Apple researchers present tri-modal masked diffusion models

Expands discrete diffusion beyond two modalities, a step toward unified multimodal generative modeling.

Apple ML researchers introduce tri-modal Masked Diffusion Models pretrained across three modalities, extending prior work that initialized and finetuned unimodal models for bi-modal generation.

Read it at Apple ML Research →

Coverage: 1 sources, in order

1. 14 Oct 00:00 Apple ML Research: The Design Space of Tri-Modal Masked Diffusion Models

← Front page