YOUTU-VL: A LIGHTWEIGHT VLM

Summary from the TLDR AI newsletter

Youtu-VL is a compact 4B-parameter vision-language model that uses a novel training strategy called Vision-Language Unified Autoregressive Supervision (VLUAS). This approach boosts general and vision-centric task performance without relying on task-specific modules.