Jing Lin

Researcher, ByteDance Inc.

1 paper at NeurIPS 2025

Homepage· OpenReview· Semantic Scholar· Google Scholar

Poster Session 4

1 paper

Thursday, December 4, 2025 · 4:30 PM → 7:30 PM

Exhibit Hall C,D,E

VaporTok: RL-Driven Adaptive Video Tokenizer with Prior & Task Awareness

#5408 · Yang Minghao, Zechen Bai, Jing Lin, Haoqian Wang, Alex Jinpeng Wang

To train an adaptive video tokenizer, we introduce probabilistic taildrop to inject visual complexity prior to the tokenizer and incorporate GRPO for post-training, which further boosts efficiency in a task-aware adaptive manner.