Don't Look Twice: Faster Video Transformers with Run-Length Tokenizationrccchoudhury.github.io·75 pts·jasondavies·16
Data movement bottlenecks to large-scale model training: Scaling past 1e28 FLOPepochai.org·12 pts·jasondavies·1
Lotus: Diffusion-Based Visual Foundation Model for High-Quality Dense Predictionlotus3d.github.io·47 pts·jasondavies·2
Machines of loving grace: How AI could transform the world for the betterdarioamodei.com·157 pts·jasondavies·129
INTELLECT–1: Launching the First Decentralized Training of a 10B Parameter Modelprimeintellect.ai·111 pts·jasondavies·36
Gaussian Haircut: Human Hair Reconstruction with Strand-Aligned 3D Gaussianseth-ait.github.io·2 pts·jasondavies·0
Everything Everywhere All at Once: LLMs Can In-Context Learn Multiple Tasksarxiv.org·1 pts·jasondavies·0
RLEF: Grounding Code LLMs in Execution Feedback with Reinforcement Learningarxiv.org·4 pts·jasondavies·0
PhysGen: Rigid-Body Physics-Grounded Image-to-Video Generationstevenlsw.github.io·1 pts·jasondavies·0
Clash of the Foundries: Gate All Around and Backside Power at 2nmsemianalysis.com·3 pts·jasondavies·0
ReKep: Spatio-Temporal Reasoning of Relational Keypoint Constraints for Robotsrekep-robot.github.io·48 pts·jasondavies·3