mT5-XXL

Training compute

8.2×10²² FLOP

Parameters

13B

Published

Oct 20, 2020

mT5-XXL is an AI model developed by Google and Google Research (United States), first published in October 2020. It works in the language domain, on tasks such as language modeling, translation, language modeling/generation and question answering. It counts among the frontier models: the systems trained with the most compute of their moment.

Training it took an estimated 8.2×10²² FLOP of compute (estimation method: operation counting). The model has 13,000,000,000 parameters. It was trained on roughly 1T datapoints. The compute alone is estimated at $179K in 2023 dollars.

Access: Open weights (unrestricted). Its weights are openly available. The reference paper has 2,892 citations. Epoch AI rates the confidence of this record as confident.

Full record

Organization: Google, Google Research
Country of organization: United States
Domain: Language
Task: Language modeling, Translation, Language modeling/generation, Question answering
Training compute: 8.2×10²² FLOP
Compute estimation method: Operation counting
Parameters: 13,000,000,000
Dataset size: 1T
Training cost (2023 USD): $179K
Model accessibility: Open weights (unrestricted)
Open weights: Yes
Citations: 2,892
Epoch confidence: Confident