← Open Source
22-hours

cabrita

Finetuning InstructLLaMA with portuguese data.

ModelsLanguage-specific modelsJupyter Notebook
Open on GitHub
Momentum
+0stars in 24 hours0.0%
558
Stars
69
Forks
+0
This week
2
Contributors
Created 2023-03-17 · Updated 2026-08-14 · #17049 today
Top developers
pedrogengopiEsposito

22-hours/cabrita is ranked 17,049 with 558 stars.

Same-category comparison

22-hours/cabrita is ranked 17,049 among these 8 peers; SKT-AI/KoGPT2 is ranked 17,034.

Same-category comparison
NameRankMetrics
xinghaochen/TinySAM17,011
Language:
Python
Stars:
561
Forks:
35
Description:
[AAAI 2025] Official PyTorch implementation of "TinySAM: Pushing the Envelope for Efficient Segment Anything Model".
01-ai/Yi-1.517,013
Stars:
560
Forks:
34
Description:
Yi-1.5 is an upgraded version of Yi, delivering stronger performance in coding, math, reasoning, and instruction-following capability.
rishikksh20/ViViT-pytorch17,026
Language:
Python
Stars:
560
Forks:
71
Description:
Implementation of ViViT: A Video Vision Transformer.
SKT-AI/KoGPT217,034
Stars:
559
Forks:
102
Description:
Korean GPT-2 pretrained cased (KoGPT2).
OpenLemur/Lemur17,113
Language:
Python
Stars:
555
Forks:
34
Description:
[ICLR 2024] Lemur: Open Foundation Models for Language Agents.
dome272/VQGAN-pytorch17,140
Language:
Python
Stars:
554
Forks:
84
Description:
Pytorch implementation of VQGAN (Taming Transformers for High-Resolution Image Synthesis) (https://arxiv.org/pdf/2012.09841.pdf).
microsoft/voxelpose-pytorch17,144
Language:
Python
Stars:
554
Forks:
91
Description:
Official implementation of "VoxelPose: Towards Multi-Camera 3D Human Pose Estimation in Wild Environment".
InternLM/InternLM-Math17,166
Language:
Python
Stars:
552
Forks:
39
Description:
State-of-the-art bilingual open-sourced Math reasoning LLMs.
Key parameters and specifications
Key parameters and specifications
ParameterValue
LanguageJupyter Notebook
CategoryModels
SubcategoryLanguage-specific models
KeywordsMultilingual
Stars558
Forks69
Contributors2
Created date2023-03-17
Updated date2026-08-14
Stars in 1 day0
Stars in 7 days0
Rank17,049
Previous rank16,843
Rank change-206
7-day stars rank15,880
Previous 7-day stars rank15,886
7-day stars rank change6
Total stars rank17,771
Previous total stars rank17,740
Total stars rank change-31
README summary

Cabrita: A portuguese finetuned instruction LLaMA This repository is intended to share all the steps and resources that we used to finetune our version of LLaMA. This model is designed for research use only, i. , cannot be used for commercial purposes or entertainment. References If I have seen further it is by standing on the sholders [sic] of Giants. For our work, we rely mainly in the works developed by: LLaMA, Stanford Alpaca, Alpaca Lora, ChatGPT and Hugging Face. So

, thank you all for the great work and open this to the world! Even this translation was not the best, the tradeoff between costs and results were. 00 to translate the full dataset to portuguese. If you want to know more about how the dataset was built go to: Stanford Alpaca. Finetuning To finetuned the LLaMA model we used the code available on Alpaca Lora, which provides code to finetune the LLaMA model using PEFT from Hugging Face. With this

, we could run our finetuning step using 1 A100 at Colab on top of LLaMA-7B. We trained during 1 hour and we found the results pretty incredible with just that much time. Example outputs Stanford Alpaca: Cabrita: Stanford Alpaca: Cabrita: You can test it using the eval notebook here. Next steps Create a better portuguese dataset Evaluate the toxicity Finetune large models Authors piEsposito pedrogengo

Frequently asked questions

How many stars and forks are recorded?

22-hours/cabrita has 558 stars and 69 forks.

What is the repository rank?

22-hours/cabrita is ranked 17,049.

What language and category are recorded?

22-hours/cabrita has 558 stars, uses Jupyter Notebook and belongs to Models.

Which repository is closest in rank?

22-hours/cabrita is compared with SKT-AI/KoGPT2, ranked 17,034 with 559 stars.

README

Cabrita: A portuguese finetuned instruction LLaMA

This repository is intended to share all the steps and resources that we used to finetune our version of LLaMA.

This model is designed for research use only, i.e., cannot be used for commercial purposes or entertainment.

References

If I have seen further it is by standing on the sholders [sic] of Giants. -- Isaac Newton

We started this section with this citation because everything we did was only possible due to the strong community and works that other people and groups did. For our work, we rely mainly in the works developed by: LLaMA, Stanford Alpaca, Alpaca Lora, ChatGPT and Hugging Face. So, thank you all for the great work and open this to the world!

Data

We translated the alpaca_data.json to portuguese using ChatGPT. Even this translation was not the best, the tradeoff between costs and results were. We paid around US$ 8.00 to translate the full dataset to portuguese. If you want to know more about how the dataset was built go to: Stanford Alpaca.

Finetuning

To finetuned the LLaMA model we used the code available on Alpaca Lora, which provides code to finetune the LLaMA model using PEFT from Hugging Face. With this, we could run our finetuning step using 1 A100 at Colab on top of LLaMA-7B. We trained during 1 hour and we found the results pretty incredible with just that much time. The notebook we used is avaible here.

Example outputs

Stanford Alpaca:

Instruction: Invente uma desculpa criativa pra dizer que não preciso ir à festa.
Response: Eu preciso se concentrar em trabalho importante para mim, por isso eu não posso ir à festa.

Cabrita:

Instrução: Invente uma desculpa criativa pra dizer que não preciso ir à festa.
Resposta: "Eu preciso ficar em casa para cuidar dos meus animais de estimação."

Stanford Alpaca:

Instruction: Como posso começar a carreira como cientista de dados? Escreva na forma de lista.
Response:
1. Comece por se especializar em um campo específico, como big data, computação, bioinformática ou estatística.
2. Obtivere um diploma em um programa de graduação relacionado a ciência de dados, como informática, matemática ou estatística.
3. Aquisite experiência prática através de internships ou projetos em empresas ou instituições.
4. Ameize sua resumo e currículo com resultados relevantes e experiência em trabalho relacionado.
5. Leia artigos, livros e atualizações sobre ciência de dados.
6. Crie um portfólio com projetos e experimentos que tenha realizado.
7. Receba um certificado de qualificação em alguma área de ciência de dados.
8. Tente obter contato com profissionais da área para conversar sobre oportunidades de emprego.
9. Acesse redes sociais, como LinkedIn, para encontrar empregos e conexões.
10. Continui sua formação atualizando-se com tecnologias e ferramentas de ciência de dados.

Cabrita:

Instrução: Como posso começar a carreira como cientista de dados? Escreva na forma de lista.
Resposta: 
1. Obtenha um mestrado em ciência de dados ou engenharia de dados.
2. Encontre um emprego como cientista de dados ou engenheiro de dados.
3. Conheça as principais ferramentas de análise de dados, como Python, R e SQL.
4. Conheça as principais técnicas de análise de dados, como regressão linear, regressão logística e análise de clustering.
5. Conheça as principais abordagens de análise de dados, como análise exploratória, análise de associação e análise de regressão.
6. Conheça as principais ferramentas de visualização de dados, como Tableau, QlikView e Power BI.
7. Conheça as principais abordagens de análise de dados, como análise de associação, análise de regressão e análise exploratória.

You can test it using the eval notebook here.

Next steps

  • Create a better portuguese dataset
  • Evaluate the toxicity
  • Finetune large models

Authors

Nearby in the ranking
Similar items