U-Net

Training compute

5.1×10¹⁶ FLOP

Parameters

37.7M

Published

May 18, 2015

U-Net is an AI model developed by University of Freiburg (Germany), first published in May 2015. It works in the vision domain, on tasks such as image segmentation.

Training it took an estimated 5.1×10¹⁶ FLOP of compute (estimation method: hardware). The model has 37,676,160 parameters. It was trained on roughly 7.9M datapoints.

Access: Open weights (unrestricted). Its weights are openly available. Epoch AI rates the confidence of this record as confident.

Full record

Organization: University of Freiburg
Country of organization: Germany
Domain: Vision
Task: Image segmentation
Training compute: 5.1×10¹⁶ FLOP
Compute estimation method: Hardware
Parameters: 37,676,160
Dataset size: 7.9M
Chips used: 1
Training time: 10 h
Model accessibility: Open weights (unrestricted)
Open weights: Yes
Epoch confidence: Confident