Paper Detail

AdaptToken: Entropy-based Adaptive Token Selection for MLLM Long Video Understanding

Haozhe Qi, Kevin Qu, Mahdi Rad, Rui Wang, Alexander Mathis, Marc Pollefeys

arxiv Score 11.8

Published 2026-03-30 · First seen 2026-03-31

General AI

Abstract

Long video understanding remains challenging for Multi-modal Large Language Models (MLLMs) due to high memory costs and context-length limits. Prior approaches mitigate this by scoring and selecting frames/tokens within short clips, but they lack a principled mechanism to (i) compare relevance across distant video clips and (ii) stop processing once sufficient evidence has been gathered. We propose AdaptToken, a training-free framework that turns an MLLM's self-uncertainty into a global control signal for long-video token selection. AdaptToken splits a video into groups, extracts cross-modal attention to rank tokens within each group, and uses the model's response entropy to estimate each group's prompt relevance. This entropy signal enables a global token budget allocation across groups and further supports early stopping (AdaptToken-Lite), skipping the remaining groups when the model becomes sufficiently certain. Across four long-video benchmarks (VideoMME, LongVideoBench, LVBench, and MLVU) and multiple base MLLMs (7B-72B), AdaptToken consistently improves accuracy (e.g., +6.7 on average over Qwen2.5-VL 7B) and continues to benefit from extremely long inputs (up to 10K frames), while AdaptToken-Lite reduces inference time by about half with comparable performance. Project page: https://haozheqi.github.io/adapt-token

Workflow Status

Review status
pending
Role
unreviewed
Read priority
now
Vote
Not set.
Saved
no
Collections
Not filed yet.
Next action
Not filled yet.

Reading Brief

No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.

Why It Surfaced

No ranking explanation is available yet.

Tags

No tags.

BibTeX

@article{qi2026adapttoken,
  title = {AdaptToken: Entropy-based Adaptive Token Selection for MLLM Long Video Understanding},
  author = {Haozhe Qi and Kevin Qu and Mahdi Rad and Rui Wang and Alexander Mathis and Marc Pollefeys},
  year = {2026},
  abstract = {Long video understanding remains challenging for Multi-modal Large Language Models (MLLMs) due to high memory costs and context-length limits. Prior approaches mitigate this by scoring and selecting frames/tokens within short clips, but they lack a principled mechanism to (i) compare relevance across distant video clips and (ii) stop processing once sufficient evidence has been gathered. We propose AdaptToken, a training-free framework that turns an MLLM's self-uncertainty into a global control },
  url = {https://arxiv.org/abs/2603.28696},
  keywords = {cs.CV, cs.AI},
  eprint = {2603.28696},
  archiveprefix = {arXiv},
}

Metadata

{}