{ "validation_date": "2026-08-12", "method": "fresh CUDA forward on every local video with release QPreferScorer", "environment": { "gpu": "NVIDIA RTX 6000 Ada Generation", "torch": "2.8.0+cu128", "transformers": "5.3.0", "peft": "0.18.1", "qwen_vl_utils": "0.0.14", "decord": "0.6.0" }, "fresh_forward_workload": { "qprefer_bench_t2v_pairs": 275, "qprefer_bench_i2v_pairs": 274, "qprefer_bench_near_pairs": 294, "human_tie_pairs": 210, "human_forced_choice_pairs": 300, "videogen_rewardbench_unique_videos": 4691, "videoscore_bench_v2_videos": 500 }, "all_runs_zero_errors": true, "all_pair_margins_and_ood_raw_scores_exact_against_legacy_d2": true, "paper_number_checks": { "main_macro_with_ties": { "observed": 0.539625, "paper": 0.54, "passed": true }, "main_macro_without_ties": { "observed": 0.73095, "paper": 0.731, "passed": true }, "near_vq_with_ties": { "observed": 0.6397, "paper": 0.64, "passed": true }, "near_vq_without_ties": { "observed": 0.66, "paper": 0.66, "passed": true }, "near_ta_with_ties": { "observed": 0.6775, "paper": 0.677, "passed": true }, "near_ta_without_ties": { "observed": 0.6196, "paper": 0.62, "passed": true }, "forced_human_macro": { "observed": 0.615, "paper": 0.615, "passed": true }, "tie_human_macro_with_ties": { "observed": 0.5833, "paper": 0.583, "passed": true }, "tie_human_macro_without_ties": { "observed": 0.6327, "paper": 0.633, "passed": true }, "videogen_vq_with_ties_percent": { "observed": 47.45581358484584, "paper": 47.5, "passed": true }, "videogen_vq_without_ties_percent": { "observed": 63.66993217784477, "paper": 63.7, "passed": true }, "videogen_ta_with_ties_percent": { "observed": 43.12039312039312, "paper": 43.1, "passed": true }, "videogen_ta_without_ties_percent": { "observed": 58.307865529998395, "paper": 58.3, "passed": true }, "videogen_overall_with_ties_percent": { "observed": 51.480204494114844, "paper": 51.5, "passed": true }, "videogen_overall_without_ties_percent": { "observed": 61.86598085440777, "paper": 61.9, "passed": true }, "vs2_vq_plcc_percent": { "observed": 49.093721381630886, "paper": 49.1, "passed": true }, "vs2_ta_plcc_percent": { "observed": 24.09226069401914, "paper": 24.1, "passed": true }, "vs2_average_plcc_percent": { "observed": 36.592991037825016, "paper": 36.6, "passed": true }, "main_t2v_vq_with": { "observed": 0.5933, "paper": 0.593, "passed": true }, "main_t2v_vq_without": { "observed": 0.7796, "paper": 0.78, "passed": true }, "main_t2v_ta_with": { "observed": 0.4647, "paper": 0.465, "passed": true }, "main_t2v_ta_without": { "observed": 0.625, "paper": 0.625, "passed": true }, "main_i2v_vq_with": { "observed": 0.5758, "paper": 0.576, "passed": true }, "main_i2v_vq_without": { "observed": 0.7791, "paper": 0.779, "passed": true }, "main_i2v_ta_with": { "observed": 0.5247, "paper": 0.525, "passed": true }, "main_i2v_ta_without": { "observed": 0.7401, "paper": 0.74, "passed": true }, "near_macro_with": { "observed": 0.6586, "paper": 0.659, "passed": true }, "near_macro_without": { "observed": 0.6398, "paper": 0.64, "passed": true }, "human_forced_t2v_vq": { "observed": 0.67, "paper": 0.67, "passed": true }, "human_forced_t2v_ta": { "observed": 0.63, "paper": 0.63, "passed": true }, "human_forced_i2v_vq": { "observed": 0.74, "paper": 0.74, "passed": true }, "human_forced_i2v_ta": { "observed": 0.68, "paper": 0.68, "passed": true }, "human_forced_near_vq": { "observed": 0.44, "paper": 0.44, "passed": true }, "human_forced_near_ta": { "observed": 0.53, "paper": 0.53, "passed": true }, "human_forced_all_vq": { "observed": 0.6167, "paper": 0.617, "passed": true }, "human_forced_all_ta": { "observed": 0.6133, "paper": 0.613, "passed": true }, "human_tie_t2v_vq": { "observed": 0.6857, "paper": 0.686, "passed": true }, "human_tie_t2v_ta": { "observed": 0.4857, "paper": 0.486, "passed": true }, "human_tie_i2v_vq": { "observed": 0.7286, "paper": 0.729, "passed": true }, "human_tie_i2v_ta": { "observed": 0.5571, "paper": 0.557, "passed": true }, "human_tie_near_vq": { "observed": 0.4286, "paper": 0.429, "passed": true }, "human_tie_near_ta": { "observed": 0.6143, "paper": 0.614, "passed": true } } }