Most AI researchers I talk to have been a bit shocked by DeepSeek-R1 and its performance. My preliminary understanding nuggets: 1. Simple post-training recipe called GRPO: Start with a good model and reward for correctness and style outcomes. No PRM, no MCTS no fancy reward

If you’re using the #ICML LaTeX template, there’s a typo in algorithmic.sty that prevents cross-referencing specific lines in the algorithm environment. The fix is simple: change \addtocounter{ALC@line}{1} to \refstepcounter{ALC@line} on Line 106. Credit: tex.stackexchange.com/questions/5234…

Today, we are publishing the first-ever International AI Safety Report, backed by 30 countries and the OECD, UN, and EU. It summarises the state of the science on AI capabilities and risks, and how to mitigate those risks. 🧵 Link to full Report: assets.publishing.service.gov.uk/media/679a0c48… 1/16

ICML DDL is over, but don’t forget about the ICLR 2025 Workshop on Deep Generative Models submission deadline coming up fast! Share your innovative work: delta-workshop.github.io #ICLR2025 #DeepGenerativeModels #ICLR2025Workshop #CallForPapers

ICML2025のDeep Generative Modelワークショップの締め切りが2月5日に迫ってきました．ぜひ投稿ください． Submission deadline: February 5 (AOE), 2025 delta-workshop.github.io

🎉 Thrilled that our paper "On the Optimization and Generalization of Two-layer Transformers with Sign Gradient Descent" is a Spotlight at #ICLR2025! Huge thanks to my collaborators & reviewers! Excited to discuss at the conference! 📄 Paper: openreview.net/forum?id=97rOQ…

Separate from the recently announced Special Postdoctoral Researcher recruitment at RIKEN, the Robot Learning Team at RIKEN AIP RIKEN Center for Advanced Intelligence Project is seeking to hire two researchers! If you are interested, please consider applying. riken.jp/en/careers/res…

グリーン・デイ 来日ライブ横浜Kアリーナ Day 2 セトリ最後まさかの観客がビリーのギターをステージで譲り受けてGood Riddanceを共演 若い日本の子がGreen Dayの曲を完璧に弾き継いでこんな幸せなファイナルある？ #greenday #グリーンデイ #setlist Green Day - Good Riddance (Time of Your Life)

Excited to announce our seminar! Join us on Mar 12, 2025, 13:00–14:30 (JST) for a hybrid talk by Prof. Difan Zou (HKU) on "Transformers: Model Depth & Attn" In-person at RIKEN AIP Nihonbashi Office & online via Zoom. Register: …c59ed978213830355fc8978.doorkeeper.jp/events/181888 #DeepLearning #Transformers

ICML 2025's rebuttal process be like🤣: 👨‍💻 Authors: spend a whole week writing a careful rebuttal ✅ Reviewer: clicks "acknowledge" without reading 🚫 Author: not allowed to reply anymore So what does acknowledge mean here? "You speak. I pretend to listen. Conversation over."🙃

Wow, Deeply Supervised Nets received the Test of Time award at AISTATS Conference 2025! It was the very first paper I submitted during my PhD. Fun fact: the paper was originally rejected by NeurIPS with scores of 8/8/7 (yes, that pain stuck with me... maybe now I can finally let it

I want to share my latest (very short) blog post: "Active Learning vs. Data Filtering: Selection vs. Rejection." What is the fundamental difference between active learning and data filtering? Well, obviously, the difference is that: 1/11

🚀【Deep Learning Theory Team Seminar】 🎙️ Talk by Prof. Wuyang Chen (SFU): Building Machines That Understand the Physics 🧠 AI meets scientific tools & physics-enriched data 📅 May 28, 15:00 JST 🔗 Details & RSVP: …c59ed978213830355fc8978.doorkeeper.jp/events/184833 #AI #DeepLearning #ScientificML #LLM

AI benchmarking culture is completely out of control. Tables with dozens of methods, datasets, and bold numbers, trying to answer a question that perhaps no one should be asking anymore.

🧵New survey: Bridging Distribution Shift and AI Safety Distribution shift and AI safety have long been studied in parallel. But how can their insights formally inform each other? We present the first comprehensive, mathematically grounded, and one-to-one aligned treatment. 1/6

Ever confused by "prompt tuning" vs "model reprogramming" vs "in-context learning"? What if they're all the same thing—just different names across ML, CV, and NLP communities? Our recent paper introduces Neural Network Reprogrammability as a unifying framework showing these

Anyone knows adam?

Announcing the ICLR 2026 Call for Papers! Abstract submission: Sept 19 (AoE) Paper submission: Sept 24 (AoE) Reviews released: Nov 11 Author/Reviewer discussion: Nov 11-Dec 3 Final decisions: Jan 22 2026 iclr.cc/Conferences/20…

Another step for China’s AI innovation! Try it yourself: AutoSurvey: github.com/AutoSurveys/Au… GLM4.5: z.ai/blog/glm-4.5 Kimi: kimi-k2.com #GLM4.5 #kimi2 #AIResearch #ZhipuAI #AutoSurvey

RLVR/RLHF libraries: • verl - ByteDance • TRL - HuggingFace • slime - Zhipu AI • prime-rl - Prime Intellect • ROLL - Alibaba • Nemo-RL - NVIDIA • AReaL - Ant Research • SkyRL - UC Berkeley • open-instruct - Allen AI • torchtune - PyTorch Any I am missing? Which do you