Spain’s Bold Move: Developing an Open-Source LLM Trained in Spanish and Regional Languages
In a surprising turn of events, Spain has announced its ambitious plan to develop an open-source large language model (LLM) trained in Spanish, including regional languages such as Basque, Catalan, Galician, and Valencian. Prime Minister Pedro Sánchez unveiled this groundbreaking project at the Mobile World Congress in Barcelona, signaling a significant leap forward for Spanish AI startups.
Sánchez emphasized the importance of inviting Latin American countries to participate in training the LLM, aiming to create a versatile tool for Spanish-speaking users worldwide. This strategic move not only fosters collaboration but also positions Spanish AI startups to compete effectively in Latin American markets and within Spanish-speaking communities in the US.
Carlos KiK, CTO and cofounder of Barcelona-based AiMA, highlighted the urgency of this initiative, stating, “We need this project very much to compete with the American tech companies. If we don’t move quickly, one of the big ones will come and impose their Spanish model on us.”
The development of the LLM will be a joint effort between the Barcelona Supercomputer Center, the Spanish Supercomputing Network, the Royal Spanish Academy, and the Association of Spanish Language Academies. Albert Cañigueral, tech transfer director at BSC, expressed optimism that the LLM will rival OpenAI’s GPT-3 model and aims for a summer release, pending the successful operation of MareNostrum 5.
Propelling the Industry Forward
Building on existing projects like Aina and Ilenia, the BSC initiative seeks to leverage data from various regions in Spain to enhance the LLM’s capabilities. The project’s second phase will focus on widespread adoption by industry and public institutions, offering a valuable resource for companies of all sizes.
Several startups and projects, including Clidrive and Latxa, are poised to benefit from the BSC initiative, fostering a collaborative ecosystem rather than a singular dominant model. Cañigueral emphasized the importance of coexistence and specialization, enabling tailored solutions for specific tasks.
Warm Welcome to Innovation
KiK praised the potential impact of the BSC LLM on AI startups in Spain, particularly in improving language accuracy and dialect adaptation. The project promises to streamline development processes, saving valuable time and effort for developers seeking natural-sounding software.
The BSC LLM’s ability to recognize and adapt to local dialects signifies a significant advancement in AI technology, offering users a more personalized and engaging experience.
Stay tuned for more updates on Spain’s groundbreaking LLM project!
