# R2M-Bench: Evaluating Revisit Memory via Relative Consistency in Interactive Video World Models

arXiv: https://arxiv.org/abs/2608.27328 · HTML: https://mlsanity.com/paper/2608.27328/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Video-generation world models and simulators, one of 11 directions in World models.
- Builds on 40 of its 47 references that are in the map.
- First posted on arXiv 27 Aug 2026.

## Placed in

- World models / [Video-generation world models and simulators](https://mlsanity.com/wm/video-generation-world-models-and-simulators/)

## Builds on (most cited first, 30 of 40)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [The Unreasonable Effectiveness of Deep Features as a Perceptual Metric](https://mlsanity.com/paper/1801.03924/) (2018)
- [Grounding DINO: Marrying DINO with Grounded Pre-Training for Open-Set Object Detection](https://mlsanity.com/paper/2303.05499/) (2023)
- [Wan: Open and Advanced Large-Scale Video Generative Models](https://mlsanity.com/paper/2503.20314/) (2025)
- [Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasets](https://mlsanity.com/paper/2311.15127/) (2023)
- [RAFT: Recurrent All-Pairs Field Transforms for Optical Flow](https://mlsanity.com/paper/2003.12039/) (2020)
- [Video Diffusion Models](https://mlsanity.com/paper/2204.03458/) (2022)
- [CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer](https://mlsanity.com/paper/2408.06072/) (2024)
- [HunyuanVideo: A Systematic Framework For Large Video Generative Models](https://mlsanity.com/paper/2412.03603/) (2024)
- [World Models](https://mlsanity.com/paper/1803.10122/) (2018)
- [SuperPoint: Self-Supervised Interest Point Detection and Description](https://mlsanity.com/paper/1712.07629/) (2017)
- [Genie: Generative Interactive Environments](https://mlsanity.com/paper/2402.15391/) (2024)
- [VideoGPT: Video Generation using VQ-VAE and Transformers](https://mlsanity.com/paper/2104.10157/) (2021)
- [Open-Sora: Democratizing Efficient Video Production for All](https://mlsanity.com/paper/2412.20404/) (2024)
- [GAIA-1: A Generative World Model for Autonomous Driving](https://mlsanity.com/paper/2309.17080/) (2023)
- [Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion](https://mlsanity.com/paper/2506.08009/) (2025)
- [LightGlue: Local Feature Matching at Light Speed](https://mlsanity.com/paper/2306.13643/) (2023)
- [MoCoGAN: Decomposing Motion and Content for Video Generation](https://mlsanity.com/paper/1707.04993/) (2017)
- [VBench-2.0: Advancing Video Generation Benchmark Suite for Intrinsic Faithfulness](https://mlsanity.com/paper/2503.21755/) (2025)
- [Diffusion for World Modeling: Visual Details Matter in Atari](https://mlsanity.com/paper/2405.12399/) (2024)
- [LongLive: Real-time Interactive Long Video Generation](https://mlsanity.com/paper/2509.22622/) (2025)
- [Rolling Forcing: Autoregressive Long Video Diffusion in Real Time](https://mlsanity.com/paper/2509.25161/) (2025)
- [Temporal Generative Adversarial Nets with Singular Value Clipping](https://mlsanity.com/paper/1611.06624/) (2016)
- [iVideoGPT: Interactive VideoGPTs are Scalable World Models](https://mlsanity.com/paper/2405.15223/) (2024)
- [VMem: Consistent Interactive Video Scene Generation with Surfel-Indexed View Memory](https://mlsanity.com/paper/2506.18903/) (2025)
- [Yume-1.5: A Text-Controlled Interactive World Generation Model](https://mlsanity.com/paper/2512.22096/) (2025)
- [BoQ: A Place is Worth a Bag of Learnable Queries](https://mlsanity.com/paper/2405.07364/) (2024)
- [VMBench: A Benchmark for Perception-Aligned Video Motion Generation](https://mlsanity.com/paper/2503.10076/) (2025)
- [Infinite-World: Scaling Interactive World Models to 1000-Frame Horizons via Pose-Free Hierarchical Memory](https://mlsanity.com/paper/2602.02393/) (2026)
- [MemRoPE: Training-Free Infinite Video Generation via Evolving Memory Tokens](https://mlsanity.com/paper/2603.12513/) (2026)

## Built on by (most cited first, 4 of 4)

- [OPIS: An Input-Grounded Benchmark for Multi-Object Memory in Video World Models](https://mlsanity.com/paper/2609.35052/) (2026)
- [DeCoPrune: Efficient KV-Cache Pruning for Autoregressive Video Diffusion via Denoising Consistency](https://mlsanity.com/paper/2609.39096/) (2026)
- [Grid or Nodes? Probing Object Memory in World Models Under Partial Observability](https://mlsanity.com/paper/2610.18637/) (2026)
- [WorldMemBench: A Comprehensive Benchmark for Memory of Video World Models](https://mlsanity.com/paper/2610.19063/) (2026)

## Abstract (arXiv)

High similarity between first-visit and return frames does not necessarily show that a video world model remembered the scene; the intervening rollout may simply have changed very little. This ambiguity makes absolute revisit scores sensitive to rendering stability, repetitive content, and failed motion. We introduce R2M-Bench (Relative Revisit Memory Benchmark), a benchmark of observable revisit-selective consistency. For every detected return, R2M-Bench compares the revisit pair with two controls from the same rollout: a gap-matched non-revisit pair that measures generic temporal stability and a short-range pair that estimates short-horizon consistency. These comparisons produce MemoryGain (MG), the revisit advantage over the temporal baseline, and the Normalized Memory Ratio (NMR), which normalizes this advantage by the short-to-baseline dynamic range. R2M-Bench combines 100 reference scenes with three leave-and-return trajectories to form 300 instances and evaluates appearance fidelity, scene and object identity, local geometry, and persistent state. Across seven action-conditioned video world models, Overall NMR correlates with human consistency judgments at Spearman's ρ=0.547 (95% CI [0.45,0.63]). Its within-model correlation magnitude with generated motion is 0.072, compared with 0.207 for raw revisit similarity, indicating that relative calibration substantially reduces the slow-motion shortcut. DreamX-World-Memo achieves the highest Overall NMR among the evaluated video models. Together, these results support same-rollout relative calibration as a practical way to distinguish revisit-specific consistency from generic temporal stability.