allenai / bilm-tf

Tensorflow implementation of contextualized word representations from bi-directional language models
Apache License 2.0
1.62k stars 452 forks source link

Shape of Embeddings? #216

Open rachuang22 opened 5 years ago

rachuang22 commented 5 years ago

The readme says the embeddings should be of shape (n_sentences, 3, max_sentence_length, 1024). After training, my embeddings came out to be an array of (n_words_in_corpus, 512). Does anyone understand how this happened or what this means? Thanks!