| license: apache-2.0 | |
| pipeline_tag: text-generation | |
| language: | |
| - fr | |
| - en | |
| - it | |
| - de | |
| - es | |
| tags: | |
| - pretrained | |
| - llama-3 | |
| - openllm-france | |
| datasets: | |
| - OpenLLM-France/Lucie-Training-Dataset | |
| widget: | |
| - text: |- | |
| Quelle est la capitale de l'Espagne ? Madrid. | |
| Quelle est la capitale de la France ? | |
| example_title: Capital cities in French | |
| group: 1-shot Question Answering | |
| # inference: | |
| # parameters: | |
| # temperature: 1.0 | |
| # top_p: 1.0 | |
| # top_k: null | |
| # max_new_tokens: null | |
| # Model Card | |
| This repository contains universal checkpoints in DeepSpeed format for the [Lucie-7B model](https://huggingface.co/OpenLLM-France/Lucie-7B), | |
| which was trained using [this repository of code](https://github.com/OpenLLM-France/Lucie-Training) | |
| based on [a fork of `Megatron-Deepspeed`](https://github.com/OpenLLM-France/Megatron-DeepSpeed). | |
| Each checkpoint is in a subbranch (revision), which names specifies the number of training steps. | |
| For instance `step0400000` corresponds to the checkpoint after 4M training steps. | |
| Those checkpoints are provided so that the model can be retrained from a given point. | |
| ## Contact | |
| [email protected] | |