LSTM LM

Training compute

1.7×10¹⁶ FLOP

Parameters

102.7M

Published

Sep 9, 2012

LSTM LM is an AI model developed by RWTH Aachen University (Germany), first published in September 2012. It works in the language domain, on tasks such as language modeling.

Training it took an estimated 1.7×10¹⁶ FLOP of compute (estimation method: operation counting). The model has 102,720,000 parameters. It was trained on roughly 27M datapoints.

Epoch AI rates the confidence of this record as speculative.

Full record

Organization: RWTH Aachen University
Country of organization: Germany
Domain: Language
Task: Language modeling
Training compute: 1.7×10¹⁶ FLOP
Compute estimation method: Operation counting
Parameters: 102,720,000
Dataset size: 27M
Epoch confidence: Speculative