CodeFuse-13B

Training compute

3.1×10²³ FLOP

Parameters

13B

Published

Oct 10, 2023

CodeFuse-13B is an AI model developed by Ant Group (China), first published in October 2023. It works in the language domain, on tasks such as code generation.

Training it took an estimated 3.1×10²³ FLOP of compute (estimation method: hardware). The model has 13,000,000,000 parameters. It was trained on roughly 1T datapoints. Training ran on 512 NVIDIA A100 SXM4 80 GB for about 960 hours.

Access: Open weights (unrestricted). Its weights are openly available. The reference paper has 29 citations. Epoch AI rates the confidence of this record as confident.

Full record

Organization: Ant Group
Country of organization: China
Domain: Language
Task: Code generation
Training compute: 3.1×10²³ FLOP
Compute estimation method: Hardware
Parameters: 13,000,000,000
Dataset size: 1T
Training hardware: NVIDIA A100 SXM4 80 GB
Chips used: 512
Training time: 960 h
Training power draw: 406.6 kW
Model accessibility: Open weights (unrestricted)
Open weights: Yes
Citations: 29
Epoch confidence: Confident