DLRM-2020

Training compute

4×10¹⁸ FLOP

Parameters

100B

Published

May 31, 2019

DLRM-2020 is an AI model developed by Facebook AI (United States), first published in May 2019. It works in the recommendation domain, on tasks such as recommender system.

Training it took an estimated 4×10¹⁸ FLOP of compute (estimation method: reported). The model has 100,000,000,000 parameters. It was trained on roughly 38.6M datapoints.

Access: Unreleased. Its weights are not openly released. The reference paper has 908 citations. Epoch AI rates the confidence of this record as confident.

Full record

Organization: Facebook AI
Country of organization: United States
Domain: Recommendation
Task: Recommender system
Training compute: 4×10¹⁸ FLOP
Compute estimation method: Reported
Parameters: 100,000,000,000
Dataset size: 38.6M
Numerical format: FP32
Model accessibility: Unreleased
Open weights: No
Citations: 908
Epoch confidence: Confident