AlexNet

Training compute

4.7×10¹⁷ FLOP

Parameters

60M

Published

Sep 30, 2012

AlexNet is an AI model developed by University of Toronto (Canada), first published in September 2012. 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 4.7×10¹⁷ FLOP of compute (estimation method: operation counting,hardware,third-party estimation). The model has 60,000,000 parameters. It was trained on roughly 2.5B datapoints. Training ran on NVIDIA GeForce GTX 580 for about 132 hours. The compute alone is estimated at $16 in 2023 dollars.

The reference paper has 125,497 citations. Epoch AI rates the confidence of this record as confident.

Full record

Organization: University of Toronto
Country of organization: Canada
Domain: Vision
Task: Image classification
Training compute: 4.7×10¹⁷ FLOP
Compute estimation method: Operation counting, Hardware, Third-party estimation
Parameters: 60,000,000
Dataset size: 2.5B
Training hardware: NVIDIA GeForce GTX 580
Training time: 132 h
Training cost (2023 USD): $16
Numerical format: FP32
Citations: 125,497
Epoch confidence: Confident