Many RSI-style papers use the coding agent to hill climb results. This works and is simple, but I wonder if greater improvements can be unlocked with…
RT Vivgreat write up, evals are hard! here are 2 broad buckets we use to evaluate agents:1. Measure the State of the World2. Agent as a…
Why is evaluating agents so difficult relative to evaluating a standard LLM?An LLM generates a single response to a prompt. An agent instead interacts with an…
So many rumors going around about Situational Awareness today, but I think Leo just wanted to buy the Jersey Mike's IPO.
Agents that understand their environment (i.e., agentic world models) are both higher performing and more efficient at inference time.After reading many recent papers on world modeling…
RT Bhavin Jawade𝗞𝗶𝗺𝗶 𝗞𝟯's 𝗽𝗼𝘀𝘁-𝘁𝗿𝗮𝗶𝗻𝗶𝗻𝗴 𝗿𝗲𝗰𝗶𝗽𝗲Kimi K3's technical report just dropped. K3 is a 2.8T-parameter open-weight MoE model. I went through the report, and here I…
RT Andrew NgRe @Mononofu @JensenHuang This is a false equivalence. Everyone has the right to keep their code private. The problem is when someone tries to…
We're living in the best possible timeline.Denny’s: @JensenHuang @nvidia
RT Jensen HuangFor my first post, I’m sharing a letter @NVIDIA signed on why open models matter.AI will transform every industry, power every company, and be…
So many things to be excited about here:- Active USG support for open models.- More than just a model, also encompasses building a harness for scientific…