NASv3 (CIFAR-10)

Training compute

2.2×10²¹ FLOP

Parameters

37.4M

Published

Nov 5, 2016

NASv3 (CIFAR-10) is an AI model developed by Google Brain (United States), first published in November 2016. It works in the vision domain, on tasks such as image classification and neural architecture search - nas. It counts among the frontier models: the systems trained with the most compute of their moment.

Training it took an estimated 2.2×10²¹ FLOP of compute (estimation method: third-party estimation,operation counting). The model has 37,400,000 parameters. It was trained on roughly 45K datapoints. The compute alone is estimated at $21K in 2023 dollars.

The reference paper has 5,894 citations. Epoch AI rates the confidence of this record as likely.

Full record

Organization: Google Brain
Country of organization: United States
Domain: Vision
Task: Image classification, Neural Architecture Search - NAS
Training compute: 2.2×10²¹ FLOP
Compute estimation method: Third-party estimation, Operation counting
Parameters: 37,400,000
Dataset size: 45K
Chips used: 800
Training cost (2023 USD): $21K
Citations: 5,894
Epoch confidence: Likely