WeNet (Penn Treebank)

Training compute

7.3×10¹⁷ FLOP

Parameters

23M

Published

Apr 8, 2019

WeNet (Penn Treebank) is an AI model developed by Amazon (United States), first published in April 2019. It works in the language domain, on tasks such as neural architecture search - nas and language modeling.

Training it took an estimated 7.3×10¹⁷ FLOP of compute (estimation method: hardware,operation counting). The model has 23,000,000 parameters. It was trained on roughly 929K datapoints. Training ran on 1 NVIDIA V100 for about 24 hours.

Access: Unreleased. Its weights are not openly released. The reference paper has 5 citations. Epoch AI rates the confidence of this record as likely.

Full record

Organization: Amazon
Country of organization: United States
Domain: Language
Task: Neural Architecture Search - NAS, Language modeling
Training compute: 7.3×10¹⁷ FLOP
Compute estimation method: Hardware, Operation counting
Parameters: 23,000,000
Dataset size: 929K
Training hardware: NVIDIA V100
Chips used: 1
Training time: 24 h
Training power draw: 339 W
Numerical format: FP16
Model accessibility: Unreleased
Open weights: No
Citations: 5
Epoch confidence: Likely