language models little-lm A 3.8B language model pretrained from scratch to 0.384 CORE for $998 vision & video XMem under stress Probing a video object segmentation model with synthetic benchmarks AI vs. radiologists on technical recall A comparative study in screening mammography, presented at RSNA 2024 curiosities The UFC as a graph What the fight record looks like when every bout is an edge Path-integral graph convolutions Borrowing from physics to aggregate over molecular graphs Solving Sudoku like a human A heuristic solver that never backtracks
