AI Paper Atlas

Research map / Video / Video-language understanding

Streaming Video Instruction Tuning

Jiaer Xia, Peixian Chen, Mengdan Zhang, Xing Sun, Kaiyang Zhou

CVPR 2026 · Video › Video-language understanding · approach: Video LLMs with fine-grained understanding · 32 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
use focal loss with frequency weights on state tokens instead of plain cross-entropy (ablation)Streamo-3B (Qwen2.5-VL-3B) · OVO-Bench Forward Active, SSRaccuracy: 20.99 -> 50.72Improved
use focal loss instead of plain cross-entropy (ablation)Streamo-3B (Qwen2.5-VL-3B) · OVO-Bench Forward Active, RECaccuracy: 6.45 -> 27.94Improved
use focal loss instead of plain cross-entropy (ablation)InternVL3-2B · OVO-Bench Forward Active, RECaccuracy: 9.46 -> 29.23Improved
use focal loss instead of fixed class weights (ablation)Streamo-3B (Qwen2.5-VL-3B) · OVO-Bench Forward Active, SSRaccuracy: 41.02 -> 50.72Improved
use fixed class weights instead of plain cross-entropy (ablation)Qwen2.5-VL-3B · OVO-Bench Forward Active, CRRaccuracy: 41.67 -> 49.17Improved
train on Streamo-Instruct-465K instead of ET-Instruct-164K (ablation)Streamo framework 3B · OVO-Bench overalloverall avg accuracy: 40.54 -> 52.33Improved
add offline LLaVA-Video data to ET-Instruct training (ablation)ET-Instruct-3B · OVO-Bench Real-Time vs Forward Activeaccuracy: Real-Time avg 46.47 -> 57.78; Forward avg 46.62 -> 40.77Mixed
evaluate at 2 fps without retraining (trained at 1 fps)Streamo-7B · OVO-Bench overalloverall avg accuracy: 55.61 -> 57.86Improved
convert Qwen2.5-VL-7B to streaming with Streamo trainingQwen2.5-VL-7B · OVO-Bench Real-Timeaccuracy: 58.8 -> 66.0Improved
convert Qwen2.5-VL-7B to streaming with Streamo trainingQwen2.5-VL-7B · TempCompassaccuracy: 71.7 -> 71.8No clear effect
convert Qwen2.5-VL-3B to streaming with Streamo trainingQwen2.5-VL-3B · offline benchmarks averageaverage accuracy: 56.6 -> 59.2Improved
train with ET-Instruct-164K onlyQwen2.5-VL-3B · offline benchmarks averageaverage accuracy: 56.6 -> 51.5Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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