Swin Transformer V2 (SwinV2-G)

Training compute

1.1×10²¹ FLOP

Parameters

3B

Published

Nov 18, 2021

Swin Transformer V2 (SwinV2-G) is an AI model developed by Microsoft Research Asia (China), first published in November 2021. It works in the vision and video domain, on tasks such as action recognition and image classification.

Training it took an estimated 1.1×10²¹ FLOP of compute (estimation method: hardware). The model has 3,000,000,000 parameters. Training ran on NVIDIA A100 SXM4 40 GB. The compute alone is estimated at $2K in 2023 dollars.

Access: Open weights (unrestricted). Its weights are openly available. The reference paper has 2,736 citations. Epoch AI rates the confidence of this record as confident.

Full record

Organization: Microsoft Research Asia
Country of organization: China
Domain: Vision, Video
Task: Action recognition, Image classification
Training compute: 1.1×10²¹ FLOP
Compute estimation method: Hardware
Parameters: 3,000,000,000
Training hardware: NVIDIA A100 SXM4 40 GB
Training cost (2023 USD): $2K
Model accessibility: Open weights (unrestricted)
Open weights: Yes
Citations: 2,736
Epoch confidence: Confident