SB-LM

Training compute

1.4×10¹⁸ FLOP

Parameters

300B

Published

Jun 22, 2007

SB-LM is an AI model developed by Google (United States), first published in June 2007. It works in the language domain, on tasks such as language modeling. It counts among the frontier models: the systems trained with the most compute of their moment.

Training it took an estimated 1.4×10¹⁸ FLOP of compute (estimation method: hardware). The model has 300,000,000,000 parameters. It was trained on roughly 1.8T datapoints.

Epoch AI rates the confidence of this record as likely.

Full record

Organization: Google
Country of organization: United States
Domain: Language
Task: Language modeling
Training compute: 1.4×10¹⁸ FLOP
Compute estimation method: Hardware
Parameters: 300,000,000,000
Dataset size: 1.8T
Chips used: 1,500
Training time: 24 h
Epoch confidence: Likely