π€ Open Source LLMs and Reinforcement Learning - Departing Ai2
This article announces a departure from Ai2 and highlights contributions to open-source LLMs and reinforcement learning, specifically TΓΌlu 3 and OLMo 2.
Key Points:
β’ Advanced open-source LLMs.
β’ Developed RLVR for grounding LLMs in real RL environments (TΓΌlu 3).
β’ Created OLMo 2, a leading fully open-source LLM.
π Resources:
β’ Twitter Thread β - Announcement and links to project details.
π‘ Reliable and Accountable LLM Interactions - ORIGen Workshop
This article announces the first ORIGen workshop focused on fostering reliable and accountable interactions with Large Language Models (LLMs).
Key Points:
β’ Addresses challenges in reliable and accountable LLM interactions.
β’ Seeks submissions from various human-centered fields.
β’ Submissions due June 20.
π Resources:
β’ Twitter Thread β - Workshop announcement and submission details.
β’ COLM Conference β - Conference hosting the workshop.
π Robotics - Scalable Robot Dataset Generation
This article describes a novel method for scaling robot datasets without teleoperation, dynamic simulation, or robot hardware.
Key Points:
β’ Scales robot datasets efficiently.
β’ Requires only a smartphone scan and a human hand demo video.
β’ Generates thousands of diverse robot trajectories.
β’ Trainable by diffusion policy and VLA models.
Image
π Resources:
β’ Twitter Thread β - Details on the new method.
π‘ AI Regulation - Sam Altman's Shifting Stance
This article discusses OpenAI CEO Sam Altman's changing views on AI regulation, as observed by Gary Marcus.
Key Points:
β’ Highlights Sam Altman's shifting stance on AI regulation.
β’ Contrasts his 2023 and 2025 Senate testimony.
β’ Includes Gary Marcus's perspective.
Image
π Resources:
β’ Twitter Thread β - Discussion of Sam Altman's changing views.
β’ Gary Marcus β - Perspective on Sam Altman's statements.
π€ AI in Mathematics - AlphaEvolve's Discoveries
This article discusses Google's AlphaEvolve AI system and its mathematical discoveries.
Key Points:
β’ Solved optimal packing problems for hexagons.
β’ Improved 4x4 matrix multiplication efficiency.
β’ Represents a significant advancement in mathematics.
Image
π Resources:
β’ Twitter Thread β - Details of AlphaEvolve's discoveries.
π‘ Social Science Research - Tips for Skepticism
This article provides tips for maintaining skepticism when evaluating social science research.
Key Points:
β’ Social and organizational life is complex and messy.
β’ Large, clean effects are rarely found without caveats.
β’ Consider firm-level incentives and organizational structures.
Image
π Resources:
β’ Twitter Thread β - Tips on evaluating social science research.
π€ AI in Materials Discovery - Assessing Productivity Claims
This article questions the productivity claims of a paper on AI in materials discovery.
Key Points:
β’ Examines a paper on AI's impact on materials science productivity.
β’ Evaluates the technical claims made in the paper.
Image
π Resources:
β’ Twitter Thread β - Discussion and analysis of the research paper.
π‘ Political Commentary - Chuck Schumer and Foreign Gifts
This article critiques Chuck Schumer's stance on accepting foreign gifts, highlighting a perceived hypocrisy.
Key Points:
β’ Contrasts Chuck Schumer's public stance on foreign gifts.
β’ Highlights a Rolls Royce gifted to him by the British.
Image
π Resources:
β’ Twitter Thread β - Commentary on Chuck Schumer's stance on foreign gifts.
π Robotics Data Collection - Intelligent Data Variation Selection
This article discusses a paper focusing on intelligent data collection strategies for robotics.
Key Points:
β’ Addresses the need for intelligent data collection in robotics.
β’ Focuses on choosing data variations efficiently.
β’ Provides a framework for resource allocation.
Image
π Resources:
β’ Twitter Thread β - Discussion of the paper on intelligent data variation selection.
β’ Lihan Zha β - Author of the paper.
β’ Apurva Badithela β - Author of the paper.
β’ mzhangio β - Author of the paper.
β’ justinlidard β - Author of the paper.
β’ Majumdar_Ani β - Author of the paper.
β’ allenzren β - Author of the paper.
β’ shahdhruv_ β - Author of the paper.
π AI Agents for the Enterprise - AI Activations Series
This article announces an AI Activations series featuring discussions with brands and thought leaders on scalable AI agents for enterprises.
Key Points:
β’ Focuses on scalable AI agents for the enterprise.
β’ Features conversations with industry leaders.
β’ Registration available via link.
Image
π Resources:
β’ Twitter Thread β - Announcement of the AI Activations series.
βοΈ Support
If you liked reading this report, please star βοΈ this repository and follow me on Github β, π (previously known as Twitter) β to help others discover these resources and regular updates.
