Dan
@dfu.bsky.social
Postdoc @ Emory
Self-assembly & AI researcher. Keyword spew: Biological circuits, omics, neural coding, network science, cognition, artificial life, stochastic modeling.
Aspiring polyglot on the side.
Self-assembly & AI researcher. Keyword spew: Biological circuits, omics, neural coding, network science, cognition, artificial life, stochastic modeling.
Aspiring polyglot on the side.
I personalize my chat LLMs to avoid human-like speech and to sound as emotionless as possible. The answers are closer now to reading encyclopedia entries.
Except, I still respond "are you serious?" or "no, you idiot" when the answers are off.
Except, I still respond "are you serious?" or "no, you idiot" when the answers are off.
November 21, 2025 at 2:07 AM
I personalize my chat LLMs to avoid human-like speech and to sound as emotionless as possible. The answers are closer now to reading encyclopedia entries.
Except, I still respond "are you serious?" or "no, you idiot" when the answers are off.
Except, I still respond "are you serious?" or "no, you idiot" when the answers are off.
Reposted by Dan
We train machine learning models on millions of proteins. But when it comes to making predictions, do we need them to understand all proteins at once? Often, we need an accurate model for the specific protein we are studying or designing. We address this with ProteinTTT arxiv.org/abs/2411.02109 1/🧵
October 23, 2025 at 1:08 PM
We train machine learning models on millions of proteins. But when it comes to making predictions, do we need them to understand all proteins at once? Often, we need an accurate model for the specific protein we are studying or designing. We address this with ProteinTTT arxiv.org/abs/2411.02109 1/🧵
Not caffeinated enough to figure out a new coffee machine. 🤦
October 13, 2025 at 6:57 PM
Not caffeinated enough to figure out a new coffee machine. 🤦
On a scale of 0 to “I only spilled a little bit, why is it everywhere”…
barbecue sauce.
barbecue sauce.
September 7, 2025 at 8:56 PM
On a scale of 0 to “I only spilled a little bit, why is it everywhere”…
barbecue sauce.
barbecue sauce.
Funny thing, I don’t think my chatbot-boosted productivity is from working faster. It’s from working happier and being focused longer. It’s social interaction. And on the other side is a hyper enthusiastic junior that has been trained slightly wrong, as a joke. I won’t let you down buddy.
July 15, 2025 at 6:57 PM
Funny thing, I don’t think my chatbot-boosted productivity is from working faster. It’s from working happier and being focused longer. It’s social interaction. And on the other side is a hyper enthusiastic junior that has been trained slightly wrong, as a joke. I won’t let you down buddy.
After trying all the premium chatbot models I can say that the sweet spot I’m comfortable with is about ~50 words max for writing as a superthesaurus and ~2-3 DSA at a time for refactoring. Generative writing is schlop and vibe coding is hit or miss at any length.
July 15, 2025 at 6:46 PM
After trying all the premium chatbot models I can say that the sweet spot I’m comfortable with is about ~50 words max for writing as a superthesaurus and ~2-3 DSA at a time for refactoring. Generative writing is schlop and vibe coding is hit or miss at any length.
Reposted by Dan
Baidu officially open-sourced its Ernie 4.5 large language model yesterday, marking one of China's biggest AI moves since DeepSeek shook global markets. The release includes 10 different models, ranging from 0.3B to 424B parameters, all available under the Apache License 2.0 for commercial use.
July 2, 2025 at 12:21 PM
Baidu officially open-sourced its Ernie 4.5 large language model yesterday, marking one of China's biggest AI moves since DeepSeek shook global markets. The release includes 10 different models, ranging from 0.3B to 424B parameters, all available under the Apache License 2.0 for commercial use.
Reposted by Dan
I have poked the @rsc.org on X/Twitter a bit over the years (sorry folks...), but huge kudos to them for leaving what has become even more of a toxic and hateful platform www.rsc.org/news-events/... 👏 #chemsky
Social media and connecting with our community
www.rsc.org
June 16, 2025 at 8:49 AM
I have poked the @rsc.org on X/Twitter a bit over the years (sorry folks...), but huge kudos to them for leaving what has become even more of a toxic and hateful platform www.rsc.org/news-events/... 👏 #chemsky
Reposted by Dan
I'm hiring! Are you a technical, senior product leader passionate about driving the next generation of biomolecular simulation software? Then come and join our team! Apply here: bit.ly/pm-bionemo-s...
Senior Product Manager, BioNeMo Simulation
NVIDIA seeks a Product Manager to lead our work in molecular dynamics (MD) and AI-accelerated simulation projects for life sciences. We are interested in finding global authorities and leaders at the ...
bit.ly
June 2, 2025 at 2:46 PM
I'm hiring! Are you a technical, senior product leader passionate about driving the next generation of biomolecular simulation software? Then come and join our team! Apply here: bit.ly/pm-bionemo-s...
Reposted by Dan
📢 Thrilled to announce the return of MoML @ MIT on Oct. 22 and that paper submissions are now open! Students & postdocs w/ accepted papers will be granted FREE admission to attend and have the chance at winning the Octavian-Eugen Ganea Prize for Best Paper! 🧵 #AcademicSky #compchem #drugdesign
May 28, 2025 at 8:28 PM
📢 Thrilled to announce the return of MoML @ MIT on Oct. 22 and that paper submissions are now open! Students & postdocs w/ accepted papers will be granted FREE admission to attend and have the chance at winning the Octavian-Eugen Ganea Prize for Best Paper! 🧵 #AcademicSky #compchem #drugdesign
Just sharing this widget I threw together. It's been very helpful for visualizing the input data I've been throwing into a model of DNA systems that I've been working on. Data generated from NUPACK. github.com/dfu99/dna-co...
May 15, 2025 at 2:50 PM
Just sharing this widget I threw together. It's been very helpful for visualizing the input data I've been throwing into a model of DNA systems that I've been working on. Data generated from NUPACK. github.com/dfu99/dna-co...
Reposted by Dan
"We'll just terraform Mars," they insist, unable to terraform Earth, a planet that is already perfect except it's like 2° too warm.
May 10, 2025 at 5:40 AM
"We'll just terraform Mars," they insist, unable to terraform Earth, a planet that is already perfect except it's like 2° too warm.
Claude's coding is suddenly horrible. It's been rewriting code artifacts, mixing the logic in each revision, and then offers a bunch of artifacts as variations, but they all only link to the most recently generated artifact.
May 9, 2025 at 8:25 PM
Claude's coding is suddenly horrible. It's been rewriting code artifacts, mixing the logic in each revision, and then offers a bunch of artifacts as variations, but they all only link to the most recently generated artifact.
Reposted by Dan
Sam Gershman writes beautifully about how theory-free neuroscience prevents the field from reaching its promise. Beautiful and true. Most folks do not test hypotheses. Running a NHST does not a hypothesis make. www.thetransmitter.org/theoretical-...
Breaking the barrier between theorists and experimentalists
Many neuroscience students are steeped in an experiment-first style of thinking. Let’s not forget how theory can guide experiments.
www.thetransmitter.org
February 24, 2025 at 6:13 PM
Sam Gershman writes beautifully about how theory-free neuroscience prevents the field from reaching its promise. Beautiful and true. Most folks do not test hypotheses. Running a NHST does not a hypothesis make. www.thetransmitter.org/theoretical-...
Reposted by Dan
This is so cool! ANNs are just like us ;) journals.plos.org/ploscompbiol...
Human-like face pareidolia emerges in deep neural networks optimized for face and object recognition
Author summary Have you ever seen a face in an inanimate object, like the foam in your coffee? This phenomenon, known as face pareidolia, demonstrates our brain’s remarkable ability to detect faces ev...
journals.plos.org
January 31, 2025 at 10:29 PM
This is so cool! ANNs are just like us ;) journals.plos.org/ploscompbiol...
Reposted by Dan
Doing good science is 90% finding a science buddy to constantly talk to about the project.
November 9, 2024 at 10:53 PM
Doing good science is 90% finding a science buddy to constantly talk to about the project.
Reposted by Dan
💯. Most aspects of the system subtly (or not-so-subtly) push you back in your lane, making interdisciplinary work very hard.
Expert's Dilemma: the more specialized you become, the less open you are to creative solutions from other fields. But the more you explore other fields, the more you risk losing credibility in your home field.
Interdisciplinary work is still not really embraced by academia.
Interdisciplinary work is still not really embraced by academia.
January 5, 2025 at 4:56 PM
💯. Most aspects of the system subtly (or not-so-subtly) push you back in your lane, making interdisciplinary work very hard.
Reposted by Dan
One main pain point: interdisciplinary work is often judged by the highest standards in *both* disciplines. This creates an almost impossible burden, effectively extinguishing creative work. Sure, some standards are there for a reason, but more often than not, they are just stock critiques.
💯. Most aspects of the system subtly (or not-so-subtly) push you back in your lane, making interdisciplinary work very hard.
Expert's Dilemma: the more specialized you become, the less open you are to creative solutions from other fields. But the more you explore other fields, the more you risk losing credibility in your home field.
Interdisciplinary work is still not really embraced by academia.
Interdisciplinary work is still not really embraced by academia.
January 5, 2025 at 5:00 PM
One main pain point: interdisciplinary work is often judged by the highest standards in *both* disciplines. This creates an almost impossible burden, effectively extinguishing creative work. Sure, some standards are there for a reason, but more often than not, they are just stock critiques.
Reposted by Dan
I hate the direction lab equipment design has gone in recent years.
I don't want touchscreens, I don't want wifi integration, I don't want centrifuges demanding to know what time zone they're in.
I just want reliable instruments - with physical buttons - that do the one task they were made for.
I don't want touchscreens, I don't want wifi integration, I don't want centrifuges demanding to know what time zone they're in.
I just want reliable instruments - with physical buttons - that do the one task they were made for.
January 2, 2025 at 4:29 PM
I hate the direction lab equipment design has gone in recent years.
I don't want touchscreens, I don't want wifi integration, I don't want centrifuges demanding to know what time zone they're in.
I just want reliable instruments - with physical buttons - that do the one task they were made for.
I don't want touchscreens, I don't want wifi integration, I don't want centrifuges demanding to know what time zone they're in.
I just want reliable instruments - with physical buttons - that do the one task they were made for.
Reposted by Dan
Probing nanomechanics by direct indentation using Nanoendoscopy-AFM reveals the nuclear elasticity transition in cancer cells https://www.biorxiv.org/content/10.1101/2024.12.13.626302v1
December 14, 2024 at 9:51 AM
Probing nanomechanics by direct indentation using Nanoendoscopy-AFM reveals the nuclear elasticity transition in cancer cells https://www.biorxiv.org/content/10.1101/2024.12.13.626302v1