iWorld-Bench supplies a large video dataset and six standardized tasks with 4.9k test samples to evaluate interactive abilities of world models via a new action generation framework.
Unlike STrajectory which relies on third-party estimators, this metric directly uses the system-built precise extrinsic sequence Egt as the benchmark
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
iWorld-Bench: A Benchmark for Interactive World Models with a Unified Action Generation Framework
iWorld-Bench supplies a large video dataset and six standardized tasks with 4.9k test samples to evaluate interactive abilities of world models via a new action generation framework.