Rather
Not specified by source
Text-to-video
This version preserves the benchmark’s model label and configuration. Its exact checkpoint and cross-benchmark identity have not been independently verified.