AI models
TransE
Training compute
1.3×10¹⁸ FLOP
Parameters
942M
Published
Dec 5, 2013
TransE is an AI model developed by Universite de Technologie de Compiègne – CNRS and Google (France and United States), first published in December 2013. It works in the language domain, on tasks such as entity embedding. It counts among the frontier models: the systems trained with the most compute of their moment.
Training it took an estimated 1.3×10¹⁸ FLOP of compute (estimation method: hardware). The model has 942,000,000 parameters. It was trained on roughly 17.5M datapoints. The compute alone is estimated at $30 in 2023 dollars.
The reference paper has 8,347 citations. Epoch AI rates the confidence of this record as speculative.
Full record
- Organization
- Universite de Technologie de Compiègne – CNRS, Google
- Country of organization
- France, United States
- Domain
- Language
- Task
- Entity embedding
- Training compute
- 1.3×10¹⁸ FLOP
- Compute estimation method
- Hardware
- Parameters
- 942,000,000
- Dataset size
- 17.5M
- Training cost (2023 USD)
- $30
- Citations
- 8,347
- Epoch confidence
- Speculative
SourceEpoch AI, 'AI Models'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/data/ai-models. Licensed under CC BY 4.0.
← All ai models