Paper proposes identifiable alignment of unpaired model representations
Neural Gromov–Monge Alignment uses pullback distances, needs no pairs or labels, and beats reported baselines.
3 stories, newest first · one card per story
Neural Gromov–Monge Alignment uses pullback distances, needs no pairs or labels, and beats reported baselines.
PixelDense separates semantic and geometric alignment teachers, improving pixel diffusion quality and training speed.
Across 13 multimodal LLMs, standard visual-text alignment scores miss corrupted visuals, an effect called the alignment illusion.
Today 1.6K readers · 2.5K views · 779 reads · all time 189K views