data science PhD student 🧬🧠🖥️ scRNA data, retina & brain | sci-comm 🤖❤️🌍 AI & sustainability | easily sidetracked by small details in plots & birds 📈🔍🦜
Tübingen, Germany
Joined March 2020
- Tweets891
- Following1.2K
- Followers642
- Likes3.8K
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
This other site is starting to gel. You can find me there.
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
I am very skeptical of the AI/ML/computational methods to predict protein-protein interactions. Prove me wrong.
I present a challenge for anyone who claims they can predict protein-protein interactions.
1/n
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
Another example of what I @b_mittelstadt @c_russl termed careless speech. Subtle hallucinations are dangerous & developers are not (yet) liable for them. We argue they should. See paper: Do LLMs have a legal duty to tell the truth? tinyurl.com/435jba5p tinyurl.com/4wpske2s
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
We updated our ICLR dataset (see github.com/berenslab/iclr-da…) with blind 2025 submissions to @iclr_conf. Over 10k submissions this year.
I like how this embedding (SBERT + tSNE) shows which ML areas are old-school and which ones are currently booming.
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
Warum lügt die @CSU?
Kurz gesagt: Um von echten Problemen abzulenken, also um euch zu verarschen.
Also kuschelt eure Haustiere, widersprecht so einem Mist und lasst uns die Probleme lösen, die euch im Alltag stören und nicht welche, die die Union erfunden hat 🐶😽❤️
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
Sequencing folks. You seem to use UMAP all the time. But everyone is making fun of it. Someone must be willing to defend it?
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
🤖🧠NOW OUT IN PNAS🧠🤖
Language models show many surprising behaviors. E.g., they can count 30 items more easily than 29
In Embers of Autoregression, we explain such effects by analyzing what LMs are trained to do
pnas.org/doi/10.1073/pnas.23…
Major updates since the preprint!
1/n
ALT At the top is the title of the paper: "Embers of autoregression show how large language models are shaped by the problem they are trained to solve". Below on the left is a screenshot of ChatGPT being asked to count how many words are in a list. The correct answer is 29, but it says 30. Next to it is a plot showing ChatGPT's accuracy at counting elements in a list; in general, it does well on multiples of 10 but poorly on other numbers. The explanation offered at the bottom of the image is: In training sets, round numbers are much more common than other numbers.
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
Datamapplot 0.4 is out now, and has far more powerful and effective interactive plots.
Here is an example of a Data Map of 2.4 million papers on ArXiv, ready to be explored.
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
Nobel prize for t-SNE!
BREAKING NEWS
The Royal Swedish Academy of Sciences has decided to award the 2024 #NobelPrize in Physics to John J. Hopfield and Geoffrey E. Hinton “for foundational discoveries and inventions that enable machine learning with artificial neural networks.”
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
The strange state of current publishing. I think some stats on our recent Cellstates paper provide interesting food for thought. The preprint was put on BioRxiv and within 3 months there were ~8500 abstract views and >2000 PDF downloads.
We also submitted it to PCB 6-11-2023 1/n
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
PSA:
If you don't want your X posts used to train Grok, you now have to explicitly opt-out.
Go to xitter.cf/settings/grok_settings and uncheck the box.
If link doesn't work, go to Settings->Privacy and Safety->Grok
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
ICML 2024 (held in Vienna) registrations vs. registrations *per million inhabitants* by country.
The barplot of registrations on the log scale was shown yesterday during the opening. I took a photo, digitized with WebPlotDigitizer, and normalized per capita.
#ICML2024 @icmlconf
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
"The Harris campaign has seen a massive influx of donations, often from people who haven't donated before. This is democracy working as intended, folks!"
Super happy to see donations, but "democracy as intended" is more about voting with your vote than your money, honestly.
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
Excited to be in Vienna at ICML to present my work with @samgreydanus on scaling _down_ deep learning with MNIST-1D!
We show how one can study serious deep learning with n=5000 and d=40.
Paper: openreview.net/forum?id=n9pr…
Code: github.com/greydanus/mnist1d…
Find me at poster session 1.
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
Few evolutionary psychologists engage with the modern genomic literature. Brendan Zietsch is an exception and he has a new paper out, criticising explanations for heritable variation based on balancing selection. He's pulling no punches.
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
Certainly, here is a possible tweet about your article: Over 10% of science is now, in part, written by LLMs. Let’s delve into the data economist.com/science-and-te…
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
Bei Messenger-Diensten sollen die Nutzer, wenn es nach der Ratspräsidentschaft geht, in Zukunft zustimmen müssen, dass ihre Inhalte durchleuchtet werden. Nur wer zustimmt, dass seine Inhalte durchsucht werden, darf visuelle Inhalte oder Internetadressen teilen.
Das soll Verschlüsselung sichern, zwingt aber jeden Nutzer dazu, seine Privatsphäre aufzugeben. Auch deshalb hat Signal bereits angekündigt, Europa verlassen zu wollen, wenn dies beschlossen würde.
Jan Lause 🟦 @janlause.bsky.social🦉 retweeted
Our paper is featured in the cover of the Patterns' issue this month. How cool is the illustration by Franz Stämmele (staemmele.com/)? 🤩
Our June issue is now live!
cell.com/patterns/issue?pii=…
On the cover this month, researchers from the Tübingen AI Center use a large language model to explore and visualization the entirety of the biomedical literature in PubMed.
Full paper: cell.com/patterns/fulltext/S…
ALT Every year, over 1.5 million scientific articles are published in the fields of biomedicine and life sciences. This vast amount of publications makes it challenging to keep up with the literature and have an overview of new developments in the field. In this issue, González-Márquez et al. introduce a map to help navigate this complex landscape. The cover image shows a woman studying a map while pondering a landscape consisting of mountains of documents. Image credit: Franz-Georg Staemmele.