NLP for French and non-English languages

And you know it! Our CEO, Michel-Marie MAUDET, is on all fronts.

A few weeks ago at the Ministry of the Interior and Overseas France, and recently at the offices of start-up La Javaness, to demonstrate our ongoing commitment to all the players in our innovation ecosystem.

As part of the first Pie & AI Paris meetup alongside Lu W., Van Tuan DANG, Jonathan Tu NGUYEN, Andrew Ng and the DeepLearning.AI community...

It was an opportunity for experts and AI enthusiasts to exchange and share ideas on the issues and problems associated with NLP , TAL.

This was of course an opportunity to present the work of OpenLLM France.

 

3 key messages to remember:


1- Under-representation of certain languages, such as French, in automatic language processing:
The Llama 2 de meta figures are clear on this point:

  • + More than 87% is in English;
  • 0.16% is in French;
  • 0.17% is in German;

2- Beyond languages, the real issue is the under-representation of our cultural, social and economic model in France and Europe.

3- Let's not be fooled: commercial LLMs from tech giants are not really Open Source and are cruelly lacking in transparency, which poses an obvious ethical problem.

 

 

How can I help you?

CAPTCHA
8 + 3 =
Solve this simple math problem and enter the result. E.g. for 1+3, enter 4.
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.