Closed nashid closed 4 years ago
You need to preprocess your embeddings, either binary or text:
python utils/preprocess_binary_word_vectors.py -v input_vectors.bin -d word2vec.en
And tell the config.py
to use them (and the same for the target lanaguge):
https://github.com/lvapeab/nmt-keras/blob/ab29848df8001f1abf58f5294c8e0c1ae78ac23e/config.py#L190
Is there code sample/example of a project how to use pretrained word2vec embeddings?