WARP: Weight-Space Analysis for Recovering Training Data Portfolios
WARP: 基于权重空间分析恢复训练数据组合
机构 * University of Wisconsin-Madison(威斯康星大学麦迪逊分校)
AI总结 提出WARP框架,通过模型合并生成伪检查点,从权重空间几何特征恢复微调模型的训练数据域混合比例,在BERT和GPT-2上平均MAE分别低至0.046和0.104。
Comments This work appears in the ICML 2026 Workshop on Weight-Space Symmetries (WSS): from Foundations to Practical Applications. Our source code is available at github.com/SprocketLab/WARP