Zoneout + Variational LSTM (WT2)

Training compute

1.6×10¹⁶ FLOP

Parameters

21M

Published

Sep 26, 2016

Zoneout + Variational LSTM (WT2) is an AI model developed by MetaMind Inc and Salesforce (United States), first published in September 2016. It works in the language domain, on tasks such as language modeling.

Training it took an estimated 1.6×10¹⁶ FLOP of compute (estimation method: operation counting). The model has 21,000,000 parameters. It was trained on roughly 2M datapoints. Training ran on NVIDIA Tesla K80.

Access: Unreleased. Its weights are not openly released. The reference paper has 4,010 citations. Epoch AI rates the confidence of this record as confident.

Full record

Organization: MetaMind Inc, Salesforce
Country of organization: United States
Domain: Language
Task: Language modeling
Training compute: 1.6×10¹⁶ FLOP
Compute estimation method: Operation counting
Parameters: 21,000,000
Dataset size: 2M
Training hardware: NVIDIA Tesla K80
Model accessibility: Unreleased
Open weights: No
Citations: 4,010
Epoch confidence: Confident