如果你在搞音乐生成，想用人类偏好来对齐模型，这个开源的奖励模型 TuneJury 可以让你直接拿来用，还附带了三种应用示例，比重新训一个省事多了。
TuneJury 是一个面向文本到音乐生成的开放实例级成对奖励模型，基于人类偏好标签（包括竞技场投票、指标对齐偏好、众包成对比较和专家美学评分）训练。它通过简单分数阈值支持数据过滤，在保留测试对和分布外基准上表现良好，并胜过先前的基线。该模型还引入锚定校准方法，以更高的数据效率恢复一致性，并在三种下游应用（推理时最优N采样、DITTO风格潜在优化和专家迭代后训练）中持续提升奖励轴收益。TuneJury 已在 GitHub 发布。
TuneJury: An Open Metric for Improving Music Generation Preference Alignment
We introduce TuneJury, an open, instance-level pairwise reward model for text-to-music that predicts a music preference score from a text prompt and an audio clip. The released checkpoint is trained on publicly available human-preference labels covering arena-style (A vs. B) votes, metric-alignment preference pairs, crowdsourced pairwise comparisons, and expert aesthetic ratings. The predicted score margin between two clips is well calibrated on our held-out test split, supporting data filtering via a simple score threshold. TuneJury generalizes to both held-out test pairs and out-of-distribution benchmarks, remaining competitive with prior baselines on the latter. For generators released after training, we introduce anchor calibration, a post-hoc, per-system Bradley-Terry calibration that recovers agreement at substantially better data efficiency than from-scratch retraining. The same frozen reward drives consistent reward-axis gains across three downstream applications: inference-time best-of-N selection, DITTO-style latent optimization, and expert-iteration post-training. TuneJury is available at https://github.com/yonghyunk1m/TuneJury.