Pushpak Kedia (@pushpakkedia) started following @shresbm on Sep 25, 2024

🚂 2 new production models: 1.5-Pro-002 and 1.5-Flash-002 with improved math, vision understanding and long context performance 🧪 Improved version of experimental 1.5-Flash-8B 🛟 Updated settings on safety filters ↗️ 2x higher rate limits on 1.5 Flash and ~3x higher on 1.5 Pro

This weekend I read the Michelangelo paper from @GoogleDeepMind (https://t.co/vOhAX4G3ZJ) which benchmarks models' ability to find relevant information needed for complex tasks spread across large context, much like Michelangelo chiseling away irrelevant material to reveal the https://t.co/n7Ls6kjWBe

RT @rao2z: A research note describing our evaluation of the planning capabilities of o1 🍓 is now on @arxiv https://t.co/B1yFtxsHR4 (thanks…

RT @rao2z: So @karthikv792 and @kayastechly stayed up until wee hours and submitted this 18-page note to arXiv; will summarize once it is m…

RT @tuicasso: This team is pretty great! Can’t wait to see what they will continue to build

RT @danijarh: Storing knowledge about the user is super important for AI to become more useful. But that'll make it hard to switch models.…

RT @rohanpaul_ai: Visual tree of thoughts for WebUI A custom Function for Open WebUI that implements a Monte Carlo Tree Search based Tree…

RT @rohanpaul_ai: New Vikhr-Nemo-12B-Instruct-R-21-09-24 dropped in @huggingface along with 2 cool datasets Summary of Vikhr-Nemo-12B-Inst…

Excited to double down with Ritwik and Karan as they scale India (and soon Asia’s) most beloved alternative accommodations brand!