Contribute to building the Aina project’s open resource ecosystem to help scale, adapt and make AI/TL more robust in Catalan.

We are looking for ideas to contribute to building Aina’s open resources ecosystem to help scale, adapt, and strengthen AI/LT in Catalan.

Some ideas that may fit within this challenge include:

• Adaptation of Aina’s foundational models (generic) to specific tasks or domains, using proprietary or public data.
• Operationalization of models in open environments: development of resources and/or modules that enable the use of Catalan in AI/LT applications and frameworks where Catalan is not included or is scarcely covered (e.g., Mycroft, Rasa, SparkNLP, Spacy, etc.).
• Development of instructions for aligning generative models to specific tasks or industry sectors.
• Development of tools and protocols to facilitate and ensure the continuous supply of data to Aina’s training corpora. For example: Data collection and sharing tools, associated tools and protocols that enable secondary use of data, implementation of protocols and/or APIs that allow programmatic access to relevant data (public or private) with permissive licenses.

The resulting products from this challenge (task- or sector-adapted models, source code, system integrations enabling model operationalization in open environments, continuous data supply mechanisms for corpora, or datasets for generative model alignment) will, once evaluated and documented, be published openly in the Aina Project’s reference repository (Aina Kit). All proposals must also include a proof of concept or a small demonstrator showcasing its usage.

This challenge is not limited to any specific field of activity, but proposals with a positive social impact will be valued positively.

Creativity and innovation are encouraged in designing solutions that significantly improve the Aina open resources ecosystem to help scale, adapt, and strengthen AI/LT in Catalan.