VGG16

Training compute

1.2×10¹⁹ FLOP

Parameters

138M

Published

Sep 4, 2014

VGG16 is an AI model developed by University of Oxford (United Kingdom), first published in September 2014. It works in the vision domain, on tasks such as image classification. It counts among the frontier models: the systems trained with the most compute of their moment.

Training it took an estimated 1.2×10¹⁹ FLOP of compute (estimation method: hardware). The model has 138,000,000 parameters. It was trained on roughly 1.3M datapoints. Training ran on 4 NVIDIA GeForce GTX Titan Black for about 504 hours. The compute alone is estimated at $239 in 2023 dollars.

The reference paper has 111,439 citations. Epoch AI rates the confidence of this record as confident.

Full record

Organization: University of Oxford
Country of organization: United Kingdom
Domain: Vision
Task: Image classification
Training compute: 1.2×10¹⁹ FLOP
Compute estimation method: Hardware
Parameters: 138,000,000
Dataset size: 1.3M
Training hardware: NVIDIA GeForce GTX Titan Black
Chips used: 4
Training time: 504 h
Training power draw: 2.1 kW
Training cost (2023 USD): $239
Numerical format: FP32
Citations: 111,439
Epoch confidence: Confident