def train_and_evaluate(self, train_generator, eval_generator, evaluator, epochs=1, eposides=1,
save_dir=None, summary_dir=None, save_summary_steps=10):
if not self.initialized:
self.session.run(tf.global_variables_initializer())
Trainer._train_and_evaluate(self, train_generator, eval_generator, evaluator, epochs=epochs,
eposides=eposides,
save_dir=save_dir, summary_dir=summary_dir, save_summary_steps=save_summary_steps)
test_batch_generator = BatchGenerator(vocab, eval_data, batch_size=60)
model_dir = base_folder + 'models/bidaf/best_weights'
model = BiDAF(vocab)
model.load(model_dir)
model.session.run(tf.local_variables_initializer())
model.inference(test_batch_generator) # inference on test data
model.evaluate(test_batch_generator, evaluator)
# evaluator.exact_match_score(prediction=,ground_truth=)
# print(SquadEvaluator.exact_match_score())
# print(SquadEvaluator.f1_score)
eval_batch_generator = test_batch_generator
eval_batch_generator.init()
eval_instances = eval_batch_generator.get_instances()
model.session.run(model.eval_metric_init_op)
eval_num_steps = (eval_batch_generator.get_instance_size() +
eval_batch_generator.get_batch_size() -
1) // eval_batch_generator.get_batch_size()
output = Trainer._eval_sess(model, eval_batch_generator, eval_num_steps, None)
pred_answer = model.get_best_answer(output, eval_instances)
print('pred_answer={}'.format(pred_answer))
score = evaluator.get_score(model.get_best_answer(output, eval_instances))
metrics_string = " ; ".join("{}: {:05.3f}".format(k, v)
for k, v in score.items())
print("- Eval metrics: " + metrics_string)
def inference(self, batch_generator):
Trainer._inference(self, batch_generator)
def evaluate(self, batch_generator, evaluator):
Trainer._evaluate(self, batch_generator, evaluator)
def evaluate(self, batch_generator, evaluator):
self.bert_embedding.init_bert()
Trainer._evaluate(self, batch_generator, evaluator)