EXAONE 3.5 7.8B

Training compute

4.2×10²³ FLOP

Parameters

7.8B

Published

Dec 9, 2024

EXAONE 3.5 7.8B is an AI model developed by LG AI Research (South Korea), first published in December 2024. It works in the language domain, on tasks such as language modeling/generation, question answering and translation.

Training it took an estimated 4.2×10²³ FLOP of compute (estimation method: reported). The model has 7,800,000,000 parameters. It was trained on roughly 9T datapoints.

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, Question answering, Translation
Training compute: 4.2×10²³ FLOP
Compute estimation method: Reported
Parameters: 7,800,000,000
Dataset size: 9T
Model accessibility: Open weights (non-commercial)
Open weights: Yes
Epoch confidence: Confident