|
--- |
|
language: |
|
- en |
|
license: apache-2.0 |
|
tags: |
|
- text-generation-inference |
|
- transformers |
|
- unsloth |
|
- gemma |
|
- trl |
|
base_model: unsloth/gemma-2b-it-bnb-4bit |
|
--- |
|
|
|
# Uploaded model |
|
|
|
<!-- Resumen del modelo. --> |
|
|
|
- **Developed by:** rovi27, sbenel, GaboTuco, iXrst, andreamorgar |
|
- **Funded by:** SomosNLP, HuggingFace |
|
- **Model type:** Language model, instruction tuned |
|
- **Language(s):** `es-CL`, `es-ES`, `es-MX`, `es-PE` |
|
- **License:** apache-2.0 |
|
- **Fine-tuned from model:** https://huggingface.co/unsloth/gemma-2b-bnb-4bit |
|
- **Dataset used:** [More Information Needed] <!-- Enlace al dataset utilizado para el ajuste --> |
|
|
|
### Model Sources |
|
|
|
- **Repository:** https://github.com/recetasdelaabuela/somosnlp |
|
- **Paper:** Comming soon! |
|
- **Demo:** https://huggingface.co/spaces/somosnlp/ComeBien_Demo |
|
- **Video presentation:** https://www.youtube.com/watch?v=WK-1F1TX5d4&list=PLTA-KAy8nxaASMwEUWkkTfMaDxWBxn-8J&index=19 |
|
|
|
This gemma model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library. |
|
|
|
More info in https://huggingface.co/datasets/somosnlp/RecetasDeLaAbuela |
|
|
|
|
|
## Uses |
|
|
|
This models aims to provide a tool to improve healthy cooking habits. |
|
|
|
### Direct Use |
|
|
|
Enhance recipes from Spanish-speaking countries with their nutritional values to help improve our relationship with food. |
|
The ultimate goal is to construct a Spanish-language specific intelligent nutrition assistant." |
|
|
|
|
|
### Out-of-Scope Use |
|
|
|
This model will not work well for nutrititional needs from under-represented areas, as well as recipes in other languages rather than Spanish. |
|
|
|
[More Information Needed] |
|
|
|
## Bias, Risks, and Limitations |
|
|
|
This model has been trained using the data provided in the training dataset without the supervision of nutrition experts. |
|
Please use it wisely and always consider the recommendations and guidelines from international health foundations. |
|
|
|
The model has been trained with recipes in Spanish from different areas of the world. |
|
Therefore, there can be biases regarding overrepresented areas, and some geographical areas may not be under-representation. |
|
We encourage the open-source community to collaborate in improving the dataset to cover as many geographic areas as possible! |
|
|
|
### Recommendations |
|
|
|
Users should be made aware of the risks, biases and limitations of the model. |
|
|
|
|
|
[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth) |
|
|
