top of page

Scientific Publications

Discover the research we've been working on lately.

In this study, we show that a modest amount of in-domain training brings significant improvements in few-shot tasks. Our best model, Sabiá-65B, outperforms ChatGPT-3.5 on average in 14 Portuguese tasks. The work has been published in the BRACIS 2023 conference.

The Sabiá-7B model is available on Hugging Face

Screen_Shot_2024-03-10_at_9.41_edited.jp

In this technical report, we introduce the new generation of Maritaca models, Sabiá-2, and present the most complete analysis of current LLMs in Portuguese tasks, covering 64 brazilian exams.

Our best model, Sabiá-2 Medium, greatly outperforms several competitor LLMs in the same price range.

Screen Shot 2024-03-28 at 2.06.24 PM.png

Juru is the first LLM trained in Brazilian legal data. In this study, we show that training Sabiá-2 Small on law-related documents from reliable websites, such as the CNPQ library, brings gains in Enade and OAB law tests.

enem.jpeg

In this study, we evaluated GPT-3.5 and GPT-4 on the ENEM and showed that using the Chain-of-Thought technique significantly improves GPT-4's performance.

 

Subsequent article about GPT-4 Vision's ability to "see" images.

Captura de Tela 2023-07-02 às 13.34.21.png

BLUEX is a dataset composed of entrance exams from two of Brazil's leading universities: USP and UNICAMP. Its main purpose is to serve as a reference for the evaluation of current and future AI models, including those with multimodal capabilities (image+text). The work has been published in the BRACIS 2023 conference.

Sabiá-2 Credits for Teaching and Research

If you are a student or researcher and want to use the Sabiá-2 models, which are LLMs specialized in Portuguese, Maritaca AI offers an API credits program to support your projects.

To participate, it's simple: fill out the form below with a summary of your project (up to one paragraph) and enter the approximate budget you foresee for using our models via API.

After submission, our team will review your order. If approved, you will receive instructions to start using the Sabiá-2 functionalities in your project, with no initial costs.

 

Promoting innovation and excellence in research and studies, Maritaca AI is committed to facilitating access to cutting-edge artificial intelligence for the academic community. Join us and bring your ideas to life with the power of Sabiá-2 models!

bottom of page