The first SomosNLP hackathon has concluded – what an experience! It has been a true pleasure to have organized an event that brought together more than 500 participants from 29 countries.
As you already know, democratizing NLP in Spanish is the main goal of our community, and in my opinion, one of the best ways to advance toward this goal is by creating more open-source NLP resources in our language. We decided to organize a hackathon open to everyone and without a fixed theme, seeking diversity both in terms of participants and projects, and honestly, we are super happy with the result.
Congratulations, teams!
First of all, a round of applause for ALL participating teams for your effort and dedication. The truth is that you made it very difficult for the jury members, and on behalf of the entire SomosNLP team, I want to say that we are very proud and we hope you learned a lot and are encouraged to participate in future editions.
Below are the winning projects from the first edition of the SomosNLP Spanish NLP hackathon:
🥇 The hackathon’s winning project was BiomedIA, developed by the Instituto de Ingeniería del Conocimiento team consisting of Alejandro Vaca, David Betancur, Álvaro Barbero, Alba Segurado, and Guillem García. BiomedIA generates, with great accuracy, answers to biomedical questions formulated both in written and oral form. BiomedIA also won the honorable mention for the project most loved by the community by receiving the most likes on the Hugging Face hub. Additionally, it gave rise to the paper “A Complete Voice-to-Voice Generative Question Answering System for the Biomedical Domain in Spanish”, which was subsequently presented at NAACL 2022, earning the Best Poster Presentation Award.
🥈 Second place went to the project Modelo Jurídico Mexicano, developed by Ana Gabriela Palomeque, Aurelio Vázquez, Cecilia Macías, and Giovanna Madariaga, with the goal of promoting legal knowledge and streamlining the work of those who administer justice. The model developed by this team is still being used a year after the hackathon by the Suprema Corte de Justicia de la Nación of Mexico.
🥉 Third place went to the project Neutralización de género, developed by Cibeles Redondo, Javier Blasco, Fernando Velasco, Madgadela Iwona, and Juan Julián Cea. This team developed a model that allows rewriting texts in an inclusive manner, a solution with a great positive impact on today’s social landscape.
💜 The honorable mention for the best project focused on one of the UN Sustainable Development Goals went to the project Detector de Sexismo for its contribution to eliminating sexist comments, a form of gender-based violence. The project was developed by María Isabel Limaylla, Manuel Rojas, Lucel Da Silva, and Roberto Del Campo.
If you would like to give visibility to these incredible projects, here is the announcement thread on Twitter.
Wondering how the teams developed these projects? Don’t miss the series of workshops on Winning Projects from the SDG Hackathon 2022!
Thank you for sharing your knowledge, speakers!
In addition to creating open-source databases and models, during the hackathon we also invited NLP experts to share their knowledge and experience with the entire community.
Below is the list of talks and workshops and the great professionals who delivered them (by date):
- “A tour of the Hugging Face ecosystem” with Manuel Romero, NLP Engineer at Narrativa and the top contributor to the Hugging Face Hub
- “Error analysis in language models” with Omar Sanseviero, ML Engineer at Hugging Face
- “Ask Me Anything (AMA)” with Manuel Romero, NLP Engineer at Narrativa and top contributor to the Hugging Face Hub
- “Training a state-of-the-art language model” with part of the Instituto de Ingeniería del Conocimiento team that developed RigoBERTa: Alejandro Vaca (Data Scientist), Helena Montoro (Computational Linguist), Nuria Aldama (Computational Linguist), and Álvaro Barbero (Chief Data Scientist)
- “Language models for social media”, by Jose Camacho Collados and Luis Espinosa-Anke, NLP Researchers at Cardiff University
- “Artificial Intelligence and Natural Language Processing, an interesting crossroads”, by Cristina Aranda, Co-Founder of Big Onion and Mujeres Tech, and PhD in Theoretical and Applied Linguistics
- “Data sampling for NLP model training” with Paulo Villegas, Senior Technology Expert at Chief Digital Office of Telefónica, Associate Professor at Universidad Autónoma de Madrid, and co-author of the BERTIN paper
- “Ask Me Anything (AMA)” with Lewis Tunstall, ML Engineer at Hugging Face and co-author of the book “Natural Language Processing with Transformers”
- “Machine translation: introduction and current challenges”, by Eva Martínez García, Senior Research Scientist at NielsenIQ and Professor in the UNIR AI Master’s program
- “Data labeling for NLP” with Daniel Vila, Co-Founder and CEO of Recognai
- “NLP considerations for minoritized languages”, by Ximena Gutierrez-Vasques, Post-doctoral Researcher at the University of Zurich and Computational Linguist
- “Inferring topics with unsupervised clustering” with Victoriano Izquierdo, Co-Founder and CEO of Graphext
- “Abstract writing workshop” with Laura N Montoya, President of LatinX in AI, and Javier Turek, Senior Research Scientist at Intel Labs
On our channel youtube.com/c/somosnlp you can find the recordings of all these events. They already have more than 5000 views!
Thank you for your support, sponsors!
Thank you for your time and for supporting us so that our initiative can reach further. Special thanks to our three gold sponsors: Paperspace provided the GPUs on which teams trained their models, Platzi offered scholarships on their online learning platform as prizes, and Hugging Face designed the swag that we gave to all participants.
Next year, bigger and better!
Thank you once again to the SomosNLP team, the sponsors, the jury members, the speakers, and above all, to everyone who participated, for helping us advance the state of the art of NLP in Spanish.
Next year, bigger and better!
Do you already have the t-shirt that Hugging Face designed for the hackathon?
It's free for all participants, check your email 🤗
Related articles:
- How to contribute to NLP, article on the Platzi blog, gold sponsor of the hackathon
- The best models of the first NLP hackathon in Spanish, article on the Narrativa blog, silver sponsor of the hackathon
- The BiomedIA project wins the 2022 Spanish NLP hackathon, article on the Instituto de Ingeniería del Conocimiento blog, where the members of the team that developed BiomedIA work