Longvila Scaling Rl Long Videos

Train vision-language models on hour-long videos using a two-stage pipeline combining supervised fine-tuning with GRPO, reaching 71% accuracy on VideoMME while supporting 8,192 frames through efficient multi-modal sequence parallelism.

adu2021 Updated 7 repo stars

File contents

adu2021/skillxiv commit 356897197d

Frequently asked questions

npx get-skill add adu2021/longvila-scaling-rl-long-videos