Model
Meta Llama 3 8B is a pretrained decoder-only, text-to-text base model. It was trained on 15 trillion tokens of data from publicly available sources by Meta.
Use the NGC CLI to download:
Copied!
| Name | Size | Updated | Actions |
|---|---|---|---|
8b_pre_trained_bf16.nemo | 14.96 GB | February 28, 2025 UTC |