HuBERT

Training compute

5.5×10²¹ FLOP

Parameters

1B

Published

Jul 27, 2021

HuBERT is an AI model developed by Facebook AI Research (United States and France), first published in July 2021. It works in the speech domain, on tasks such as speech recognition (asr).

Training it took an estimated 5.5×10²¹ FLOP of compute (estimation method: hardware). The model has 1,000,000,000 parameters. It was trained on roughly 864M datapoints.

Access: Open weights (unrestricted). Its weights are openly available. The reference paper has 4,518 citations. Epoch AI rates the confidence of this record as speculative.

Full record

Organization: Facebook AI Research
Country of organization: United States, France
Domain: Speech
Task: Speech recognition (ASR)
Training compute: 5.5×10²¹ FLOP
Compute estimation method: Hardware
Parameters: 1,000,000,000
Dataset size: 864M
Numerical format: FP16
Model accessibility: Open weights (unrestricted)
Open weights: Yes
Citations: 4,518
Epoch confidence: Speculative