question about embeddings in different languages

[DHOFM]

Hi and thanks for sharing your code. I would like to know, how you train with raw embeddings, because your code looks for checkpoint. If I would like to use the tf hub nnlm model for German embeddings https://tfhub.dev/google/tf2-preview/nnlm-de-dim128-with-normalization/1 where should this be done in your code. I find loadings of already mappings with vocal and embedding files I guess you created the embeddings but I do not find where to load raw embeddings and new vocab.

Thanks and kind regards

Dirk