AI models
KEPLER
Training compute
1.7×10²¹ FLOP
Parameters
125M
Published
Nov 23, 2020
KEPLER is an AI model developed by Tsinghua University, Mila - Quebec AI (originally Montreal Institute for Learning Algorithms), HEC, CIFAR AI Research, Princeton University and University of Montreal / Université de Montréal (China, Canada, France and United States), first published in November 2020. It works in the language domain, on tasks such as relation extraction.
Training it took an estimated 1.7×10²¹ FLOP of compute (estimation method: hardware). The model has 125,000,000 parameters. It was trained on roughly 3.5B datapoints.
Access: Unreleased. Its weights are not openly released. It is built on top of RoBERTa Base. The reference paper has 819 citations. Epoch AI rates the confidence of this record as confident.
Full record
- Organization
- Tsinghua University, Mila - Quebec AI (originally Montreal Institute for Learning Algorithms), HEC, CIFAR AI Research, Princeton University, University of Montreal / Université de Montréal
- Country of organization
- China, Canada, France, United States
- Domain
- Language
- Task
- Relation extraction
- Training compute
- 1.7×10²¹ FLOP
- Compute estimation method
- Hardware
- Parameters
- 125,000,000
- Dataset size
- 3.5B
- Numerical format
- FP16
- Model accessibility
- Unreleased
- Open weights
- No
- Base model
- RoBERTa Base
- Citations
- 819
- Epoch confidence
- Confident
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