AFM-on-device

Training compute

4.5×10²³ FLOP

Parameters

2.7B

Published

Jul 29, 2024

AFM-on-device is an AI model developed by Apple (United States), first published in July 2024. It works in the language domain, on tasks such as language modeling/generation.

Training it took an estimated 4.5×10²³ FLOP of compute (estimation method: operation counting). The model has 2,730,000,000 parameters. It was trained on roughly 7.6T datapoints. Training ran on 2,048 Google TPU v5p.

Access: Hosted access (no API). Its weights are not openly released. Epoch AI rates the confidence of this record as confident.

Full record

Organization: Apple
Country of organization: United States
Domain: Language
Task: Language modeling/generation
Training compute: 4.5×10²³ FLOP
Compute estimation method: Operation counting
Parameters: 2,730,000,000
Dataset size: 7.6T
Training hardware: Google TPU v5p
Chips used: 2,048
Training power draw: 2.2 MW
Numerical format: BF16
Model accessibility: Hosted access (no API)
Open weights: No
Epoch confidence: Confident