Bielik 7B

Parameters

7B

Published

Oct 24, 2024

Bielik 7B is an AI model developed by SpeakLeash and Cyfronet AGH (Poland), first published in October 2024. It works in the language domain, on tasks such as language modeling/generation and question answering.

Epoch AI has no training-compute estimate for this model. The model has 7,000,000,000 parameters. It was trained on roughly 36B datapoints. Training ran on 256 NVIDIA GH200.

Access: Open weights (unrestricted). Its weights are openly available. It is built on top of Mistral 7B. Epoch AI rates the confidence of this record as confident.

Full record

Organization: SpeakLeash, Cyfronet AGH
Country of organization: Poland
Domain: Language
Task: Language modeling/generation, Question answering
Compute estimation method: Operation counting, Hardware
Parameters: 7,000,000,000
Dataset size: 36B
Training hardware: NVIDIA GH200
Chips used: 256
Chip-hours: 2.2K
Training power draw: 352.7 kW
Numerical format: BF16
Model accessibility: Open weights (unrestricted)
Open weights: Yes
Base model: Mistral 7B
Epoch confidence: Confident