Basics of Deep RL tutorial I am still very happy with, as good a day as any to re-post :) youtube.com/playlist?list=…

Very happy to share the almost-final version of our new online control book: arxiv.org/abs/2211.09619

We present InverseBench, a framework for benchmarking plug-and-play diffusion approaches for inverse problems in physical sciences. (#ICLR2025 Spotlight) PnP diffusion approaches are attractive for scientific inverse problems because they offer flexibility in incorporating

Zero-shot forecasting of chaotic systems. arxiv.org/abs/2409.15771

I'm happy to announce that v2 of my RL tutorial is now online. I added a new chapter on multi-agent RL, and improved the sections on 'RL as inference' and 'RL+LLMs' (although latter is still WIP), fixed some typos, etc. arxiv.org/abs/2412.05265…

the best researchers from Meta, Yale, Stanford, Google DeepMind, and Microsoft laid out all we know about Agents in a 264-page paper [book], here are some of their key findings:

#L4DC2025 will feature 4 oral presentation sessions on System Identification, Safe Learning, Reinforcement Learning, and Learning-Based Control; and 4 poster sessions with 120 posters. Register here: tinyurl.com/583u2wrz Early bird due: May 2 Student travel grant due: Apr 21

Excited about this work with Asher Trockman Yash Savani (and others) on antidistillation sampling. It uses a nifty trick to efficiently generate samples that makes student models _worse_ when you train on samples. I spoke about it at Simons this past week. Links below.

🚀We have two exciting works at ICLR! 1) Oral: Behavioral Economics in Decision Making within Robust Multi-Agent Systems — Computationally Tractable Equilibrium in Games! 2) Poster: Robust Gymnasium — A Module Benchmark for Robust Reinforcement Learning with Over 60 Tasks!

Aligned Multi-Objective Optimization (A-🐮) has been accepted at #ICML2025! 🎉 We explore optimization scenarios where objectives align rather than conflict, introducing new scalable algorithms with theoretical guarantees. #MachineLearning #AIResearch #Optimization #MLCommunity

Harvard’s AI Research Experience free course book by Pranav Rajpurkar covers the essentials and tips on doing research: - VSCode, Git, Conda - PyTorch, W&B - AWS, colab - LLMs and VLMs - reading AI papers - research progress and organization this is a must read!

Introducing your arXiv Research Agent A personal research assistant with access to arXiv + bioRxiv + medRxiv + Semantic Scholar. Upload drafts, conduct literature reviews, get insights across millions of papers MCP support coming soon 🚀

New lecture recordings on RL+LLM! 📺 This spring, I gave a lecture series titled **Reinforcement Learning of Large Language Models**. I have decided to re-record these lectures and share them on YouTube. (1/7)

We’re now accepting applications for the 2026–2027 NVIDIA Graduate Fellowships! If you’re passionate about advancing cutting-edge reasoning models for Physical AI applications 🚗🤖, apply here: research.nvidia.com/graduate-fello… — and be sure to select “Autonomous Vehicles.” NVIDIA AI

Introducing GitHub star rankings for arXiv papers We indexed every arXiv paper's codebase so you can instantly filter by implementation quality Find the most popular, production-ready research in seconds 🚀

'Data-Driven Performance Guarantees for Classical and Learned Optimizers', by Rajiv Sambharya, Bartolomeo Stellato. jmlr.org/papers/v26/24-… #optimizers #optimization #minimize

We found a troubling emergent behavior in LLM. 💬When LLMs compete for social media likes, they start making things up 🗳️When they compete for votes, they turn inflammatory/populist When optimized for audiences, LLMs inadvertently become misaligned—we call this Moloch’s Bargain

Notion 的创始人 Ivan Zhao 的精彩文章：《钢铁、蒸汽机与无限大脑》 越来越清晰的感觉到，我们正处于 AI 革命的早期阶段，对于未来谁也不知道会怎么样，所以都喜欢从历史中、去工业革命、互联网革命中寻找规律，以期望能对未来有所启发。

Claude Code 是我今年最推荐的 AI 产品，没有之一。 Claude Code 虽然叫 Code，但它的功能绝对不止是写代码，而是一款真正意义上的通用 Agent。 很多电脑上的很繁琐的工作，都是 CC 一句话的事情。 我用它做问答、写作、写网页、写软件，数据分析、甚至拆分工资条。

It's a progression from our AdvPrompter paper (from ICML 2025), which explores attacks and does SFT-based alignment. Our presentation is available here: bamos.github.io/presentations/… And 🧵: x.com/brandondamos/s…