Our recent work: academic.oup.com/bioinformatics…, accepted at ISMB 2024, introduces Truffle - a method that enables accurate discovery of endotypes from clinical trials data, thus helping advance precision medicine. #ISMB #clinicaltrials #precisionmedicine

Really important research! Every year of private renting is associated with an extra 2.4 weeks of biological aging Renting has a bigger impact on premature aging than being unemployed, obesity and being a former smoker jech.bmj.com/content/78/1/40

Data generation through simulation is one reason why the whole idea of ML for science is so exciting.

🚀 The Segment Anything Model (SAM) has been upgraded to SAM2, featuring an efficient image encoder for segmenting images and videos. But does SAM2 outperform SAM1 in medical image and video segmentation? We're thrilled to present our paper "Segment Anything in Medical Images

Today we release a new model series for math-specific language models, Qwen2-Math, which is based on Qwen2. The flagship model, Qwen2-Math-72B-Instruct, outperforms proprietary models, including GPT-4o and Claude 3.5, in math related downstream tasks! Feel free to check our blog

Check out NotebookLM! Create a notebook, upload one or more sources (e.g. PDFs of research papers, your favorite PhD thesis, a newspaper article, etc) then click on 'Generate' to create a podcast of two voices talking about the content you've uploaded. blog.google/technology/ai/…

Excited to share our latest research, “Recovering Time-Varying Networks from Single-Cell Data”, now available on arXiv! In this work, we introduce Marlene, a framework designed to infer dynamic gene regulatory networks from scRNA-seq data. arxiv.org/abs/2410.01853

Now that Hopfield and Hinton won the Nobel Prize, Boltzmann machines seem to be interesting again, and I want to share this project I did with Ezra Winston and Zico Kolter (who wrote most of the hard code) Monotone deep Boltzmann machines. (1/4) Paper: arxiv.org/pdf/2307.04990

Inference with VLMs is costly, thanks to 500+ image tokens. So… should you use a smaller model or run a bigger model on fewer tokens? Our 📢new findings left me genuinely amazed: processing just ONE compressed visual token with the largest LLM gives compute-optimal inference.🧵

🧵 Are "medical" LLMs/VLMs *adapted* from general-domain models, always better at answering medical questions than the original models? In our oral presentation at #EMNLP2024 today (2:30pm in Tuttle), we'll show that surprisingly, the answer is "no". arxiv.org/abs/2411.04118

Contrastive VLMs (CLIP) lack the structure of text embeddings, like satisfying analogies via arithmetic (king - man = queen). We enhance CLIP’s *reasoning abilities* on such tasks by finetuning w/ text descriptions of image differences! w/ D. Willmott, J.Semedo, Zico Kolter 1/🧵

Just in time for your holiday reading!! Our senescent cell signature of the lung is out TriState SenNet TMC SenNet Consortium Euxhen Hasanaj PittDeptofMed. SenSet, a novel human lung senescence cell gene signature, identifies cell-specific senescence mechanisms biorxiv.org/content/10.110…

We have just released SenSet, a novel list of 106 senescence marker genes. We hope this resource accelerates discoveries in aging research, cancer biology, and regenerative medicine. #senescence #aging #pulearning #gene-set #SenNet biorxiv.org/content/10.110…

To trust LLMs in deployment (e.g., agentic frameworks or for generating synthetic data), we should predict how well they will perform. Our paper shows that we can do this by simply asking black-box models multiple follow-up questions! w/ Marc Finzi and Zico Kolter 1/ 🧵

The Illusion of Thinking [1] has been generating a lot of commotion recently, and for good reason. It's an important study showing how LLMs struggle with high-complexity puzzles. But its focus on perfect step-by-step execution may offer too narrow a view of reasoning. Strategy

We will be presenting our paper on Multimodal Benchmarking of Foundation Model Representations for Cellular Perturbation Response Prediction at two #ICML2025 Workshops this week: FM4LS (fm4ls.github.io) and Generative AI and Biology (genbio-workshop.github.io/2025/). We

🤔 What if protein design models knew where they’re uncertain and adapted their predictions accordingly? 🚀 We introduce uncertainty-aware discrete diffusion for structure-conditioned protein design: doi.org/10.1101/2025.0…. 📍Catch us at #ICML2025 FM4LS workshop this week! 1/4

Good software should be fast, reliable, reusable, and maintainable. A lot of BioML benchmarking is uh… not. But biology doesn’t standardize to a few data types like language, audio, or images. We’re constantly inventing new ways to measure life... 1/n

1/ 🚀 Introducing AIDO.StructureDiffusion: A generative model for structural protein design—enabling high-quality, controllable generation of monomers, complexes, and antibodies. 🧵

Excited to be at #ISMB2025 in Liverpool, UK this week! If you’re around, I’d love to connect and chat about foundation models, networks, or anything in between. I’ll be presenting my work on Recovering Time-Varying Networks from Single-Cell Data; come say hi if that’s up your