Decoupled weight decay regularization

Training compute

4.7×10¹⁷ FLOP

Parameters

36.5M

Published

Jan 4, 2019

Decoupled weight decay regularization is an AI model developed by University of Freiburg (Germany), first published in January 2019. It works in the vision domain, on tasks such as image classification.

Training it took an estimated 4.7×10¹⁷ FLOP of compute (estimation method: operation counting). The model has 36,500,000 parameters. It was trained on roughly 50K datapoints.

Access: Open weights (unrestricted). Its weights are openly available. The reference paper has 2,658 citations. Epoch AI rates the confidence of this record as confident.

Full record

Organization: University of Freiburg
Country of organization: Germany
Domain: Vision
Task: Image classification
Training compute: 4.7×10¹⁷ FLOP
Compute estimation method: Operation counting
Parameters: 36,500,000
Dataset size: 50K
Model accessibility: Open weights (unrestricted)
Open weights: Yes
Citations: 2,658
Epoch confidence: Confident