一篇题为“FID彩票”的新论文调查了用于生成模型评估的Fréchet Inception Distance (FID)指标的可复现性。研究表明，使用不同种子重新训练模型对FID的影响比简单地从固定网络重采样大3.2倍。这种差异归因于随机初始化、数据排序和流匹配损失等因素。该研究提出了一个修订版的FID评估协议，包括每个单元格的最佳引导和跨多个训练种子的误差条，以解释这种固有的随机性。 AI
影响 强调了生成式AI中需要更鲁棒的评估协议，这可能会影响模型性能的报告和比较方式。
排序理由 该集群包含一篇学术论文，详细介绍了关于生成式AI中标准评估指标的新研究发现。
在 Hugging Face Daily Papers 阅读 →
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
Analysis of FID variance across different training and sampling seeds reveals significant reproducibility issues in image generation evaluation, with retraining causing larger fluctuations than resampling, and recommends updated evaluation protocols with error bars and optimal gu…
arXiv:2606.20536v1 Announce Type: new Abstract: The Frechet Inception Distance (FID) is the de facto arbiter of image generation, yet most papers report just a single number from a single trained model using a single sampling seed. How reproducible is that number if we retrain th…
The Frechet Inception Distance (FID) is the de facto arbiter of image generation, yet most papers report just a single number from a single trained model using a single sampling seed. How reproducible is that number if we retrain the model, or merely resample from it? In this pap…