arXiv · cs.AI · Moonwon Choi, Kisung Nam, Seunggeun Lee · 9 小时前模型评分44
帕累托自适应压缩(PACE,Pareto-Adaptive Compression for Efficient Native MLLMs):用重复人工评分预测代理优化下的人类收益曲线
Predicting Human-Gain Curves under Local Proxy Optimization from Repeated Ratings
导读
研究者提出 帕累托自适应压缩(PACE,Pareto-Adaptive Compression for Efficient Native MLLMs)(Proxy Alignment and Curvature Expansion),一种优化前诊断方法,利用重复人工评分预测固定代理目标带来的人类感知收益曲线,将局部收益近似为 A√d+Bd。
分类置信度
中 · 待人工复核
- 只有单一媒体来源,需要人工复核。
来源级别:媒体与个人
事件类型:研究发布
证据摘录
- “We propose PACE (Proxy Alignment and Curvature Expansion), a pre-optimization diagnostic that uses repeated human ratings to predict the human-gain curve induced by a fixed proxy.”
- “The resulting diagnostic is local and one-sided: it characterizes gentle optimization along the positive tilt branch of a fixed item pool.”
来源:arXiv · cs.AI · arxiv.org