Pleias 1.0 350m

Training compute

2.7×10²¹ FLOP

Parameters

350M

Published

Dec 5, 2024

Pleias 1.0 350m is an AI model developed by PleIAs (France), 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 2.7×10²¹ FLOP of compute (estimation method: hardware,operation counting). The model has 350,000,000 parameters. Training ran on 64 NVIDIA H100 SXM5 80GB for about 46 hours.

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

Full record

Organization: PleIAs
Country of organization: France
Domain: Language
Task: Language modeling/generation, Question answering, Translation
Training compute: 2.7×10²¹ FLOP
Compute estimation method: Hardware, Operation counting
Parameters: 350,000,000
Training hardware: NVIDIA H100 SXM5 80GB
Chips used: 64
Training time: 46 h
Training power draw: 88.1 kW
Model accessibility: Open weights (unrestricted)
Open weights: Yes
Epoch confidence: Confident