EXAONE 3.0

Training compute

4×10²³ FLOP

Parameters

7.8B

Published

Aug 7, 2024

EXAONE 3.0 is an AI model developed by LG AI Research (South Korea), first published in August 2024. It works in the language domain, on tasks such as language modeling/generation, code generation and question answering.

Training it took an estimated 4×10²³ FLOP of compute (estimation method: reported,operation counting). The model has 7,820,000,000 parameters. It was trained on roughly 8T datapoints. Training ran on NVIDIA H100 SXM5 80GB.

Access: Open weights (non-commercial). Its weights are openly available. Epoch AI rates the confidence of this record as confident.

Full record

Organization: LG AI Research
Country of organization: South Korea
Domain: Language
Task: Language modeling/generation, Code generation, Question answering
Training compute: 4×10²³ FLOP
Compute estimation method: Reported, Operation counting
Parameters: 7,820,000,000
Dataset size: 8T
Training hardware: NVIDIA H100 SXM5 80GB
Model accessibility: Open weights (non-commercial)
Open weights: Yes
Epoch confidence: Confident