Datamase

Article Overviews

Every "Article Overview" post from Datamase — short summaries of papers, model releases, and technical reports worth knowing about. Search to find a specific one.

65 article overviews

Kimi K3: Open Frontier Intelligence

Kimi K3: Open Frontier Intelligence

Summary of “Kimi K3: Open Frontier Intelligence”, by researchers at Kimi

JoyAI-VL-Interaction: Real-Time Vision-Language Interaction Intelligence

JoyAI-VL-Interaction: Real-Time Vision-Language Interaction Intelligence

Summary of “JoyAI-VL-Interaction: Real-Time Vision-Language Interaction Intelligence”, by researchers at Jingdong Group

GigaWorld-1: A Roadmap to Build World Models for Robot Policy Evaluation

GigaWorld-1: A Roadmap to Build World Models for Robot Policy Evaluation

Summary of “GigaWorld-1: A Roadmap to Build World Models for Robot Policy Evaluation”, by GigaAI, Tsinghua University

Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs

Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs

Summary of “Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs”, by researchers from Yale University and Google Research

Scaling the Horizon, Not the Parameters: Reaching Trillion-Parameter Performance with a 35B Agent

Scaling the Horizon, Not the Parameters: Reaching Trillion-Parameter Performance with a 35B Agent

Summary of “Scaling the Horizon, Not the Parameters: Reaching Trillion-Parameter Performance with a 35B Agent”, by the Agents-A1 Team, Shanghai Artificial Intelligence Laboratory

ResearchStudio-Idea: An Evidence-Grounded Research-Ideation Skill Suite from ML Conference Outcomes

ResearchStudio-Idea: An Evidence-Grounded Research-Ideation Skill Suite from ML Conference Outcomes

Summary of “ResearchStudio-Idea: An Evidence-Grounded Research-Ideation Skill Suite from ML Conference Outcomes”, by Researchers from Microsoft Research and affiliated universities

Embodied.cpp: A Portable Inference Runtime of Embodied AI Models on Heterogeneous Robots

Embodied.cpp: A Portable Inference Runtime of Embodied AI Models on Heterogeneous Robots

Summary of “Embodied.cpp: A Portable Inference Runtime of Embodied AI Models on Heterogeneous Robots”, by Researchers from Microsoft Research and affiliated universities

AI-Trader: Benchmarking Autonomous Agents in Real-Time Financial Markets

AI-Trader: Benchmarking Autonomous Agents in Real-Time Financial Markets

Summary of “AI-Trader: Benchmarking Autonomous Agents in Real-Time Financial Markets”, by Researchers from the University of Hong Kong

Harness Engineering for Self-Improvement

Harness Engineering for Self-Improvement

Summary of “Harness Engineering for Self-Improvement”, by Lilian Weng

Autoresearch: The feedback loop behind self-improving agents

Autoresearch: The feedback loop behind self-improving agents

Summary of “Autoresearch: The feedback loop behind self-improving agents”, an interview of Roland Gavrilescu by Latent Space

The Twilight of the Chatbots

The Twilight of the Chatbots

Summary of “The Twilight of the Chatbots” by Ethan Mollick

The AI All-You-Can-Eat Buffet is Ending

The AI All-You-Can-Eat Buffet is Ending

Summary of “The AI All-You-Can-Eat Buffet is Ending” from an interview of Gary Marcus by Steve Eisman

A New Look at AI's Impact on Jobs

A New Look at AI's Impact on Jobs

Summary of “A New Look at AI's Impact on Jobs” by researchers from Ramp and Revelio Labs

Accurate Decoding of Natural Sentences from Non-Invasive Brain Recordings

Accurate Decoding of Natural Sentences from Non-Invasive Brain Recordings

Summary of “Accurate Decoding of Natural Sentences from Non-Invasive Brain Recordings” by researchers from Meta and affiliated universities

RoboReward: General-Purpose Vision-Language Reward Models for Robotics

RoboReward: General-Purpose Vision-Language Reward Models for Robotics

Summary of “RoboReward: General-Purpose Vision-Language Reward Models for Robotics” by researchers from Stanford and UCBerkeley

Residual Context Diffusion Language Models

Residual Context Diffusion Language Models

Summary of “Residual Context Diffusion Language Models” by researchers from UCBerkeley and Apple

Qwen-AgentWorld: Language World Models for General Agents

Qwen-AgentWorld: Language World Models for General Agents

Summary of “Qwen-AgentWorld: Language World Models for General Agents” by the Qwen Team

PorTAL: Portal Task Adapters for LLMs

PorTAL: Portal Task Adapters for LLMs

Summary of “PorTAL: Portal Task Adapters for LLMs” by Ben Geist, Ramp Labs

Introducing LongCat-2.0

Introducing LongCat-2.0

Summary of “Introducing LongCat-2.0” by Researchers from Meituan

Can Scale Save Us From Plasticity Loss in Large Language Models?

Can Scale Save Us From Plasticity Loss in Large Language Models?

Summary of “Can Scale Save Us From Plasticity Loss in Large Language Models?” by Researchers from Zyphra

LFM2.5-230M: Built to Run Anywhere

LFM2.5-230M: Built to Run Anywhere

Summary of “LFM2.5-230M: Built to Run Anywhere” by Researchers from Liquid AI

DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence

DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence

Summary of “DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence” by Researchers from DeepSeek-AI

Arctic RL: A Unified, Open Source RL Backend for Enterprise Post-Training

Arctic RL: A Unified, Open Source RL Backend for Enterprise Post-Training

Summary of “Arctic RL: A Unified, Open Source RL Backend for Enterprise Post-Training” by Researchers from Snowflake AI Research

Return on Tokens (ROT)

Return on Tokens (ROT)

Summary of "Return on Tokens (ROT)" by Packy McCormick and Markie Wagner

First steps Towards Automated AI Research

First steps Towards Automated AI Research

Summary of "First steps Towards Automated AI Research" by Researchers from Recursive AI

Mistral OCR 4 Technical Report

Mistral OCR 4 Technical Report

Summary of "Mistral OCR 4 Technical Report" by Researchers from Mistral AI

Laguna M.1/XS.2 Technical Report

Laguna M.1/XS.2 Technical Report

Summary of "Laguna M.1/XS.2 Technical Report" by Researchers from Poolside AI

Krea 2 Technical Report

Krea 2 Technical Report

Summary of “Krea 2 Technical Report" by Researchers from Krea AI

Reinforcement learning towards broadly and persistently beneficial models

Reinforcement learning towards broadly and persistently beneficial models

Summary of “Reinforcement learning towards broadly and persistently beneficial models" by Researchers from OpenAI

LifeSciBench: Evaluating Language Models on Realistic, Expert-Level Tasks in the Life Sciences

LifeSciBench: Evaluating Language Models on Realistic, Expert-Level Tasks in the Life Sciences

Summary of “LifeSciBench: Evaluating Language Models on Realistic, Expert-Level Tasks in the Life Sciences" by Researchers from OpenAI and Tacit Labs

AtomMem: Building Simple and Effective Memory System for LLM Agents via Atomic Facts

AtomMem: Building Simple and Effective Memory System for LLM Agents via Atomic Facts

Summary of “AtomMem: Building Simple and Effective Memory System for LLM Agents via Atomic Facts" by Researchers from USTC and Anhui University

How Long Until AI Doesn't Need Humans?

How Long Until AI Doesn't Need Humans?

Summary of “How Long Until AI Doesn't Need Humans?" by Ajeya Cotra and Timothy B. Lee

Is AI ruining our skills? Early results are in — and they’re not good

Is AI ruining our skills? Early results are in — and they’re not good

Summary of “Is AI ruining our skills? Early results are in — and they’re not good" by Mariana Lenharo at Nature

AI Systems Out-Persuade Expert Humans

AI Systems Out-Persuade Expert Humans

Summary of “AI Systems Out-Persuade Expert Humans" by Researchers at Oxford, UKAI, Stanford, and LSE

GDM AI Control Roadmap

GDM AI Control Roadmap

Summary of “GDM AI Control Roadmap" by Researchers at Google DeepMind

VibeThinker-3B: Exploring the Frontier of Verifiable Reasoning in Small Language Models

VibeThinker-3B: Exploring the Frontier of Verifiable Reasoning in Small Language Models

Summary of “VibeThinker-3B: Exploring the Frontier of Verifiable Reasoning in Small Language Models" by Researchers at Sina Weibo Inc.

The Untrainable

The Untrainable

Summary of “The Untrainable" by Sarah Guo

Olmo-Eval: An evaluation workbench for the model development loop

Olmo-Eval: An evaluation workbench for the model development loop

Summary of “Olmo-Eval: An evaluation workbench for the model development loop" by Researchers from Ai2

Agentic coding and persistent returns to expertise

Agentic coding and persistent returns to expertise

Summary of “Agentic coding and persistent returns to expertise" by Researchers from Anthropic

Dialogues with AI Reduce Beliefs in Misinformation but Build No Lasting Discernment Skills

Dialogues with AI Reduce Beliefs in Misinformation but Build No Lasting Discernment Skills

Summary of “Dialogues with AI Reduce Beliefs in Misinformation but Build No Lasting Discernment Skills" by Researchers at MIT

Qwen-RobotWorld Technical Report: Unifying Embodied World Modeling through Language-Conditioned Video Generation

Qwen-RobotWorld Technical Report: Unifying Embodied World Modeling through Language-Conditioned Video Generation

Summary of “Qwen-RobotWorld Technical Report: Unifying Embodied World Modeling through Language-Condition" by Researchers from the Qwen Team

MolmoMotion: Forecasting Point Trajectories in 3D with Languge Instruction

MolmoMotion: Forecasting Point Trajectories in 3D with Languge Instruction

Summary of “MolmoMotion: Forecasting Point Trajectories in 3D with Language Instruction" by Researchers from the Allen Institute for AI, University of Washington, and UNC-Chapel Hill

Loop Engineering

Loop Engineering

Summary of “Loop Engineering" by Addy Osmani

IndexCache: Accelerating Sparse Attention via Cross-Layer Index Reuse

IndexCache: Accelerating Sparse Attention via Cross-Layer Index Reuse

Summary of “IndexCache: Accelerating Sparse Attention via Cross-Layer Index Reuse”, by Researchers from Tsinghua University and Z.ai

AI demands more engineering discipline. Not less

AI demands more engineering discipline. Not less

Summary of “AI demands more engineering discipline. Not less”, by Charity Majors

From AGI to ASI

From AGI to ASI

Summary of “AGI to ASI”, by researchers from Google DeepMind, the University of Waterloo, Australian National University, and University College London

Molmo2: Open Weights and Data for Vision-Language Models with Video Understanding and Grounding

Molmo2: Open Weights and Data for Vision-Language Models with Video Understanding and Grounding

Summary of “Molmo2: Open Weights and Data for Vision-Language Models with Video Understanding and Grounding”, by researchers from Allen Institute for AI and University of Washington

ParseBench: A Document Parsing Benchmark for AI Agents

ParseBench: A Document Parsing Benchmark for AI Agents

Summary of “ParseBench: A Document Parsing Benchmark for AI Agents”, by researchers from LlamaIndex

Large Language Models Hack Rewards, and Society

Large Language Models Hack Rewards, and Society

Summary of “Large Language Models Hack Rewards, and Society”, by researchers from King’s College London, Fudan University, and the Alan Turing Institute

AI Dark Output: The Visible Cost of Invisible Output

AI Dark Output: The Visible Cost of Invisible Output

Summary of “AI Dark Output: The Visible Cost of Invisible Output”, by Malcolm Spittler and Dylan Patel

Democratic Governance of Frontier AI: A Blueprint for a Federal Framework

Democratic Governance of Frontier AI: A Blueprint for a Federal Framework

Summary of "Democratic Governance of Frontier AI: A Blueprint for a Federal Framework" by researchers at OpenAI

CubePart: An Open-Vocabulary Part-Controllable 3D Generator

CubePart: An Open-Vocabulary Part-Controllable 3D Generator

Summary of “CubePart: An Open-Vocabulary Part-Controllable 3D Generator" by researchers at Roblox, Stanford, and Carnegie Mellon University

MiMo-V2.5-Pro-UltraSpeed: Pushing 1T-Parameter Model Generation Speed to 1000 TPS

MiMo-V2.5-Pro-UltraSpeed: Pushing 1T-Parameter Model Generation Speed to 1000 TPS

Summary of “MiMo-V2.5-Pro-UltraSpeed: Pushing 1T-Parameter Model Generation Speed to 1000 TPS”, by researchers from Xiaomi

Introducing North Mini Code: Cohere’s First Model For Developers

Introducing North Mini Code: Cohere’s First Model For Developers

Summary of “Introducing North Mini Code: Cohere’s First Model For Developers”, by the Cohere Code Agents Team

Releasing the MisoTTS

Releasing the MisoTTS

Summary of “Releasing the MisoTTS”, by Aoden Teo & Cassidy Dalva, Miso Labs

Language Models Need Sleep: Learning to Self-Modify and Consolidate Memories

Language Models Need Sleep: Learning to Self-Modify and Consolidate Memories

Summary of “Language Models Need Sleep: Learning to Self-Modify and Consolidate Memories", by researchers from Google Research and Cornell University

Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning

Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning

Summary of “Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning" by researchers at NVIDIA

When AI Builds Itself

When AI Builds Itself

Summary of “When AI Builds Itself”, by researchers from Anthropic

GPIC: A Giant Permissive Image Corpus for Visual Generation

GPIC: A Giant Permissive Image Corpus for Visual Generation

Summary of “GPIC: A Giant Permissive Image Corpus for Visual Generation”, by researchers from Stanford, Radical Numerics, University of Michigan, and Salesforce Research

Adapting the Interface, Not the Model: Runtime Harness Adaptation for Deterministic LLM Agents

Adapting the Interface, Not the Model: Runtime Harness Adaptation for Deterministic LLM Agents

Summary of “Adapting the Interface, Not the Model: Runtime Harness Adaptation for Deterministic LLM Agents”, by Tianshi Xu, Huifeng Wen, Meng Li

Automated alignment is harder than you think

Automated alignment is harder than you think

Summary of “Automated alignment is harder than you think”, by Aleksandr Bowkis, Marie Davidsen Buhl, Jacob Pfau, Geoffrey Irving

Cosmos 3: Omnimodal World Models for Physical AI

Cosmos 3: Omnimodal World Models for Physical AI

Summary of “Cosmos 3: Omnimodal World Models for Physical AI”, by NVIDIA AI Research

EMERGENCE WORLD: A Laboratory for Evaluating Long-horizon Agent Autonomy

EMERGENCE WORLD: A Laboratory for Evaluating Long-horizon Agent Autonomy

Summary of “EMERGENCE WORLD: A Laboratory for Evaluating Long-horizon Agent Autonomy”, by researchers at Emergence AI

Measuring the AI Economy

Measuring the AI Economy

Summary of “Measuring the AI economy”, by Anton Korinek (PIIE) and Patrick McKelvey (Bank of Canada)

Building a Hill Climbing Machine

Building a Hill Climbing Machine

Summary of "MAI-Thinking-1: Building a Hill-Climbing Machine", by the Microsoft AI Team.

No article overviews match your search.

All posts on Substack → New posts on AI, video, and the tech world.