How deeply does Sutskever's List cover multilingual machine translation?

Compares models from roughly 400 million to 6 billion parameters and reports quality trends, without describing translation architecture internals.

Multilingual machine translation

As explained in Sutskever's List

Multilingual machine translation is the automated conversion of text between multiple human languages using a shared translation system.

A single Transformer can translate many languages into one target language. One 102-to-English system trained on 25 billion examples tested how GPipe made much larger models practical to train. section 6.3.2

Scaling the translation model

Models ranged from roughly 400 million to 6 billion parameters. Translation quality generally improved with size, with diminishing returns at the largest scales. Low-resource languages gained especially strongly over bilingual baselines; at 1.3 billion parameters, a deep model outperformed a wider one. section 6.3.2

How Sutskever's List covers it

liveBook manning.com

Read Sutskever's List on liveBook Get the book at manning.com

Related concepts

machine translationkind of · Machine translation is the automated computational conversion of text or speech from one natural language into another.
transformer architectureuses · A transformer architecture is a neural-network architecture that processes sequences or other structured inputs using attention mechanisms and related transformations.

This explainer was generated by Manning's knowledge graph from the text of Sutskever's List. The section links point to the parts of the book it draws on.