professors as well. It's not just me.
Found this in my notebook. Never been more appropriate. 🥰
November 18, 2025 at 5:59 PM
Found this in my notebook. Never been more appropriate. 🥰
Perks of having meetings in Paris: if you plan ahead, there's always something interesting to see. For example a stroll in the Montparnasse cemetery, among some of the greatest writers.
November 18, 2025 at 9:58 AM
Perks of having meetings in Paris: if you plan ahead, there's always something interesting to see. For example a stroll in the Montparnasse cemetery, among some of the greatest writers.
Reposted by David Picard
A model for detecting LLM generated text is under review at #ICLR2026. The website using this model detects one of the reviews for this paper as LLM generated ...
🍿🍿🍿🍿🍿🍿🍿🍿🍿
🍿🍿🍿🍿🍿🍿🍿🍿🍿
(As an aside, it looks like the model EditLens that is used for these scores is itself under review at ICLR and classifies one of its reviews as 100% AI which makes the whole thing very meta)
November 17, 2025 at 1:55 PM
A model for detecting LLM generated text is under review at #ICLR2026. The website using this model detects one of the reviews for this paper as LLM generated ...
🍿🍿🍿🍿🍿🍿🍿🍿🍿
🍿🍿🍿🍿🍿🍿🍿🍿🍿
Reposted by David Picard
#KostasThoughts: Please stop using AI tools to write reviews. I want your authentic thoughts, even if they contain spelling, grammar, & factual mistakes. An LLM should not influence your judgment. The purpose of multiple reviewers is to provide diverse views and error checking.
November 17, 2025 at 1:50 AM
#KostasThoughts: Please stop using AI tools to write reviews. I want your authentic thoughts, even if they contain spelling, grammar, & factual mistakes. An LLM should not influence your judgment. The purpose of multiple reviewers is to provide diverse views and error checking.
Reposted by David Picard
There's a reviewer at ICLR who apparently always writes *exactly* 40 weaknesses and comments no matter what paper he's reviewing.
Exhibit A: openreview.net/forum?id=8qk...
Exhibit B: openreview.net/forum?id=GlX...
Exhibit C: openreview.net/forum?id=kDh...
Exhibit A: openreview.net/forum?id=8qk...
Exhibit B: openreview.net/forum?id=GlX...
Exhibit C: openreview.net/forum?id=kDh...
November 15, 2025 at 2:42 PM
There's a reviewer at ICLR who apparently always writes *exactly* 40 weaknesses and comments no matter what paper he's reviewing.
Exhibit A: openreview.net/forum?id=8qk...
Exhibit B: openreview.net/forum?id=GlX...
Exhibit C: openreview.net/forum?id=kDh...
Exhibit A: openreview.net/forum?id=8qk...
Exhibit B: openreview.net/forum?id=GlX...
Exhibit C: openreview.net/forum?id=kDh...
Reposted by David Picard
左右のリングは反時計回りに同じ速度で回転しているが、右のリングは反時計回りに回転して見えるが、左のリングは（特に中心視では）ゆっくり反時計回りに見えたり、（周辺視では）時計回りに回転して見える。
November 16, 2025 at 12:40 AM
左右のリングは反時計回りに同じ速度で回転しているが、右のリングは反時計回りに回転して見えるが、左のリングは（特に中心視では）ゆっくり反時計回りに見えたり、（周辺視では）時計回りに回転して見える。
But is this list of advice relevant "in the current impetuous and indicates society"?
November 16, 2025 at 8:57 AM
But is this list of advice relevant "in the current impetuous and indicates society"?
Reposted by David Picard
Crossposting because the QT is there but the audience for this is here
November 15, 2025 at 10:39 PM
Crossposting because the QT is there but the audience for this is here