Nuno Loureiro, professor and director of MIT’s Plasma Science and Fusion Center, dies at 47 In his 10 years at MIT, Loureiro helped illuminate the physics occurring at the center of fusion vacuum chambers and at the edges of the universe. news.mit.edu/2025/nuno-lo...

Donating the Model Context Protocol and establishing the Agentic AI Foundation www.anthropic.com/news/donatin...

Titans + MIRAS: Helping AI have long-term memory share.google/YCYv9OoHPoHR...

Enduring patterns in world's languages: One-third of grammatical 'universals' stand up to rigorous testing share.google/YggCLd1m2vzn...

OpenTSLM: Time-Series Language Models for Reasoning over Multivariate Medical Text- and Time-Series Data Stanford BDHG GitHub share.google/pfh467ea3XmK...

Genomic divergence across the tree of life 🧪🌐https://www.pnas.org/doi/10.1073/pnas.2319389122

NLD approaches for characterizing multi-sector climate impacts under irreducible uncertainty share.google/MOVW6RNPmbpl...

Clearing significant hurdle to quantum computing — Harvard Gazette Harvard physicists working to develop game-changing tech demonstrate 3,000 quantum-bit system capable of continuous operation share.google/nrKNRYy2UYH6...

Parallel-R1: Towards Parallel Thinking via Reinforcement Learning arxiv.org/abs/2509.07980

Ultrabroadband on-chip photonics for full-spectrum wireless communications www.nature.com/articles/s41...

Graph theory in state space Moduli space youtu.be/YGLNyHd2w10?...

Ergodicity detection algorithms: Scaling of ergodicity in random symbolic dynamics arxiv.org/html/2508.08...

Relay-BP: The world’s fastest, most accurate decoder for qLDPC error correction codes www.ibm.com/quantum/blog...

Pure quantum state without the need for cooling - ETH Zurich ethz.ch/en/news-and-...

as exemplified by the success of convolutional networks in the image domain. arxiv.org/abs/2406.06248

How Attention Sinks Keep Language Models Stable hanlab.mit.edu/blog/streami...

On the Generalization of SFT: A Reinforcement Learning Perspective with Reward Rectification arxiv.org/abs/2508.05629

Existence of smooth solutions of the Navier-Stokes equations in three-dimensional Euclidean space arxiv.org/abs/2507.18063

Persona Vectors: Monitoring and Controlling Character Traits in Language Models arxiv.org/abs/2507.21509

Hierarchical Reasoning Model arxiv.org/abs/2506.21734

Persona Vectors brb 👀👀👀👀👀👀 Anthropic just dropped this paper. They can steer models quite effectively, and even detect training data that elicits a certain (e.g. evil) persona arxiv.org/abs/2507.21509

RiemannLoRA: A Unified Riemannian Framework for Ambiguity-Free LoRA Optimization share.google/UupYlhvHZwz4...

NIST and Partners Use Quantum Mechanics to Make a Factory for Random Numbers www.nist.gov/news-events/...

NIST Offers 19 Ways to Build Zero Trust Architectures www.nist.gov/news-events/...

#newpaper from my QEC colleagues. Check this out: arxiv.org/abs/2506.14688 #qec #quantum #quantumcomputing #quantinuum

AlphaOne: Reasoning Models Thinking Slow and Fast at Test Time arxiv.org/abs/2505.24863

RareFold: Structure prediction and design of proteins with noncanonical amino acids www.biorxiv.org/content/10.1...

This is what a Cold Air Front looks like from Space

Physicists confirm the fascinating existence of "second sound" Source: Earth.com search.app?link=https:/... via the Google news

Input-Driven Plasticity (IDP) model aims to address the lack of clarity with a mechanism that gradually integrates past and new information, guiding the memory retrieval process to the correct memory techxplore.com/news/2025-05...

Meta Introduces KernelLLM: An 8B LLM that Translates PyTorch Modules into Efficient Triton GPU Kernels Trained on ~25K PyTorch-Triton pairs, it simplifies GPU programming by generating optimized kernels Article: www.marktechpost.com/2025/05/20/m... Hugging Face: huggingface.co/facebook/Ker...

The Open Molecules 2025 effort required 6 billion compute hours to create Meta released a massive trove of chemistry data to supercharge scientific research o Built an #AI model that can speed up the time it takes to create new drugs and materials www.semafor.com/article/05/1... #HPC #LBNL

Google DeepMind Introduces AlphaEvolve: A Gemini-Powered Coding AI Agent for Algorithm Discovery and Scientific Optimization www.marktechpost.com/2025/05/14/g...

MIT engineers advance toward a fault-tolerant quantum computer Researchers achieved a type of coupling between artificial atoms and photons that could enable readout and processing of quantum information in a few nanoseconds news.mit.edu/2025/mit-eng...

Wiring of the brain during development is like a dance where neurons try different partners, then choose the perfect one. Fekrije Selimi explains how the parallel fibers and the climbing fibers find their Purkinje cell partners in the cerebellum.

Scalar fields: the secret sauce of theoretical physics Ever wonder what the real deal is behind scalars, vectors, and tensors? You just might understand what they are after reading this. (And no, a tensor isn't "something that transforms like a tensor.") bigthink.com/starts-with-...

What is MoE 2.0? Update Your Knowledge about Mixture-of-experts huggingface.co/blog/Ksenias...

I-Con: A Unifying Framework for Representation Learning Introduces a single equation that unifies >20 machine learning methods into a periodic table. This framework can be use to make a state-of-the-art unsupervised image classifier mhamilton.net/icon#:~:text....

Beating the Ramsey limit on sensing with deterministic qubit control a new protocol to enhance the sensitivity of a qubit frequency measurement in the presence of decoherence by applying a continuous drive to stabilize one component of the Bloch vector. www.nature.com/articles/s41...

Designing a new way to optimize complex coordinated systems Using diagrams to represent interactions in multipart systems can provide a faster way to design software improvements. news.mit.edu/2025/designi...

Probabilistic approach to the problem of network alignment, as well as the corresponding inference algorithms. www.nature.com/articles/s41...

How to find the hypergraphs underlying dynamical systems Scientists usually use a hypergraph model to predict dynamic behaviors. But the opposite problem is interesting, too. www.santafe.edu/news-center/...

just gonna leave this here. www.nature.com/articles/s41...

Researchers at Pacific Northwest National Laboratory have developed a new algorithm, Picasso, that reduces quantum data preparation time by 85%, addressing a key bottleneck in hybrid quantum-classical computing. thequantuminsider.com/2025/04/24/s...

Loving this poster session energy! #ICLR2025

Holy shit. The National Science Foundation (NSF) has halted the awarding of previously approved proposals to ensure compliance with presidential directives following the arrival of DOGE at the science funding agency.

Computer vision researcher Suguru Onda just had his US visa revoked. x.com/adamsmallksl...

Reading the transcript of Princeton's president Eisengruber on the Daily and what a cogent, powerful, and brave defense of #academic #freedom. @nytimes.com www.nytimes.com/2025/04/09/p...? A few highlights, but it's worth a read.

Great example of generate -> validate loop + error analysis > "the most effective route to improve outcomes was brute force: retry steps until they passed or reached a limit. We give the validation errors ... to the LLM and built a loop runner"

“Neither Harvard nor any other private university can allow itself to be taken over by the federal government,” Harvard attorneys wrote to federal officials. “Accordingly, Harvard will not accept the government’s terms.” If you’ve never given to Harvard, GIVE NOW! search.app/M76p3GzTmMuc...