AI Paper Atlas

Research map / Video / Video generation

Towards Memory-Efficient Autoregressive Video Generation via Instance-Specific Parametric Absorption

Xiaomeng Fu, Jia Li, Yiming Hu, Yong Wang, Hayden So, Jiao Dai, Xiangxiang Chu, Jizhong Han

ECCV 2026 · Video › Video generation · approach: Autoregressive long video generation · 6 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 instance-specific weight absorption (Delta W) to local-attention layers (ablation)ISPA (LongLive) · qualitative 30s generationsubject consistencyImproved
keep sink frame in local attention (ablation)ISPA (LongLive) · qualitative 30s generationvisual sharpnessImproved
evict 50% of KV cache with ISPA-0.5LongLive 1.3B · VBench 5s (MovieGen prompts)aesthetic quality: 64.43 -> 64.48No clear effect
evict 50% of KV cache with ISPA-0.5Krea-Realtime 14B · VBench 30s (MovieGen prompts)aesthetic quality: 62.15 -> 63.93Improved
evict 50% of KV cache with ISPA-0.5Krea-Realtime 14B · VBench 5s (MovieGen prompts)subject consistency: 94.90 -> 92.32Hurt
evict 50% of KV cache with ISPA-0.5LongLive 1.3B · VBench 30s (MovieGen prompts)aesthetic quality: 63.59 -> 62.98Hurt
evict 50% of KV cache with ISPA-0.5Reward-Forcing 1.3B · VBench 30s (MovieGen prompts)dynamic degree: 66.92 -> 65.09Hurt
raise L-Layer fraction beyond 0.5 (K=0.7N) (ablation)ISPA (LongLive) · VBench 30soverall qualityHurt
use warmup T_warm=6 instead of 12 (ablation)ISPA (LongLive) · VBench 30squality metricsHurt
increase warmup from 12 to 24 frames (ablation)ISPA (LongLive) · VBench 30squality metricsNo clear effect
add W8A8 quantization to ISPAISPA · Inference speedoverall speedup: 1.35x -> 1.86xImproved
add W8A8 quantization to ISPAISPA · LongLive generationaesthetic and image qualityNo clear effect

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