CNN Committee (MNIST)

Training compute

5.2×10¹⁶ FLOP

Parameters

120.6K

Published

Sep 18, 2011

CNN Committee (MNIST) is an AI model developed by IDSIA (Switzerland), first published in September 2011. 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 5.2×10¹⁶ FLOP of compute (estimation method: hardware,operation counting). The model has 120,620 parameters. It was trained on roughly 420K datapoints. Training ran on 4 NVIDIA GeForce GTX 580,NVIDIA GeForce GTX 480 for about 98 hours.

Access: Unreleased. Its weights are not openly released. Epoch AI rates the confidence of this record as likely.

Full record

Organization: IDSIA
Country of organization: Switzerland
Domain: Vision
Task: Image classification
Training compute: 5.2×10¹⁶ FLOP
Compute estimation method: Hardware, Operation counting
Parameters: 120,620
Dataset size: 420K
Training hardware: NVIDIA GeForce GTX 580, NVIDIA GeForce GTX 480
Chips used: 4
Training time: 98 h
Model accessibility: Unreleased
Open weights: No
Epoch confidence: Likely