AI models
Meta Pseudo Labels
Training compute
4.8×10²² FLOP
Parameters
480M
Published
Mar 1, 2021
Meta Pseudo Labels is an AI model developed by Google Brain and Google AI (United States), first published in March 2021. It works in the vision domain, on tasks such as image classification.
Training it took an estimated 4.8×10²² FLOP of compute (estimation method: hardware). The model has 480,000,000 parameters. It was trained on roughly 131.3M datapoints. Training ran on 1,024 Google TPU v3 for about 264 hours. The compute alone is estimated at $54K in 2023 dollars.
Access: Unreleased. Its weights are not openly released. The reference paper has 766 citations. Epoch AI rates the confidence of this record as likely.
Full record
- Organization
- Google Brain, Google AI
- Country of organization
- United States
- Domain
- Vision
- Task
- Image classification
- Training compute
- 4.8×10²² FLOP
- Compute estimation method
- Hardware
- Parameters
- 480,000,000
- Dataset size
- 131.3M
- Training hardware
- Google TPU v3
- Chips used
- 1,024
- Training time
- 264 h
- Chip-hours
- 270.3K
- Training power draw
- 934.4 kW
- Training cost (2023 USD)
- $54K
- Model accessibility
- Unreleased
- Open weights
- No
- Citations
- 766
- Epoch confidence
- Likely
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