AI models
VideoMAE V2
Training compute
9.7×10²¹ FLOP
Parameters
1B
Published
Mar 29, 2023
VideoMAE V2 is an AI model developed by Nanjing University, Shenzhen Institute of Advanced Technology and Shanghai AI Lab (China), first published in March 2023. It works in the video domain, on tasks such as action recognition.
Training it took an estimated 9.7×10²¹ FLOP of compute (estimation method: hardware). The model has 1,000,000,000 parameters. It was trained on roughly 1.2B datapoints. Training ran on 64 NVIDIA A100 SXM4 80 GB for about 336 hours. The compute alone is estimated at $18K in 2023 dollars.
Access: Open weights (unrestricted). Its weights are openly available. It is built on top of ViT-G/14. The reference paper has 664 citations. Epoch AI rates the confidence of this record as confident.
Full record
- Organization
- Nanjing University, Shenzhen Institute of Advanced Technology, Shanghai AI Lab
- Country of organization
- China
- Domain
- Video
- Task
- Action recognition
- Training compute
- 9.7×10²¹ FLOP
- Compute estimation method
- Hardware
- Parameters
- 1,000,000,000
- Dataset size
- 1.2B
- Training hardware
- NVIDIA A100 SXM4 80 GB
- Chips used
- 64
- Training time
- 336 h
- Training power draw
- 51.0 kW
- Training cost (2023 USD)
- $18K
- Numerical format
- FP16
- Model accessibility
- Open weights (unrestricted)
- Open weights
- Yes
- Base model
- ViT-G/14
- Citations
- 664
- 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