wave2vec 2.0 LARGE

Training compute

3.9×10²¹ FLOP

Parameters

317M

Published

Oct 22, 2020

wave2vec 2.0 LARGE is an AI model developed by Facebook (United States), first published in October 2020. It works in the speech domain, on tasks such as speech completion.

Training it took an estimated 3.9×10²¹ FLOP of compute (estimation method: hardware). The model has 317,000,000 parameters. It was trained on roughly 4.6B datapoints. Training ran on NVIDIA Tesla V100 DGXS 32 GB. The compute alone is estimated at $5K in 2023 dollars.

Access: Open weights (unrestricted). Its weights are openly available. The reference paper has 8,294 citations. Epoch AI rates the confidence of this record as confident.

Full record

Organization: Facebook
Country of organization: United States
Domain: Speech
Task: Speech completion
Training compute: 3.9×10²¹ FLOP
Compute estimation method: Hardware
Parameters: 317,000,000
Dataset size: 4.6B
Training hardware: NVIDIA Tesla V100 DGXS 32 GB
Training cost (2023 USD): $5K
Model accessibility: Open weights (unrestricted)
Open weights: Yes
Citations: 8,294
Epoch confidence: Confident