Model's retraining

[szhitansky (Sergey Zhitansky)]

Hello,

For example, I’ve created model based on some sources. After that I got additional materials and I know that I can run training started from latest Epoch. How I can extend model by additional vocabulary.

In other words - can model be retrained by additional materials ?

Many thanks !

[guillaumekln (Guillaume Klein)]

Hi,

You can’t change the vocabulary of a trained model. A common approach is to use sub-tokenization like BPE to have an “open” vocabulary.

See for example:

Training Romance Multi-Way model

Multi-Way Neural Machine Translation Model Following (Toward Multilingual Neural Machine Translation with Universal Encoder and Decoder, Thanh-Le Ha 2016) and (Google's Multilingual Neural Machine Translation System: Enabling Zero-Shot Translation, Johnson 2016), this tutorial shows how to train a multi-source and multi-target NMT model. We have chosen here Spanish, Italian, French, Portuguese and Romanian: all 5 languages are part of the same "Romance" language family (we could have extende…