Any company active in content generation in French that previously mostly relied on English-trained models can now leverage the largest and publicly available French language model released to date, available in beta version on cedille.ai. Developed by digital agency Coteries, Cedille is four times larger than the current biggest existing French language model.
With its six billion parameters, Cedille can generate texts in French with unprecedented quality and competes with multilingual models such as GPT-3, the current benchmark in the field. Cedille is especially powerful in generating articles, summarizing long reports, rewriting texts, translating contents, brainstorming ideas, building chatbots, and summing up complex concepts.
Based at the EPFL Innovation Park in Lausanne (canton of Vaud) and active in the B2B business, Coteries aims to develop applications based on the Cedille model for media agencies and companies producing a lot of content in French. Speaking of the targeted market, Sébastien Flury, co-founder of Coteries, explains : “The platform is a test tool to demonstrate the potential of the model, not a "ready-to-use" tool. It is intended for any type of company that wants to improve or accelerate the creation of textual content.”
Key support from Google
The Cedille project was launched with support from the Google TRC Program, which provided training for several months on Tensor Processing Units (TPUs), custom chips made by Google to speed up AI computation.
Relying on this infrastructure ensured that the training process was carbon neutral. This is a major achievement as most training processes for such models require a tremendous amount of energy, drastically increasing carbon emissions.
