x_test = test_data y_test = test_labels x_train_de = de_train_data y_train_de = de_train_labels x_val_de = de_dev_data y_val_de = de_dev_labels x_test_de = de_test_data y_test_de = de_test_labels # tests # print(x_train[:3]) # print(x_test[:3]) EMBEDDING_DIM = 100 embeddings_index = utils.load_embs_2_dict('EMBEDDINGS/EN_ES.txt.w2v') embedding_matrix = utils.build_emb_matrix(num_embedding_vocab=vocab_size, embedding_dim=EMBEDDING_DIM, word_index=tokenizer.word_index, embeddings_index=embeddings_index) global_en_mic_train = 0 global_de_mic_train = 0 global_en_mac_train = 0 global_de_mac_train = 0 global_en_mic_tune = 0 global_de_mic_tune = 0 global_en_mac_tune = 0 global_de_mac_tune = 0 num_iterations = 8
x_test_es = es_test_data y_test_es = es_test_labels x_test_hu = hu_test_data y_test_hu = hu_test_labels x_test_sk = sk_test_data y_test_sk = sk_test_labels x_test_sv = sv_test_data y_test_sv = sv_test_labels x_test_it = it_test_data y_test_it = it_test_labels x_test_pt = pt_test_data y_test_pt = pt_test_labels EMBEDDING_DIM = 300 embeddings_index = utils.load_embs_2_dict( 'EMBEDDINGS/EN_DE_ES_HU_SK_SV_IT_PT.txt', dim=EMBEDDING_DIM) embedding_matrix = utils.build_emb_matrix(num_embedding_vocab=vocab_size, embedding_dim=EMBEDDING_DIM, word_index=tokenizer.word_index, embeddings_index=embeddings_index) global_en_mic_train = 0 global_en_mac_train = 0 global_de_mic_train = 0 global_de_mac_train = 0 global_es_mic_train = 0 global_es_mac_train = 0
y_train_de = de_train_labels x_val_de = de_dev_data y_val_de = de_dev_labels x_test_de = de_test_data y_test_de = de_test_labels # tests print(x_train[:3]) print(x_test[:3]) EMBEDDING_DIM = 200 # embeddings_index = utils.load_embs_2_dict('EMBEDDINGS/EN_DE.txt.w2v') # embeddings_index = utils.load_embs_2_dict('EMBEDDINGS/EN_ES.txt.w2v') # embeddings_index = utils.load_embs_2_dict('EMBEDDINGS/EN_DE_HU_SK_SV.txt', dim=300) embeddings_index = utils.load_embs_2_dict( 'EMBEDDINGS/glove.twitter.27B.200d.txt', dim=EMBEDDING_DIM) embedding_matrix = utils.build_emb_matrix(num_embedding_vocab=vocab_size, embedding_dim=EMBEDDING_DIM, word_index=tokenizer.word_index, embeddings_index=embeddings_index) global_en_mic_train = 0 global_de_mic_train = 0 global_en_mac_train = 0 global_de_mac_train = 0 global_en_mic_tune = 0 global_de_mic_tune = 0 global_en_mac_tune = 0 global_de_mac_tune = 0 num_iterations = 8
y_train = np.concatenate((train_labels, hu_labels, sk_labels, sv_labels)) x_val = dev_data y_val = dev_labels x_test = test_data y_test = test_labels x_train_de = de_train_data y_train_de = de_train_labels x_val_de = de_dev_data y_val_de = de_dev_labels x_test_de = de_test_data y_test_de = de_test_labels EMBEDDING_DIM = 300 embeddings_index = utils.load_embs_2_dict('EMBEDDINGS/EN_DE_HU_SK_SV.txt', dim=300) embedding_matrix = utils.build_emb_matrix(num_embedding_vocab=vocab_size, embedding_dim=EMBEDDING_DIM, word_index=tokenizer.word_index, embeddings_index=embeddings_index) global_en_mic_train = 0 global_de_mic_train = 0 global_en_mac_train = 0 global_de_mac_train = 0 global_en_mic_tune = 0 global_de_mic_tune = 0 global_en_mac_tune = 0 global_de_mac_tune = 0 num_iterations = 10 for i in range(num_iterations): print('training iteration:', i + 1)
y_train = np.concatenate((train_labels, it_labels, pt_labels)) x_val = dev_data y_val = dev_labels x_test = test_data y_test = test_labels x_train_de = de_train_data y_train_de = de_train_labels x_val_de = de_dev_data y_val_de = de_dev_labels x_test_de = de_test_data y_test_de = de_test_labels EMBEDDING_DIM = 300 embeddings_index = utils.load_embs_2_dict('EMBEDDINGS/EN_ES_IT_PT.txt', dim=300) embedding_matrix = utils.build_emb_matrix(num_embedding_vocab=vocab_size, embedding_dim=EMBEDDING_DIM, word_index=tokenizer.word_index, embeddings_index=embeddings_index) global_en_mic_train = 0 global_de_mic_train = 0 global_en_mac_train = 0 global_de_mac_train = 0 global_en_mic_tune = 0 global_de_mic_tune = 0 global_en_mac_tune = 0 global_de_mac_tune = 0 num_iterations = 10