AI Paper Atlas

Research map / Video / Video-language understanding

Enhancing Visual Token Representations for Video Large Language Models via Training-free Spatial-Temporal Pooling and Gridding

Bingjun Luo, Tony Wang, Hanqi Chen, Xinpeng Ding

ICLR 2026 · Video › Video-language understanding · approach: Long video understanding · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add ST-GridPool (training-free) to LLaVA-Video-7BLLaVA-Video-7B · LongVideoBenchaccuracy: 58.2 -> 60.1Improved
add ST-GridPool to LLaVA-OneVision-7BLLaVA-OneVision-7B · EgoSchemaaccuracy: 60.1 -> 62.1Improved
add pyramid temporal gridding on top of norm-based pooling (ablation)LLaVA-Video-7B · MVBenchaccuracy: 58.8 -> 59.8Improved
add norm-based spatial pooling on top of PTG (ablation)LLaVA-Video-7B · MVBenchaccuracy: 59.1 -> 59.8Improved
use ST-GridPool as token reducer at 30% budget instead of FrameFusionLLaVA-Video-7B · EgoSchemaaccuracy: 53.0 -> 56.0Improved
use ST-GridPool at 50% budget vs FrameFusion on VideoMMELLaVA-Video-7B · VideoMME (50% budget)accuracy: 62.6 -> 62.5No clear effect
apply IG-VLM image gridding to LLaVA-VideoLLaVA-Video-7B · MVBenchaccuracy: 58.6 -> 54.2Hurt
apply TS-LLaVA to LLaVA-VideoLLaVA-Video-7B · MVBenchaccuracy: VideoMME 63.3 -> 63.8; MVBench 58.6 -> 57.1Mixed
token-level gridding instead of image-level gridding (ablation)ST-GridPool on LLaVA-Video-7B · VideoMMEaccuracy: 62.8 -> 64.2Improved
increase softmax temperature beta to 5 or 10 in NSP (ablation)ST-GridPool · VideoMME / LongVideoBenchaccuracyHurt
use norm order p above 2 in NSP (ablation)ST-GridPool · VideoMME / LongVideoBenchaccuracyHurt
use spatial pooling kernel larger than 2x2 (ablation)ST-GridPool · VideoMME / LongVideoBenchaccuracyHurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy