Transfusion
Early-fusion multimodal architecture
Episodes
-
#6311: How Transformers Process Images and VideoA Vision Transformer divides a standard image into a grid of 16-by-16 pixel patches rather than processing each individual pixel as a token. -
#5626: What Multimodal Models Actually Learn From VideoLate fusion, early fusion, and vision laziness — what actually happens when models train on video, audio, and text together. -
#5503: Inside the Hidden Image Generation PipelineThat one-click image generator is secretly a graph of many models. We reconstruct the hidden pipeline behind Gemini and ChatGPT.