logging.info("Model saved in %s", MODEL_PATH)
else :
logging.info("Loading existing model from %s...",MODEL_PATH)
model = load_model(MODEL_PATH)
logging.info("Completed loading model from file")
logging.info("Predicting on test set...")
output = model.predict(x=test_x, verbose=1)
logging.debug("Shape of output array: %s",np.shape(output))
obtained_tokens = postprocessing.undo_sequential(output)
obtained_words = postprocessing.get_words(test_doc,obtained_tokens)
precision = metrics.precision(test_answer,obtained_words)
recall = metrics.recall(test_answer,obtained_words)
f1 = metrics.f1(precision,recall)
print("### Obtained Scores ###")
print("### (full dataset) ###")
print("###")
print("### Precision : %.4f" % precision)
print("### Recall : %.4f" % recall)
print("### F1 : %.4f" % f1)
print("### ###")
keras_precision = keras_metrics.keras_precision(test_y,output)
keras_recall = keras_metrics.keras_recall(test_y,output)
keras_f1 = keras_metrics.keras_f1(test_y,output)
model.save(MODEL_PATH)
logging.info("Model saved in %s", MODEL_PATH)
else:
logging.info("Loading existing model from %s...", MODEL_PATH)
model = load_model(MODEL_PATH)
logging.info("Completed loading model from file")
logging.info("Predicting on test set...")
output = model.predict(x=test_x, verbose=1)
logging.debug("Shape of output array: %s", np.shape(output))
obtained_tokens = postprocessing.undo_sequential(output)
obtained_words = postprocessing.get_words(test_doc, obtained_tokens)
precision = metrics.precision(test_answer, obtained_words, STEM_MODE)
recall = metrics.recall(test_answer, obtained_words, STEM_MODE)
f1 = metrics.f1(precision, recall)
print("### Obtained Scores ###")
print("### (full dataset) ###")
print("###")
print("### Precision : %.4f" % precision)
print("### Recall : %.4f" % recall)
print("### F1 : %.4f" % f1)
print("### ###")
keras_precision = keras_metrics.keras_precision(test_y, output)
keras_recall = keras_metrics.keras_recall(test_y, output)
keras_f1 = keras_metrics.keras_f1(test_y, output)
