def __init__(self, params):
# transfer parameters to self
for key, val in params.items():
setattr(self, key, val)
self.agent = Agent(params)
self.save_path = None
self.train_environment = env(params, 'train') # loaded train data here
self.dev_test_environment = env(params, 'dev') # loaded dev data here
self.test_test_environment = env(params,
'test') # loaded test data here
self.test_environment = self.dev_test_environment
self.rev_relation_vocab = self.train_environment.grapher.rev_relation_vocab
self.rev_entity_vocab = self.train_environment.grapher.rev_entity_vocab
self.max_hits_at_10 = 0
self.disc_size = 5
self.ePAD = self.entity_vocab['PAD']
self.rPAD = self.relation_vocab['PAD']
# optimize
self.baseline = ReactiveBaseline(l=self.Lambda)
self.optimizer = tf.train.AdamOptimizer(self.learning_rate)
self.input_dir = params['data_input_dir']
self.disc_embedding_size = 2 * params['embedding_size']
self.discriminator = Discriminator(self.disc_size,
self.disc_embedding_size)
self.num_rollouts = params['num_rollouts']
self.num_iter = params['total_iterations']
def __init__(self, params):
self.batch_size = params['batch_size']
self.num_rollouts = params['num_rollouts']
self.action_vocab_size = len(params['relation_vocab'])
self.entity_vocab_size = len(params['entity_vocab'])
self.embedding_size = params['embedding_size']
self.train_env = env(params,'train')
self.test_env = env(params,'test')
self.eval_every = params['eval_every']
self.learning_rate = params['learning_rate_judge']
self.total_iteration = params['total_iterations']
self.optimizer = tf.train.AdamOptimizer(self.learning_rate)
def __init__(self, params):
# transfer parameters to self
for key, val in params.items(): setattr(self, key, val);
self.agent = Agent(params)
self.save_path = None
self.train_environment = env(params, 'train')
self.dev_test_environment = env(params, 'dev')
self.test_test_environment = env(params, 'test')
self.test_environment = self.dev_test_environment
self.rev_relation_vocab = self.train_environment.grapher.rev_relation_vocab
self.rev_entity_vocab = self.train_environment.grapher.rev_entity_vocab
self.max_hits_at_10 = 0
self.ePAD = self.entity_vocab['PAD']
self.rPAD = self.relation_vocab['PAD']
# optimize
self.baseline = ReactiveBaseline(l=self.Lambda)
self.optimizer = tf.train.AdamOptimizer(self.learning_rate)
def __init__(self, params):
# transfer parameters to self
for key, val in params.items():
setattr(self, key, val)
self.agent = Agent(params)
self.save_path = None
self.train_environment = env(params, 'train')
self.dev_test_environment = env(params, 'dev')
self.test_test_environment = env(params, 'test')
self.test_environment = self.dev_test_environment
self.rev_relation_vocab = self.train_environment.grapher.rev_relation_vocab
self.rev_entity_vocab = self.train_environment.grapher.rev_entity_vocab
self.max_hits_at_10 = 0
self.ePAD = self.entity_vocab['PAD']
self.rPAD = self.relation_vocab['PAD']
self.global_step = 0
self.decaying_beta = tf.keras.optimizers.schedules.ExponentialDecay(
self.beta, decay_steps=200, decay_rate=0.90, staircase=True)
# optimize
self.baseline = ReactiveBaseline(l=self.Lambda)
# self.optimizer = tf.compat.v1.train.AdamOptimizer(self.learning_rate)
self.optimizer = tf.keras.optimizers.Adam(self.learning_rate)
