Grover-Mega
Grover-Mega is an AI model developed by University of Washington (United States), first published in May 2019. It works in the language domain, on tasks such as language modeling/generation. It counts among the frontier models: the systems trained with the most compute of their moment.
Training it took an estimated 4.6×10²¹ FLOP of compute (estimation method: hardware,operation counting). The model has 1,500,000,000 parameters. It was trained on roughly 32B datapoints. Training ran on 128 Google TPU v3 for about 336 hours. The compute alone is estimated at $16K in 2023 dollars.
Access: Open weights (unrestricted). Its weights are openly available. The reference paper has 1,231 citations. Epoch AI rates the confidence of this record as confident.
- Organization
- University of Washington
- Country of organization
- United States
- Domain
- Language
- Task
- Language modeling/generation
- Training compute
- 4.6×10²¹ FLOP
- Compute estimation method
- Hardware, Operation counting
- Parameters
- 1,500,000,000
- Dataset size
- 32B
- Training hardware
- Google TPU v3
- Chips used
- 128
- Training time
- 336 h
- Training power draw
- 118.5 kW
- Training cost (2023 USD)
- $16K
- Model accessibility
- Open weights (unrestricted)
- Open weights
- Yes
- Citations
- 1,231
- Epoch confidence
- Confident