maxime retweeted
Being unemployed doesn't prevent us from doing important science. We took things into our own hands to test which fizzy water was the best, and yOu wON't bElieVE the conclusions. This might not go in my thesis but @vmaxmc2 made a nice article.
maximevidal.com/sparkling-wa…
maxime retweeted
we've just achieved one of the highest iron purity available on the market.
insanely proud of the team, thank you to the day one believers!
maxime retweeted
🥰 Self-supervised, label-free 3D cell imaging is here actu.epfl.ch/news/self-super… #epfl #elife
Congrats to @AChRD4 on this very cool first, first author publication 🏆! This started when he was a bachelor's student (and he's now completing his masters!).
And many thanks to my other awesome co-authors: @TheVetFuturist @vmaxmc2 @CYHSM Yves Paychere Colin Hofmann @itsasimiqbal @SebHausmann and @stephane_pages! 🙏
elifesciences.org/articles/9…
maxime retweeted
sim2real achieved! mini shoggoth can now (slowly) track hands. to what purpose i do not yet know
stack is PPO + mujoco + dynamics randomization + yolo11n
maxime retweeted
just realized the Westworld show was actually about a private attempt at distributed group relative policy optimization with real-world reward functions at scale to solve llm sim2real
maxime retweeted
Announcing Nautilus: A three-month gap program where you get paid –no strings attached– to dive into your craziest, most ambitious project.
Yes, you read that right. Come live with us in San Francisco and build the next Renaissance.
How it works: – at San Francisco, CA
maxime retweeted
#CellSeg3D hits 85K downloads, and we’ve updated our preprint to add a few more key details and added new demo videos 🎥!
Check out the demos from @AChRD4 and @SebHausmann 🔥
youtu.be/3UOvvpKxEAo?feature…
m.youtube.com/watch?v=U2a9Ib…
biorxiv.org/content/10.1101/…
maxime retweeted
🦀🐝🧠🤖 How should we make decisions when faced with systems of uncertain sentience? My new book The Edge of Sentience: Risk and Precaution in Humans, Other Animals, and AI is available online for free at academic.oup.com/book/57949. Please download it! Just click on PDF to get a PDF of the whole book. Individual chapters are also available as PDFs - just click on whichever chapters you want. The book is completely free to everyone and I'm grateful for all shares and forwards.
maxime retweeted
🚨 Introducing "ColPali: Efficient Document Retrieval with Vision Language Models" !
We use Vision LLMs + late interaction to improve document retrieval (RAG, search engines, etc.), solely using the image representation of document pages ! arxiv.org/abs/2407.01449
🧵(1/N)
maxime retweeted
I am thrilled to share that our paper "SuperAnimal pretrained pose estimation models for behavioral analysis" has just been published at Nature Communications! @mwmathislab nature.com/articles/s41467-0…. More details about our animal pose foundation models:
maxime retweeted
𝗙𝗶𝗻𝗮𝗹𝗹𝘆 𝗽𝘂𝗯𝗹𝗶𝘀𝗵𝗲𝗱 𝗶𝗻 @CellCellPress !
How does the brain processes proprioceptive information?🧠
We use a task-driven modeling approach by combining biomechanics 🦾 and neural network modeling💻 to investigate different hypotheses. 1/N
cell.com/cell/pdf/S0092-8674…
maxime retweeted
> wget big stuff
> oh takes more than 5sec, let’s open X
> wget finishes 2min later
> 20min later: wait, is my download complete ?
Can relate ? github.com/aamorel/cmd-learn monitors long commands in your terminal, opens a browser window closes it you when the long command is over.
maxime retweeted
.@karpathy's micrograd is really neat. To turbo-teach myself cuda, I extended it with cuda kernels on 2D tensors. It’s a bit heftier, but it’s still very readable for anyone wanting to quickly understand how gpu acceleration works.
github.com/mlecauchois/micro…
maxime retweeted
🔔Updated on arXiv: it’s a pretty big overhaul of presentation and new SOTA results, plus even better model performance for quadrupeds! #PoseEstimation #ComputerVision #neuroscience #MachineLearning
arxiv.org/abs/2203.07436
🐁🐘🐿🔥SuperAnimals for #Pose
-No human labeling
- Video analysis on over 45 species with only 2 global classes of animal pose models
-If SA needs fine-tuning, its 10× more data efficient & 2X outperforms prior transfer learning
📝@shaokaiyeah et al arxiv.org/abs/2203.07436
🧵⬇️
maxime retweeted
Pleased to share Spellburst: an LLM-powered creative coding tool, accepted at @ACMUIST! Artists can move between semantic (high level) and syntactic (low level) ideas and explore many branches in parallel
Paper: bit.ly/47sIJ9y
More on the @Replit/@StanfordHCI collab:
maxime retweeted
Check our latest work that, for the first time, enables people to do behavior analysis using natural language alone!
🔮Ever wished you could just say what behavioral analysis you wanted? ... "Plot the animal's trajectory" "Count the head hips in the open arm"
👀🪄we present AmadeusGPT🎻 to do this ...
🗞@shaokaiyeah @jessy_lauer @zhoumu53 @trackingplumes and me arxiv.org/abs/2307.04858
🧵⬇️
maxime retweeted
🐁🐘🐿🔥SuperAnimals for #Pose
-No human labeling
- Video analysis on over 45 species with only 2 global classes of animal pose models
-If SA needs fine-tuning, its 10× more data efficient & 2X outperforms prior transfer learning
📝@shaokaiyeah et al arxiv.org/abs/2203.07436
🧵⬇️




















