Benchmark Radar™
Star —

R2M-Bench

Explore score history ↗

Evaluates revisit memory in interactive video world models using relative consistency: for each detected return, compares revisit pair with gap-matched non-revisit and short-range control pairs to compute MemoryGain and Normalized Memory Ratio across appearance, identity, geometry, and state.

Released
2026-08-27
Institutions
Tongji University · Alibaba Group (China) · Shanghai Jiao Tong University
Evaluates
General AI, Vision & 3D, cs.CV
Openness
unknown
Importer
Claire Radar
Review status
ai-reviewed

R2M-Bench paper and code

No reported scores are on record for this benchmark yet.

Which sources cite R2M-Bench?

Related General AI, Vision & 3D, cs.CV benchmarks