TorchMetrics - Measuring Reproducibility in PyTorch

Citations Over TimeTop 1% of 2022 papers

Abstract

A main problem with reproducing machine learning publications is the variance of metric implementations across papers. A lack of standardization leads to different behavior in mechanisms such as checkpointing, learning rate schedulers or early stopping, that will influence the reported results. For example, a complex metric such as Frchet inception distance (FID) for synthetic image quality evaluation

Related Papers

• → Short- and long-term reproducibility of radioisotopic examination of gastric emptying(1990)20 cited
• The reproducibility of reported height and body weight in repeated questionnaire surveys.(1995)
• The reproducibility of scores from three video formats.(1988)
• Comparisons of probing depth measurements with Florida probe and conventional periodontal probe(2005)
• Susquehanna Chorale Spring Concert "Roots and Wings"(2017)