---
license: apache-2.0
---

# LLaMA 3-8B Instruct - Galician Language Model

This repository hosts a variation of LLaMA 3-8B Instruct model that has been fine-tuned to understand and generate the Galician language. This model is built on Meta's LLaMA (Large Language Model Architecture) and has been fine-tuned using LLama-Factory advanced training tools to enhance its capabilities specifically for the Galician language.

## Model Description

The LLaMA 3-8B Instruct model is one of the advanced versions of Meta's Large Language Model Architecture, initially designed to understand and generate human-like text across various languages. However, its proficiency in Galician is limited. This specific model has been fine-tuned to specialize in the Galician language, aiming to provide more accurate and contextually appropriate responses in Galician.
This model is the result of merging the [Unsloth Llama3-8B-Instruct model](https://huggingface.co/unsloth/llama-3-8b-Instruct-bnb-4bit) and the [abrahammg/Llama3-8B-Galician-Chat-Lora Lora Adapter](https://huggingface.co/abrahammg/Llama3-8B-Galician-Chat-Lora) 

#### Technical Details

- **Base Model**: Unsloth Meta's LLaMA 3 8B Instruct (https://huggingface.co/unsloth/llama-3-8b-Instruct-bnb-4bit)
- **Fine-Tuning Platform**: LLaMA Factory
- **Infrastructure**: Finisterrae III Supercomputer, CESGA (Galicia-Spain)
- **Dataset**: [irlab-udc/alpaca_data_galician](https://huggingface.co/datasets/irlab-udc/alpaca_data_galician) (with modifications)
- **Fine-Tuning Objective**: To improve text comprehension and generation in Galician.

### Trainning parameters

The project is still in the testing phase, and the training parameters will continue to vary to find the values that result in a more accurate model. Currently, the model is trained with a set of entries from the dataset and the following values:

- num_train_epochs=3.0
- finetuning_type="lora"
- per_device_train_batch_size=2
- gradient_accumulation_steps=4
- lr_scheduler_type="cosine"
- learning_rate=5e-5
- max_grad_norm=1.0

## Model tests

Below, we have several examples of question-answer pairs comparing the pure Llama3 model with the Llama3 model fine tunned for the Galician language.

*Llama3-8B* 👎
```bash
User: Cal é a capital de Canadá?
Assistant: Não! Cal não é a capital de Canadá. A capital de Canadá é Ottawa.
```
*Llama3-8B-Galician-Chat-Lora* ✅ 👍
```bash
User: Cal é a capital de Canadá?
Assistant: A capital de Canadá é Ottawa.
```
*Llama3-8B* 👎
```bash
User: Cantos habitantes ten Galicia?
Assistant: Según el Instituto Nacional de Estadística (INE) de España, la población residente en Galicia al 1 de enero de 2021 era de aproximadamente 2.79 millones de personas.
```
*Llama3-8B-Galician-Chat-Lora* ✅ 👍
```bash
User: Cantos habitantes ten Galicia?
Assistant: Segundo as últimas estimacións, Galicia ten uns 2,8 millóns de habitantes.
```

# Using the Model 💬

To utilize this model within LM Studio, simply input the URL https://huggingface.co/abrahammg/Llama3-8B-Galician-Instruct-GGUF into the search box. For the best performance, ensure you set the template to LLama3.

Or pull it in **Ollama** with the command:
```bash
ollama run abrahammg/llama3-gl-chat
```

# Lora Adapter

There is also a Lora Adapter for Llama3-8B-Instruct models in this repo [https://huggingface.co/abrahammg/Llama3-8B-Galician-Chat-Lora](https://huggingface.co/abrahammg/Llama3-8B-Galician-Chat-Lora)


## Acknowledgement

- [meta-llama/llama3](https://github.com/meta-llama/llama3)
- [hiyouga/LLaMA-Factory](https://github.com/hiyouga/LLaMA-Factory)
- [irlab-udc/alpaca_data_galician](https://huggingface.co/datasets/irlab-udc/alpaca_data_galician)