July 28, 2022
NLLB-200: Meta’s model capable of translating into 200 languages
By Sofía Sánchez González
In recent weeks, nothing else has been more talked about: Meta has just launched a model that is capable of translating into more than 200 languages. The NLLB-200 (No Language Left Behind) is intended to give people the opportunity to access and share web content in their native language and to communicate with anyone, anywhere, regardless of their language preferences.
Although it is an open source model for everyone, it stands that it is necessary to have a good capacity in a computer to be able to test it (especially larger checkpoints). But at Narrativa we solved the problem for you. We have created the first site where you can try it out, without having to install anything beforehand. Keep reading to discover the link!
This is how the NLLB-200 model can be used
Here are some facts that show that the NLLB-200 is the most powerful model for translation to date:
- 54-billion parameters
- LASER3 used for training
- Average 44% score improvement compares to other models
- New evaluation dataset: FLORES-200
According to Meta, many of these minority languages were not supported well or at all by the best existing translation tools today, such as Asturian, Luganda, Urdu, Kamba, Lao…
Here are some of its real-world applications:
- Applying AI techniques to Facebook and Instagram for translation of low-resource languages
- Building for an inclusive metaverse
- Translating Wikipedia for everyone
Minority languages
The main difference between NLLB-200 and other translation models is its focus on minority languages. In fact, it equals or improves the advances that already existed for these types of languages and in Meta they have more than fulfilled what they promised.
- Fewer than 25 African languages are currently supported by widely used translation tools — many of which are of poor quality. In contrast, NLLB-200 supports 55 African languages with high-quality results.
- In total, NLLB-200’s BLEU scores improve on the previous state of the art by an average of 44 percent across all 10k directions of the FLORES-101 benchmark. For some African and Indian languages, the increase is greater than 70 percent over recent translation systems.
Democratization of AI by large companies
Some powerful technology companies are already realizing that if progress only reaches a few, society will not be able to advance as a whole. With the launch of NLLB-200 we find ourselves faced with a new scenario in which the democratization of translation models is revolutionized.
Why is it a revolution?
Until now, the language models were quite poor at a technical level and did not work correctly due to the large amount of resources that were needed. We always had to resort to third-party APIs like Google and obviously the requests were limited.
Now Meta changes everything. In fact, they detail all the features and functionality of the model in a post with a non-technical language adapted for all users who are not so AI-savvy.
On the downside, if we want to test the model we have to use the Meta libraries and not all of us have access to a machine to host this library. However…
Try our NLLB-200 model demo!
At Narrativa we have created the first site where you can test the NLLB-200 model without having to install anything previously on your devices. You can try it on our Hugging Face profile!
As you already know, we seek advances in natural language processing to reach everyone and you can also find other models such as the detection of hate speech in social networks.
About Narrativa
Narrativa is an internationally recognized content services company that uses its proprietary artificial intelligence and machine learning platforms to build and deploy digital content solutions for enterprises. Its technology suite, consisting of data extraction, data analysis, natural language processing (NLP) and natural language generation (NLG) tools, all seamlessly work together to power a lineup of smart content creation, automated business intelligence reporting and process optimization products for a variety of industries.
Contact us to learn more about our solutions!
Share