Tsuki0512/avrm-case-study

Case study data This directory contains the two requested case-study packages, frozen from the existing AVC and paper-evaluation artifacts. 01_avcbenchmark_reward_model/ contains the 706 AVC benchmark rows, their 706 paired MP4/MP3 files, the reward-model/evaluation CSVs, and the relevant inference code. 02_paper_benchmark_ltx2_omninft_ours_step876/ contains VABench (778), AVGenBench (235), and AV-Phys-Bench (321) samples for LTX-2, OmniNFT, and Ours ID06/step876, plus prompts… See the full description on the dataset page: https://huggingface.co/datasets/Tsuki0512/avrm-case-study.