RQ-Transformer (LSUN-cat dataset)

Training compute

2.9×10¹⁸ FLOP

Parameters

612M

Published

Mar 3, 2022

RQ-Transformer (LSUN-cat dataset) is an AI model developed by Kakao and POSTECH (South Korea), first published in March 2022. It works in the image generation domain, on tasks such as text-to-image and image generation.

Training it took an estimated 2.9×10¹⁸ FLOP of compute (estimation method: hardware). The model has 612,000,000 parameters. It was trained on roughly 106.1M datapoints. Training ran on 4 NVIDIA A100 for about 216 hours.

The reference paper has 763 citations. Epoch AI rates the confidence of this record as likely.

Full record

Organization: Kakao, POSTECH
Country of organization: South Korea
Domain: Image generation
Task: Text-to-image, Image generation
Training compute: 2.9×10¹⁸ FLOP
Compute estimation method: Hardware
Parameters: 612,000,000
Dataset size: 106.1M
Training hardware: NVIDIA A100
Chips used: 4
Training time: 216 h
Chip-hours: 864
Training power draw: 3.2 kW
Citations: 763
Epoch confidence: Likely