Dive into mathematical frameworks, training/inference techniques, & applications across language, vision-language, & bio domains: huggingface.co/papers/2506.13… Repo: github.com/LiQiiiii/DLLM-…

Great excuse to share something I really love: 1-Lipschitz nets. They give clean theory, certs for robustness, the right loss for W-GANs, even nicer grads for explainability!! Yet are still niche. Here’s a speed-run through some of my favorite papers on the field. 🧵👇

🚨 The era of infinite internet data is ending, So we ask: 👉 What’s the right generative modelling objective when data—not compute—is the bottleneck? TL;DR: ▶️Compute-constrained? Train Autoregressive models ▶️Data-constrained? Train Diffusion models Get ready for 🤿 1/n

"Deep Researcher with Test-Time Diffusion" This paper treats report writing as an iterative retrieval‑augmented diffusion process that can be enhanced by component‑wise self‑evolution. This demonstrates SoTA on multi‑hop search‑and‑reasoning benchmarks.

Our 500+ page AI4Science paper is finally published: Artificial Intelligence for Science in Quantum, Atomistic, and Continuum Systems. Foundations and Trends® in Machine Learning, Vol. 18, No. 4, 385–912, 2025 nowpublishers.com/article/Detail…

Coupling all-electron full-potential density functional theory with grid-based continuum embeddings. arxiv.org/abs/2507.17672

From Atoms to Dynamics: Learning the Committor Without Collective Variables. arxiv.org/abs/2507.17700

Code release! 🚀 Following up on our IMO 2025 results with the public LLM Gemini 2.5 Pro — here’s the full pipeline & general (non-problem-specific) prompts. 👉 [github.com/lyang36/IMO25] Have fun exploring! #AI #Math #LLMs #IMO2025

Interested in some foundation aspects? Waiting or unhappy about NeurIPS reviews? Plz consider NeurIPS workshop DynaFront: Dynamics at the Frontiers of Optimization, Sampling, and Games sites.google.com/view/dynafront… Yuejie Chi Andrea Montanari Taiji Suzuki Tatjana Chavdarova ++ Sponsor appreciated!

We're delighted to invite submissions to the #NeurIPS2025 workshop "Tackling Climate Change with Machine Learning". Important dates: ▶️Mentorship program: Jul 28 ▶️Papers, proposal & tutorial submissions: Aug 20 ▶️Workshop: Dec 6 or 7 Learn more: climatechange.ai/events/neurips…

📢 Excited to announce that GenMol is now open-sourced. GenMol: A Drug Discovery Generalist with Discrete Diffusion Paper: arxiv.org/abs/2501.06158 Code: github.com/NVIDIA-Digital…

Anyone knows adam?

Proud to introduce Group Sequence Policy Optimization (GSPO), our stable, efficient, and performant RL algorithm that powers the large-scale RL training of the latest Qwen3 models (Instruct, Coder, Thinking) 🚀 📄 huggingface.co/papers/2507.18…

Atomistic Generative Diffusion for Materials Modeling. arxiv.org/abs/2507.18314

"Group Sequence Policy Optimization" Qwen introduces Group Sequence Policy Optimization (GSPO), a RL algorithm for LLMs that uses sequence-level importance weighting and clipping instead of token-level methods, which provides high performing training for MoE models.

Anthropic's recent blog post provides a good summary of some of the recent work applying mechanistic interpretability to biology. I am very bullish on this research direction!

Phase 1 of Physics of Language Models code release ✅our Part 3.1 + 4.1 = all you need to pretrain strong 8B base model in 42k GPU-hours ✅Canon layers = strong, scalable gains ✅Real open-source (data/train/weights) ✅Apache 2.0 license (commercial ok!) 🔗github.com/facebookresear…

Yep, TRPO ignores the mismatch of state distribution. (and the 2002 reference shows that this is a classic approximation). GSPO just says: why we need to ignore it?

Enjoyed reading this piece of writing: statmodeling.stat.columbia.edu/2025/07/11/opp…

Higher order calculations in perturbation theory quickly become complex and difficult, but progress is being made. This highlight uses modern techniques from computational algebraic geometry to analytically compute five-point three-loop diagrams. #EdSugg journals.aps.org/prd/abstract/1…