Paper Detail
Haowen Hou, Zhen Huang, Zheming Liang, Qingyi Si, Chenglin Li, Shuai Dong, Kele Shao, Ruilin Li, Dianyi Wang, Nan Duan, Jiaqi Wang
Video is temporally redundant: adjacent frames usually share most objects, background, and layout. Yet existing video multimodal large language models (video MLLMs) usually encode each sampled frame as an independent RGB image, causing visual tokens to repeat content already present in earlier frames. This suggests a more direct video interface: send a full reference frame only when the scene cannot be predicted well from prior context, and otherwise transmit a compact description of inter-frame changes. We call this interface a predictive visual code, and instantiate it for video MLLMs as AdaCodec. AdaCodec spends full visual tokens on a reference frame only when its conditional predictive cost is high; otherwise, it encodes inter-frame changes, including motion and prediction residuals, as compact P-tokens. Across all eleven benchmarks, AdaCodec improves over the Qwen3-VL-8B per-frame RGB baseline at a matched visual-token budget. Even at 1/7 the budget, AdaCodec with 32k tokens surpasses the 224k baseline on all long-video benchmarks; on five general-video benchmarks, it raises the average score while substantially cutting time-to-first-token from 9.26s to 1.62s.
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@misc{hou2026adacodec,
title = {AdaCodec: A Predictive Visual Code for Video MLLMs},
author = {Haowen Hou and Zhen Huang and Zheming Liang and Qingyi Si and Chenglin Li and Shuai Dong and Kele Shao and Ruilin Li and Dianyi Wang and Nan Duan and Jiaqi Wang},
year = {2026},
abstract = {Video is temporally redundant: adjacent frames usually share most objects, background, and layout. Yet existing video multimodal large language models (video MLLMs) usually encode each sampled frame as an independent RGB image, causing visual tokens to repeat content already present in earlier frames. This suggests a more direct video interface: send a full reference frame only when the scene cannot be predicted well from prior context, and otherwise transmit a compact description of inter-frame},
url = {https://huggingface.co/papers/2606.02569},
keywords = {video multimodal large language models, visual tokens, predictive visual code, P-tokens, conditional predictive cost, inter-frame changes, motion encoding, prediction residuals, huggingface daily},
eprint = {2606.02569},
archiveprefix = {arXiv},
}
{}