DeepSeekMoE-16B

Training compute

3.4×10²² FLOP

Parameters

16B

Published

Jan 11, 2024

DeepSeekMoE-16B is an AI model developed by DeepSeek (China), first published in January 2024. It works in the language domain, on tasks such as chat.

Training it took an estimated 3.4×10²² FLOP of compute (estimation method: operation counting). The model has 16,000,000,000 parameters. It was trained on roughly 2T datapoints. Training ran on NVIDIA A100,NVIDIA H800 SXM5.

Access: Open weights (restricted use). Its weights are openly available. The reference paper has 844 citations. Epoch AI rates the confidence of this record as likely.

Full record

Organization: DeepSeek
Country of organization: China
Domain: Language
Task: Chat
Training compute: 3.4×10²² FLOP
Compute estimation method: Operation counting
Parameters: 16,000,000,000
Dataset size: 2T
Training hardware: NVIDIA A100, NVIDIA H800 SXM5
Model accessibility: Open weights (restricted use)
Open weights: Yes
Citations: 844
Epoch confidence: Likely