Workshop / Overview

You have an NLP system like a search engine, chat bot, text classifier or NER model working for one language, like English, and now you want to make it work for more languages.

We'll go through the 3 basic approaches:

  • Machine translating at inference (or query) time
  • Machine translating labelled training data (or search indices), and training a multilingual model
  • Zero-shot approaches with a massively multilingual language model like BERT or LASER

Workshop / Outcome

  • Know when to use which approach
  • Hands-on experience with each approach

Workshop / Difficulty

Intermediate level

Workshop / Prerequisites

  • A Google Colab account
  • Coding skills in a language like Python
  • A GitHub account, a ModelFront account and a Unix system are recommended for applying these approaches in production
  • If you'd like to check out the materials, code or datasets, you can take a peek at the workshop repo
  • You can read up on multilingual search at

Track / Co-organizers

Adam Bittlingmayer

CEO & Co-founder, ModelFront

Nerses Nersesyan

Junior AI Engineer, Polixis

AMLD EPFL 2021 / Workshops

Towards ethical AI – practical tools for responsible data scientists

With Johan Rochel & Lea Strohm

10:00-11:30 November 10Online

Deep Learning-Driven Text Summarization & Explainability with Reuters News Data

With Nadja Herger, Nina Hristozova & Andreea Iuga

15:00-17:30 March 02Online

Building Interpretable AI for Digital Pathology

With Guillaume Jaume, Pushpak Pati & Mara Graziani

09:00-12:00 April 27Online

AMLD / Global partners