This repository is used to collect and browse sampled results from different models across four benchmark tasks. Results are stored as .zip archives, and each task folder includes an HTML page that can be opened directly in a browser for convenient side-by-side comparison.
The project contains 4 parts (all under sample_results/):
vbench_t2v (Text-to-Video): sample_results/vbench_t2v/
vbench_i2v (Image-to-Video): sample_results/vbench_i2v/
ibench_t2i (Text-to-Image):… See the full description on the dataset page:
https://huggingface.co/datasets/Liveme/model_sample_results.
