Sparse Transformer (Enwik8)

Training compute

4.1×10¹⁸ FLOP

Parameters

95M

Published

Apr 23, 2019

Sparse Transformer (Enwik8) is an AI model developed by OpenAI (United States), first published in April 2019. It works in the language domain, on tasks such as language modeling.

Training it took an estimated 4.1×10¹⁸ FLOP of compute (estimation method: operation counting). The model has 95,000,000 parameters. It was trained on roughly 90M datapoints. Training ran on 8 NVIDIA V100.

Access: Unreleased. Its weights are not openly released. Epoch AI rates the confidence of this record as confident.

Full record

Organization: OpenAI
Country of organization: United States
Domain: Language
Task: Language modeling
Training compute: 4.1×10¹⁸ FLOP
Compute estimation method: Operation counting
Parameters: 95,000,000
Dataset size: 90M
Training hardware: NVIDIA V100
Chips used: 8
Training power draw: 4.9 kW
Model accessibility: Unreleased
Open weights: No
Epoch confidence: Confident