According to METR, the length of tasks that AI can solve ("time-horizon of autonomy") is doubling every 7 months But that leaves many questions unanswered ▸ What abilities in LLMs did it take to increase autonomy that much?
▸ How will we train LLMs to keep this progress
LLMS
-
Doubling AI Task Autonomy and LLM Training Challenges
By
–
-
Landing page update for B2B AI listening
By
–
Have you seen our landing page? We updated it this morning. It's a pure B2B problem. Basically, we allow brands to know what AIs think of them. It's like SEO or social listening but for AI models. AI listening, in short.
-
Metis M.2 Max Doubles LLM VLM Performance With Enhanced Security
By
–
Thanks to @techintelpro for covering our latest exciting #AI news The new Metis M.2 Max is doubles LLM and VLM performance, with up to 16GB, enhanced security and all on a VERY small, efficient device. Excellent write-up here → https://
lnkd.in/gXH-QCWY -
Chinese AI Models Gaining Edge Over Western Censored Competitors
By
–
The Chinese AI models will win Ironically because the Western models censor and the Chinese ones don't
-
Simulation of human behaviors by AI using LLM at Google France
By
–
Today, I had the pleasure of presenting at @GoogleFR, to the tech teams of @YouTube and to the partners present, my thoughts, tests and prototyping on the simulation of human behaviors by AI using LLM, and on how this could be applied for
-
Learning Anything in Two Weeks with LLMs and Research
By
–
you can learn anything in 2 weeks btw between deep research and arguing with llms, learning things just got way smoother, faster, and actually fun
-
Agentic RL Breakthrough: Solving Long Rollout Collapse
By
–
HUGE if you care about agentic RL most agent RL setups collapse under long rollouts either you spam tools endlessly or CoT devolves into junk in this paper, they fix it: > memory overflow detection > single-turn rollout format > length-normalized REINFORCE RECOMMENDED READ
-
Old stack vs New stack: consultants+tools to LLM+clarity
By
–
Old stack: consultants + tools. New stack: LLM + clarity.
