def __init__(self, is_Setting_Embedder=True):
self.vocab_size = 85
self.cell_Dec_fw = tf.nn.rnn_cell.BasicLSTMCell(128)
self.cell_Dec_bw = tf.nn.rnn_cell.BasicLSTMCell(128)
self.cell_Output_fw = tf.nn.rnn_cell.BasicLSTMCell(self.vocab_size)
self.cell_Output_bw = tf.nn.rnn_cell.BasicLSTMCell(self.vocab_size)
self.Weight_Hidden = self.weight_variable(shape=[50, 128], name='Weight_Hidden_Tagger')
self.Bias_Hidden = self.bias_variable(shape=[128], name='Bias_Hiddne_Tagger')
self.Weight_Embedding = self.weight_variable(shape=[128, 256], name='Weight_Embedding_Tagger')
self.Bias_Embedding = self.bias_variable(shape=[256], name='Bias_Embedding_Tagger')
self.Batch = 500
self.P_Length = 100
self.C_Length = 20
self.Char_Onehot = 128
self.Embedding_Size = 50
self.X_Input = tf.placeholder(shape=[None, self.P_Length, self.Embedding_Size], dtype=tf.float32, name='Tagger_Input')
if is_Setting_Embedder:
self.pos_Embed = POS_Embed.POS_Embedder()
self.DataHolder = data_processor.Data_holder(is_Just_Embedding=True)
self.DataHolder.set_batch()
self.pos_Embed.set_Data_Hlder(self.DataHolder)
def __init__(self):
self.dataset = dp.Data_holder()
self.dataset.set_batch()
self.dataset.set_sentence_batch()
self.question = 0
self.paragraph = 0
self.start_index = 0
self.stop_index = 0
self.batch = 500
self.p_length = 100
self.q_length = 30
self.embedding_size = 50
self.paragraph, self.question, self.start_index, self.stop_index = self.dataset.get_next_batch(
)
self.cell_Qr_f = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_Pr_f = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_stack_f = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_stack_b = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_stack_f2 = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_stack_b2 = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_stack_f3 = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_stack_b3 = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_output_f = tf.nn.rnn_cell.BasicLSTMCell(1)
self.cell_output_b = tf.nn.rnn_cell.BasicLSTMCell(1)
self.cell_Qr_f_ = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_Pr_f_ = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_stack_f_ = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_stack_b_ = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_stack_f2_ = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_stack_b2_ = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_stack_f3_ = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_stack_b3_ = tf.nn.rnn_cell.BasicLSTMCell(self.embedding_size)
self.cell_output_f_ = tf.nn.rnn_cell.BasicLSTMCell(1)
self.cell_output_b_ = tf.nn.rnn_cell.BasicLSTMCell(1)
self.output_Start = None
self.output_Stop = None
self.x_q_holer = tf.placeholder(
dtype=tf.float32,
shape=[self.batch, self.q_length, self.embedding_size])
self.x_p_holer = tf.placeholder(
dtype=tf.float32,
shape=[self.batch, self.p_length, self.embedding_size])
def __init__(self):
self.dataset = dp.Data_holder()
self.dataset.set_batch()
self.question = 0
self.paragraph = 0
self.start_index = 0
self.stop_index = 0
self.batch = 1000
self.p_length = 125
self.q_length = 40
self.seq_length = [40] * 1000
self.seq_length_ = [125] * 1000
self.paragraph, self.question, self.start_index, self.stop_index = self.dataset.get_next_batch(
)
self.cell_Qr = tf.nn.rnn_cell.BasicLSTMCell(100)
self.cell_Pr = tf.nn.rnn_cell.BasicLSTMCell(100)
self.cell_output_START = tf.nn.rnn_cell.BasicLSTMCell(1)
self.cell_output_STOP = tf.nn.rnn_cell.BasicLSTMCell(1)
self.W_conv1 = self.weight_variable([2, 100, 1, 4])
self.b_conv1 = self.bias_variable([4])
self.W_conv2 = self.weight_variable([3, 100, 1, 4])
self.b_conv2 = self.bias_variable([4])
self.W_conv3 = self.weight_variable([4, 100, 1, 4])
self.b_conv3 = self.bias_variable([4])
self.W_conv4 = self.weight_variable([5, 100, 1, 4])
self.b_conv4 = self.bias_variable([4])
self.W_conv5 = self.weight_variable([6, 100, 1, 4])
self.b_conv5 = self.bias_variable([4])
self.W_fc_Q = self.weight_variable([20, 90])
self.b_fc = self.bias_variable([90])
self.output_Start = None
self.output_Stop = None
self.x_q_holer = tf.placeholder(dtype=tf.float32,
shape=[self.batch, 40, 100])
self.x_p_holer = tf.placeholder(dtype=tf.float32,
shape=[self.batch, 125, 100])
def __init__(self, is_Para=True):
self.dataset = dp.Data_holder()
self.dataset.set_batch()
#self.dataset.get_POS()
self.question = 0
self.paragraph = 0
self.POS_Embeddings = 0
self.POS_Q_Embeddings = 0
self.start_index = 0
self.stop_index = 0
self.attention_Label = 0
self.embedding_size = 50
self.POS_Embedding_Size = 128
self.batch = 500
self.p_length = 125
self.q_length = 30
self.paragraph, self.question, self.start_index, self.stop_index, self.POS_Embeddings = self.dataset.get_next_batch(
)
self.cell_Q_Enc_fw = tf.nn.rnn_cell.BasicLSTMCell(32)
self.cell_Q_Enc_bw = tf.nn.rnn_cell.BasicLSTMCell(32)
self.cell_P_Enc_fw = tf.nn.rnn_cell.BasicLSTMCell(32)
self.cell_P_Enc_bw = tf.nn.rnn_cell.BasicLSTMCell(32)
self.cell_POS_Enc_fw = tf.nn.rnn_cell.BasicLSTMCell(32)
self.cell_POS_Enc_bw = tf.nn.rnn_cell.BasicLSTMCell(32)
self.cell_output_Index_fw = tf.nn.rnn_cell.BasicLSTMCell(1)
self.cell_output_Index_bw = tf.nn.rnn_cell.BasicLSTMCell(1)
self.cell_modeling_fw = tf.nn.rnn_cell.BasicLSTMCell(64)
self.cell_modeling_bw = tf.nn.rnn_cell.BasicLSTMCell(64)
self.cell_modeling_fw_ = tf.nn.rnn_cell.BasicLSTMCell(32)
self.cell_modeling_bw_ = tf.nn.rnn_cell.BasicLSTMCell(32)
self.W_conv_2 = self.weight_variable([2, 1, 1, 4])
self.W_conv_3 = self.weight_variable([3, 1, 1, 4])
self.W_conv_4 = self.weight_variable([4, 1, 1, 4])
self.W_conv_5 = self.weight_variable([5, 1, 1, 4])
self.Weight_Paragraph = self.weight_variable(shape=[1, 64, 64])
self.Weight_POS = self.weight_variable(shape=[1, 64, 64])
self.Bias_Paragraph = self.bias_variable(shape=[64])
self.Bias_POS = self.bias_variable(shape=[64])
self.Weight_Dec_Input = self.weight_variable(shape=[1, 64 * 5, 128])
self.Bias_Dec_Input = self.bias_variable(shape=[128])
self.Weight_D_Out = self.weight_variable(shape=[1, 128, 1])
self.Bias_D_Out = self.weight_variable(shape=[1])
self.Weight_D_Out_Stop = self.weight_variable(shape=[1, 128, 1])
self.Bias_D_Out_Stop = self.weight_variable(shape=[1])
self.W_fc = self.weight_variable([1, 16, 1])
self.b_fc = self.bias_variable([1])
self.output_Start = None
self.output_Stop = None
self.x_q_holer = tf.placeholder(
dtype=tf.float32,
shape=[self.batch, self.q_length, self.embedding_size],
name='x_q_holer')
self.x_p_holer = tf.placeholder(
dtype=tf.float32,
shape=[self.batch, self.p_length, self.embedding_size],
name='x_p_holer')
self.x_pos_q_holder = tf.placeholder(
dtype=tf.float32,
shape=[self.batch, self.q_length, self.POS_Embedding_Size],
name='x_POS_holer')
self.x_pos_holder = tf.placeholder(
dtype=tf.float32,
shape=[self.batch, self.p_length, 85],
name='x_POS_holer')
fileName = 'C:\\Users\\Administrator\\Desktop\\qadataset\\pos_embed'
f = open(fileName, 'w')
for a in range(84):
string = ""
for i in range(128):
string = string + str(embeddings[a, 0, i]) + " "
string = string + "#"
f.write(string)
f.close()
pos_Embed = POS_Embed.POS_Embedder()
dataholder = data_processor.Data_holder()
dataholder.set_batch()
pos_Embed.set_Data_Hlder(dataholder)
"""
pos_Modeler = Combined_POS_Processor.POS_Processor(is_Setting_Embedder=False)
pos_Modeler.setEmbedder(pos_Embed)
pos_Embed.set_POS_Tagger(pos_Modeler)
a, b = pos_Embed.read_POS_From_Embed()
pos_Embed.get_Fianl_POS(dataholder.paragraph_arr, dataholder.numberOf_available_question, a)
"""
print('Complete')
def __init__(self, is_Para=True):
self.dataset = dp.Data_holder()
self.dataset.set_batch()
if is_Para:
self.dataset.set_sentence_batch()
else:
self.dataset.set_sentence_batch_para()
self.question = 0
self.paragraph = 0
self.POS_Embeddings = 0
self.POS_Q_Embeddings = 0
self.start_index = 0
self.stop_index = 0
self.attention_Label = 0
self.embedding_size = 50
self.POS_Embedding_Size = 128
self.batch = 500
self.p_length = 100
self.q_length = 30
self.paragraph, self.question, self.start_index, self.stop_index, self.attention_Label, \
self.POS_Embeddings, self.POS_Q_Embeddings = self.dataset.get_next_batch()
self.cell_Q_Enc_fw = tf.nn.rnn_cell.BasicLSTMCell(100)
self.cell_Q_Enc_bw = tf.nn.rnn_cell.BasicLSTMCell(100)
self.cell_P_Enc_fw = tf.nn.rnn_cell.BasicLSTMCell(100)
self.cell_P_Enc_bw = tf.nn.rnn_cell.BasicLSTMCell(100)
self.cell_POS_Enc_fw = tf.nn.rnn_cell.BasicLSTMCell(50)
self.cell_POS_Enc_bw = tf.nn.rnn_cell.BasicLSTMCell(50)
self.cell_POS_Enc_fw_Q = tf.nn.rnn_cell.BasicLSTMCell(50)
self.cell_POS_Enc_bw_Q = tf.nn.rnn_cell.BasicLSTMCell(50)
self.cell_output_Index_fw = tf.nn.rnn_cell.BasicLSTMCell(1)
self.cell_output_Index_bw = tf.nn.rnn_cell.BasicLSTMCell(1)
self.cell_modelling_fw = tf.nn.rnn_cell.BasicLSTMCell(1)
self.cell_modelling_bw = tf.nn.rnn_cell.BasicLSTMCell(1)
self.W_conv_2 = self.weight_variable([2, 1, 1, 4])
self.W_conv_3 = self.weight_variable([3, 1, 1, 4])
self.W_conv_4 = self.weight_variable([4, 1, 1, 4])
self.W_conv_5 = self.weight_variable([5, 1, 1, 4])
self.W_fc = self.weight_variable([16, 1])
self.b_fc = self.bias_variable([1])
self.output_Start = None
self.output_Stop = None
self.x_q_holer = tf.placeholder(
dtype=tf.float32,
shape=[self.batch, self.q_length, self.embedding_size],
name='x_q_holer')
self.x_p_holer = tf.placeholder(
dtype=tf.float32,
shape=[self.batch, self.p_length, self.embedding_size],
name='x_p_holer')
self.x_pos_q_holder = tf.placeholder(
dtype=tf.float32,
shape=[self.batch, self.q_length, self.POS_Embedding_Size],
name='x_POS_holer')
self.x_pos_holder = tf.placeholder(
dtype=tf.float32,
shape=[self.batch, self.p_length, self.POS_Embedding_Size],
name='x_POS_holer')
self.P_Length = 100