def __init__(self, model_name, data_name, cv_runs, params_dict, logger,
eval_by_rel):
dataset = DataSet(config.DATASET[data_name])
self.train_triples, self.valid_triples, self.test_triples = dataset.load_data(
)
self.e2id, self.r2id = dataset.load_idx()
self.model_name = model_name
self.data_name = data_name
self.cv_runs = cv_runs
self.params_dict = params_dict
self.hparams = AttrDict(params_dict)
self.logger = logger
self.n_entities = len(self.e2id)
self.n_relations = len(self.r2id)
if eval_by_rel:
self.scorer = RelationScorer(self.train_triples,
self.valid_triples, self.test_triples,
self.n_relations)
else:
self.scorer = Scorer(self.train_triples, self.valid_triples,
self.test_triples, self.n_entities)
self.model = self._get_model()
self.saver = tf.train.Saver(tf.global_variables())
checkpoint_path = os.path.abspath(config.CHECKPOINT_PATH)
if not os.path.exists(checkpoint_path):
os.makedirs(checkpoint_path)
self.checkpoint_prefix = os.path.join(checkpoint_path, self.__str__())
def __init__(self, model_name, data_name, cv_runs, params_dict, logger, eval_by_rel):
dataset = DataSet(config.DATASET[data_name])
self.train_triples, self.valid_triples, self.test_triples = dataset.load_data()
self.e2id, self.r2id = dataset.load_idx()
self.groundings = dataset.load_groundings() if "SoLE" in model_name else None
self.model_name = model_name
self.data_name = data_name
self.cv_runs = cv_runs
self.params_dict = params_dict
self.hparams = AttrDict(params_dict)
if "batch_size" not in self.hparams :
if "batch_num" in self.hparams :
self.hparams["batch_size"] = int(len(self.train_triples)/self.hparams["batch_num"])
else:
raise AttributeError("Need parameter batch_size or batch_num! (Check model_param_space.py)")
self.logger = logger
self.n_entities = len(self.e2id)
self.n_relations = len(self.r2id)
if eval_by_rel:
self.scorer = RelationScorer(
self.train_triples, self.valid_triples, self.test_triples, self.n_relations)
else:
self.scorer = Scorer(
self.train_triples, self.valid_triples, self.test_triples, self.n_entities)
self.model = self._get_model()
self.saver = tf.train.Saver(tf.global_variables(), max_to_keep=1)
checkpoint_path = os.path.abspath(config.CHECKPOINT_PATH + "/" + self.__str__() +
("_NNE_" if "NNE_enable" in self.hparams and self.hparams.NNE_enable == True else "_noNNE_") + data_name)
if not os.path.exists(checkpoint_path):
os.makedirs(checkpoint_path)
self.checkpoint_prefix = os.path.join(checkpoint_path, self.__str__())
print(self.hparams)