Addressable Memory for Video World Models
Summary
This paper introduces WorldTrace, a training-free memory framework for long-horizon video world models that keeps compressed cache addressable, plus LoopBench, a benchmark for episodic recall after long detours. It improves temporal consistency by +15.5% and episodic recall by +19.5% on LoopBench.
View Cached Full Text
Cached at: 08/10/26, 02:12 AM
Paper page - Addressable Memory for Video World Models
Source: https://huggingface.co/papers/2608.07408
Abstract
Westudyvisualpersistenceininteractivevideoworldmodels.ThesemodelsrelyonaKey-Value(KV)cacheasagrowingvisualmemorytocarryforwardpreviouslygeneratedframes.However,wefindthatmodelscannolongerreliablyaddressstoredcontentoncerolloutsextendbeyondthetraininghorizon,becausetemporalRotaryPositionalEmbeddings(RoPE)offsetsthenfalloutsidetherangeseenduringtrainingandthemodelstrugglestoretrievetherelevantvisualinformationthroughattention.Moreover,naivelycompressingthecacheintheRoPE-rotatedspacecorruptsmemorybyaveragingtogetherincompatiblepositionalphases.Toaddressthis,weproposeWorldTrace,atraining-freememoryframeworkforlong-horizonvisualpersistence.WorldTracekeepscompressedmemoryaddressablebyassigningeachsummaryslotadistinct,in-distributionvirtualposition.Withinthisaddressablecache,westudytwomemorycompressionapproaches:WorldTrace-Fieldcompresseshistoryfortemporalcoherence,whileWorldTrace-Landmarkstoresverbatimscenetracesatdetectedtransitionsforepisodicrecall.WefurtherintroduceLoopBench,abenchmarkevaluatingwhetheracompressedcachecanreconstructapreviouslyvisitedsceneafteralongdetour.WorldTrace-Fieldimprovestemporalconsistencyby+15.5%,andWorldTrace-Landmarkimprovesepisodicrecallby+19.5%onLoopBench,extendingvisuallypersistentgenerationwithoutretraining.
View arXiv pageView PDFProject pageAdd to collection
Get this paper in your agent:
hf papers read 2608\.07408
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2608.07408 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2608.07408 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2608.07408 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
Memory in Video World Models (6 minute read)
A best-paper research from NVIDIA and collaborators introduces WorldTrace, a training-free framework that keeps compressed memory addressable in autoregressive video world models by assigning fixed slot-rank positions, enabling coherent long rollouts and long-range recall beyond the training horizon.
MBench: A Comprehensive Benchmark on Memory Capability for Video World Models
This paper introduces MBench, a benchmark for evaluating the memory capabilities of video world models across entity, environment, and causal consistency over long temporal horizons.
Latent Spatial Memory for Video World Models
This paper introduces latent spatial memory for video world models, storing 3D scene information directly in diffusion latent space to avoid costly pixel-space reconstruction. The proposed Mirage framework achieves up to 10.57x faster generation and 55x memory reduction while achieving state-of-the-art performance on WorldScore and RealEstate10K.
ReWorld: An Interactive World Model with Long-Horizon Memory
ReWorld presents an interactive world model that decouples short-horizon control from long-horizon memory using mixed attention windows and distribution-matching LoRA distillation, enabling real-time video generation with high action fidelity and recall.
Execution Before Memory: Diagnosing Memory Failures in Video World Models
The paper introduces MIST, an evaluation that separates scene setup from change execution to diagnose memory failures in video world models, and proposes M3-Agent, which recursively revises instructions to triple MIST reveal scores and boost WRBench and MBench-T results.