Factorized vector quantization matches continuous visual tokens on understanding
The discrete model beats UniAR and Emu3-Chat by 22.8% and 59.2% on a nine-metric average.
58
The discrete model beats UniAR and Emu3-Chat by 22.8% and 59.2% on a nine-metric average.
Today 1.3K readers · 1.6K views · 427 reads · all time 171K views