Zip CNN

Training compute

1.5×10¹² FLOP

Parameters

9.8K

Published

Dec 1, 1989

Zip CNN is an AI model developed by AT&T and Bell Laboratories (United States), first published in December 1989. It works in the vision domain, on tasks such as character recognition (ocr). It counts among the frontier models: the systems trained with the most compute of their moment.

Training it took an estimated 1.5×10¹² FLOP of compute (estimation method: operation counting). The model has 9,760 parameters. It was trained on roughly 7.3K datapoints.

The reference paper has 11,725 citations. Epoch AI rates the confidence of this record as confident.

Full record

Organization: AT&T, Bell Laboratories
Country of organization: United States
Domain: Vision
Task: Character recognition (OCR)
Training compute: 1.5×10¹² FLOP
Compute estimation method: Operation counting
Parameters: 9,760
Dataset size: 7.3K
Citations: 11,725
Epoch confidence: Confident