SCOPD: Sparse-Context On-Policy Self-Distillation for Efficient Vision-Language Models
SCOPD self-distillation helps vision-language models retain accuracy when most visual tokens are pruned.
1 stories, newest first · one card per story
SCOPD self-distillation helps vision-language models retain accuracy when most visual tokens are pruned.
Today 1.6K readers · 2.5K views · 779 reads · all time 189K views