In conversation.py
Shared_Embedding = Embedding(output_dim=word_embedding_size, input_dim=dictionary_size, weights=[embedding_matrix], input_length=maxlen_input, name='Shared')
however it produces the error above, should we reimport embeddings?