Handwritten digit recognition network

Training compute

1.8×10¹¹ FLOP

Parameters

2.6K

Published

Nov 27, 1989

Handwritten digit recognition network is an AI model developed by AT&T (United States), first published in November 1989. It works in the vision domain, on tasks such as digit recognition. It counts among the frontier models: the systems trained with the most compute of their moment.

Training it took an estimated 1.8×10¹¹ FLOP of compute (estimation method: hardware). The model has 2,578 parameters. It was trained on roughly 9.8K datapoints.

Epoch AI rates the confidence of this record as confident.

Full record

Organization: AT&T
Country of organization: United States
Domain: Vision
Task: Digit recognition
Training compute: 1.8×10¹¹ FLOP
Compute estimation method: Hardware
Parameters: 2,578
Dataset size: 9.8K
Training time: 72 h
Epoch confidence: Confident