实现beam_search部分
基于tensorflow1.4 Seq2seq的实现
import helpers
import tensorflow as tf
from tensorflow.python.util import nest
from tensorflow.contrib import seq2seq,rnn
tf.__version__
tf.reset_default_graph()
sess = tf.InteractiveSession()
PAD = 0
EOS = 1
vocab_size = 10
input_embedding_size = 20
encoder_hidden_units = 25
decoder_hidden_units = encoder_hidden_units
import helpers as data_helpers
batch_size = 10
# 一个generator，每次产生一个minibatch的随机样本
batches = data_helpers.random_sequences(length_from=3, length_to=8,
vocab_lower=2, vocab_upper=10,
batch_size=batch_size)
print('产生%d个长度不一（最短3，最长8）的sequences, 其中前十个是:' % batch_size)
for seq in next(batches)[:min(batch_size, 10)]:
print(seq)
tf.reset_default_graph()
sess = tf.InteractiveSession()
mode = tf.contrib.learn.ModeKeys.TRAIN
产生10个长度不一（最短3，最长8）的sequences, 其中前十个是:
[8, 2, 7, 2]
[6, 7, 9, 2, 5, 3]
[3, 7, 5, 3, 2, 3, 4]
[7, 4, 6, 4, 2]
[9, 5, 5, 5, 7, 8]
[8, 7, 7, 6, 8, 6, 2]
[9, 6, 3, 5, 3, 8, 5, 4]
[7, 8, 4, 8]
[7, 8, 7, 5, 4]
[4, 7, 9, 4, 2, 7]
1.使用seq2seq库实现seq2seq模型
1. 计算图的数据的placeholder
with tf.name_scope('minibatch'):
encoder_inputs = tf.placeholder(tf.int32, [None, None], name='encoder_inputs')
encoder_inputs_length = tf.placeholder(tf.int32, [None], name='encoder_inputs_length')
decoder_targets = tf.placeholder(tf.int32, [None, None], name='decoder_targets')
decoder_inputs = tf.placeholder(shape=(None, None),dtype=tf.int32,name='decoder_inputs')
#decoder_inputs_length和decoder_targets_length是一样的
decoder_inputs_length = tf.placeholder(shape=(None,),
dtype=tf.int32,
name='decoder_inputs_length')
def _create_rnn_cell():
def single_rnn_cell(encoder_hidden_units):
# 创建单个cell，这里需要注意的是一定要使用一个single_rnn_cell的函数，不然直接把cell放在MultiRNNCell
# 的列表中最终模型会发生错误
single_cell = rnn.LSTMCell(encoder_hidden_units)
#添加dropout
single_cell = rnn.DropoutWrapper(single_cell, output_keep_prob=0.5)
return single_cell
#列表中每个元素都是调用single_rnn_cell函数
#cell = rnn.MultiRNNCell([single_rnn_cell() for _ in range(self.num_layers)])
cell = rnn.MultiRNNCell([single_rnn_cell(encoder_hidden_units) for _ in range(1)])
return cell
2.定义encoder 部分
with tf.variable_scope('encoder'):
# 创建LSTMCell
encoder_cell = _create_rnn_cell()
# 构建embedding矩阵,encoder和decoder公用该词向量矩阵
embedding = tf.get_variable('embedding', [vocab_size,input_embedding_size])
encoder_inputs_embedded = tf.nn.embedding_lookup(embedding,encoder_inputs)
# 使用dynamic_rnn构建LSTM模型，将输入编码成隐层向量。
# encoder_outputs用于attention，batch_size*encoder_inputs_length*rnn_size,
# encoder_state用于decoder的初始化状态，batch_size*rnn_szie
encoder_outputs, encoder_state = tf.nn.dynamic_rnn(encoder_cell, encoder_inputs_embedded,
sequence_length=encoder_inputs_length,
dtype=tf.float32)
3.定义decoder 部分(训练阶段)
with tf.variable_scope('decoder'):
decoder_cell = _create_rnn_cell()
#定义decoder的初始状态
decoder_initial_state = encoder_state
#定义output_layer
output_layer = tf.layers.Dense(vocab_size,kernel_initializer=tf.truncated_normal_initializer(mean=0.0, stddev=0.1))
decoder_inputs_embedded = tf.nn.embedding_lookup(embedding, decoder_inputs)
# 训练阶段，使用TrainingHelper+BasicDecoder的组合，这一般是固定的，当然也可以自己定义Helper类，实现自己的功能
training_helper = seq2seq.TrainingHelper(inputs=decoder_inputs_embedded,
sequence_length=decoder_inputs_length,
time_major=False, name='training_helper')
training_decoder = seq2seq.BasicDecoder(cell=decoder_cell, helper=training_helper,
initial_state=decoder_initial_state,
output_layer=output_layer)
# 调用dynamic_decode进行解码，decoder_outputs是一个namedtuple，里面包含两项(rnn_outputs, sample_id)
# rnn_output: [batch_size, decoder_targets_length, vocab_size]，保存decode每个时刻每个单词的概率，可以用来计算loss
# sample_id: [batch_size], tf.int32，保存最终的编码结果。可以表示最后的答案
max_target_sequence_length = tf.reduce_max(decoder_inputs_length, name='max_target_len')
decoder_outputs, _, _ = seq2seq.dynamic_decode(decoder=training_decoder,
impute_finished=True,
maximum_iterations=max_target_sequence_length)
decoder_logits_train = tf.identity(decoder_outputs.rnn_output)
sample_id = decoder_outputs.sample_id
max_target_sequence_length = tf.reduce_max(decoder_inputs_length, name='max_target_len')
mask = tf.sequence_mask(decoder_inputs_length,max_target_sequence_length, dtype=tf.float32, name='masks')
print('\t%s' % repr(decoder_logits_train))
print('\t%s' % repr(decoder_targets))
print('\t%s' % repr(sample_id))
loss = seq2seq.sequence_loss(logits=decoder_logits_train,targets=decoder_targets, weights=mask)
3.定义decoder 部分(测试阶段)
beam_search.png
with tf.variable_scope('decoder',reuse=True):
start_tokens = tf.ones([batch_size, ], tf.int32)*1 #[batch_size] 数值为1
encoder_state = nest.map_structure(lambda s: seq2seq.tile_batch(s, 3),
encoder_state)
inference_decoder = tf.contrib.seq2seq.BeamSearchDecoder(cell=decoder_cell, embedding=embedding,
start_tokens=start_tokens,
end_token=1,
initial_state=encoder_state,
beam_width=3,
output_layer=output_layer)
beam_decoder_outputs, _, _ = seq2seq.dynamic_decode(decoder=inference_decoder,maximum_iterations=10)
train_op = tf.train.AdamOptimizer(learning_rate = 0.001).minimize(loss)
sess.run(tf.global_variables_initializer())
def next_feed():
batch = next(batches)
encoder_inputs_, encoder_inputs_length_ = data_helpers.batch(batch)
decoder_targets_, decoder_targets_length_ = data_helpers.batch(
[(sequence) + [EOS] for sequence in batch]
)
decoder_inputs_, decoder_inputs_length_ = data_helpers.batch(
[[EOS] + (sequence) for sequence in batch]
)
# 在feedDict里面，key可以是一个Tensor
return {
encoder_inputs: encoder_inputs_.T,
decoder_inputs: decoder_inputs_.T,
decoder_targets: decoder_targets_.T,
encoder_inputs_length: encoder_inputs_length_,
decoder_inputs_length: decoder_inputs_length_
}
x = next_feed()
print('encoder_inputs:')
print(x[encoder_inputs][0,:])
print('encoder_inputs_length:')
print(x[encoder_inputs_length][0])
print('decoder_inputs:')
print(x[decoder_inputs][0,:])
print('decoder_inputs_length:')
print(x[decoder_inputs_length][0])
print('decoder_targets:')
print(x[decoder_targets][0,:])
encoder_inputs:
[6 9 9 3 4 7 0 0]
encoder_inputs_length:
6
decoder_inputs:
[1 6 9 9 3 4 7 0 0]
decoder_inputs_length:
7
decoder_targets:
[6 9 9 3 4 7 1 0 0]
loss_track = []
max_batches = 6001
batches_in_epoch = 200
try:
# 一个epoch的learning
for batch in range(max_batches):
fd = next_feed()
_, l = sess.run([train_op, loss], fd)
loss_track.append(l)
if batch == 0 or batch % batches_in_epoch == 0:
print('batch {}'.format(batch))
print(' minibatch loss: {}'.format(sess.run(loss, fd)))
predict_ = sess.run(beam_decoder_outputs.predicted_ids, fd)
#print(predict_)
for i, (inp, pred) in enumerate(zip(fd[encoder_inputs], predict_)):
print(' sample {}:'.format(i + 1))
print(' input > {}'.format(inp))
print(' predicted > {}'.format(pred))
if i >= 2:
break
print()
except KeyboardInterrupt:
print('training interrupted')
batch 0
minibatch loss: 0.32762354612350464
sample 1:
input > [2 8 9 4 6 3 5 0]
predicted > [[ 2 2 2]
[ 8 8 8]
[ 9 9 9]
[ 4 4 4]
[ 6 6 6]
[ 3 5 9]
[ 5 3 3]
[ 1 1 1]
[-1 -1 -1]]
sample 2:
input > [2 5 4 2 6 0 0 0]
predicted > [[ 2 2 2]
[ 5 3 5]
[ 4 4 2]
[ 2 5 4]
[ 6 9 6]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [5 8 3 0 0 0 0 0]
predicted > [[ 5 5 5]
[ 8 3 2]
[ 3 8 8]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 200
minibatch loss: 0.23138172924518585
sample 1:
input > [4 9 2 0 0 0 0 0]
predicted > [[ 4 4 4]
[ 9 9 9]
[ 2 7 9]
[ 1 2 1]
[-1 1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [7 7 6 5 2 4 9 3]
predicted > [[7 7 7]
[7 7 7]
[6 6 6]
[5 5 2]
[4 2 5]
[2 4 9]
[3 9 4]
[9 3 5]
[1 1 1]]
sample 3:
input > [4 9 3 3 7 0 0 0]
predicted > [[ 4 4 4]
[ 9 9 9]
[ 3 3 3]
[ 3 7 3]
[ 7 3 9]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 400
minibatch loss: 0.21507926285266876
sample 1:
input > [7 7 8 7 2 2 3]
predicted > [[7 7 7]
[7 7 7]
[8 8 8]
[4 7 2]
[7 2 7]
[3 2 7]
[2 3 5]
[1 1 1]]
sample 2:
input > [2 7 9 6 0 0 0]
predicted > [[ 2 2 2]
[ 7 7 9]
[ 9 6 7]
[ 6 9 6]
[ 1 9 1]
[-1 1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [2 8 2 0 0 0 0]
predicted > [[ 2 8 8]
[ 8 2 8]
[ 2 2 2]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 600
minibatch loss: 0.3271256387233734
sample 1:
input > [2 9 9 3 9 8 5 0]
predicted > [[ 9 7 9]
[ 2 8 2]
[ 2 4 2]
[ 3 5 7]
[ 9 2 5]
[ 9 9 9]
[ 7 5 3]
[ 1 1 9]
[-1 -1 1]]
sample 2:
input > [8 4 7 5 4 0 0 0]
predicted > [[ 8 8 8]
[ 4 2 2]
[ 7 7 7]
[ 5 4 4]
[ 4 6 5]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [4 6 9 8 8 7 0 0]
predicted > [[ 4 4 4]
[ 6 9 9]
[ 9 6 6]
[ 8 8 8]
[ 8 8 8]
[ 7 3 6]
[ 1 6 1]
[-1 1 -1]
[-1 -1 -1]]
batch 800
minibatch loss: 0.3914913535118103
sample 1:
input > [3 5 5 9 0 0 0]
predicted > [[ 3 5 3]
[ 5 3 5]
[ 5 3 5]
[ 9 9 7]
[ 1 1 9]
[-1 -1 1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [4 5 4 8 4 5 2]
predicted > [[4 4 4]
[4 4 5]
[5 5 4]
[5 5 8]
[8 8 4]
[8 4 5]
[4 8 2]
[1 1 1]]
sample 3:
input > [9 4 7 5 6 5 0]
predicted > [[ 9 5 5]
[ 4 4 4]
[ 7 7 7]
[ 5 9 9]
[ 6 7 9]
[ 1 9 7]
[-1 1 1]
[-1 -1 -1]]
batch 1000
minibatch loss: 0.2255089282989502
sample 1:
input > [2 3 9 0 0 0 0 0]
predicted > [[ 2 8 2]
[ 3 3 3]
[ 9 2 7]
[ 1 4 1]
[-1 1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [5 9 8 0 0 0 0 0]
predicted > [[ 5 5 5]
[ 9 8 8]
[ 8 9 9]
[ 1 9 1]
[-1 1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [4 7 7 9 0 0 0 0]
predicted > [[ 4 4 4]
[ 7 7 9]
[ 7 9 7]
[ 9 7 7]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 1200
minibatch loss: 0.3723776340484619
sample 1:
input > [8 2 6 8 6 0 0 0]
predicted > [[ 8 8 8]
[ 2 2 4]
[ 6 6 8]
[ 8 8 5]
[ 6 1 8]
[ 1 -1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [4 6 3 7 6 6 6 0]
predicted > [[ 4 4 4]
[ 6 6 6]
[ 3 3 3]
[ 7 6 6]
[ 6 7 7]
[ 6 7 6]
[ 1 6 9]
[-1 1 1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [9 7 5 4 5 0 0 0]
predicted > [[ 9 9 9]
[ 7 7 7]
[ 5 2 2]
[ 4 9 9]
[ 5 5 5]
[ 1 1 6]
[-1 -1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 1400
minibatch loss: 0.2460956871509552
sample 1:
input > [2 6 2 0 0 0 0]
predicted > [[ 2 8 5]
[ 6 2 8]
[ 2 5 2]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [2 8 4 8 4 3 0]
predicted > [[ 2 2 2]
[ 4 8 4]
[ 8 4 8]
[ 8 8 8]
[ 8 4 2]
[ 3 3 3]
[ 1 1 8]
[-1 -1 1]
[-1 -1 -1]]
sample 3:
input > [3 6 2 8 3 6 5]
predicted > [[ 3 3 3]
[ 6 6 6]
[ 2 8 8]
[ 8 3 2]
[ 5 2 5]
[ 3 5 3]
[ 6 4 6]
[ 1 1 1]
[-1 -1 -1]]
batch 1600
minibatch loss: 0.21721991896629333
sample 1:
input > [2 4 7 5 3 0 0]
predicted > [[ 2 2 2]
[ 4 7 7]
[ 7 4 4]
[ 5 5 5]
[ 3 4 8]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [9 5 2 6 3 8 7]
predicted > [[5 5 5]
[9 9 9]
[8 8 8]
[2 2 7]
[7 7 5]
[7 5 2]
[5 7 2]
[1 1 1]]
sample 3:
input > [4 2 7 8 0 0 0]
predicted > [[ 4 2 8]
[ 2 4 4]
[ 8 7 7]
[ 7 3 2]
[ 1 1 4]
[-1 -1 1]
[-1 -1 -1]
[-1 -1 -1]]
batch 1800
minibatch loss: 0.2289828509092331
sample 1:
input > [8 8 2 0 0 0 0]
predicted > [[ 8 8 2]
[ 8 2 8]
[ 2 8 8]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [9 8 4 0 0 0 0]
predicted > [[ 9 8 9]
[ 8 9 9]
[ 4 9 2]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [2 2 4 5 0 0 0]
predicted > [[ 2 2 2]
[ 2 2 2]
[ 4 9 9]
[ 5 4 2]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 2000
minibatch loss: 0.35617268085479736
sample 1:
input > [6 9 2 4 0 0 0 0]
predicted > [[ 6 6 9]
[ 9 9 6]
[ 2 4 4]
[ 4 2 2]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [3 7 8 8 5 7 2 0]
predicted > [[ 3 3 3]
[ 7 7 8]
[ 8 8 7]
[ 8 8 7]
[ 2 2 5]
[ 7 7 2]
[ 9 5 8]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [2 6 9 0 0 0 0 0]
predicted > [[ 2 2 2]
[ 6 7 8]
[ 9 6 6]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 2200
minibatch loss: 0.44849294424057007
sample 1:
input > [3 7 5 6 7 9 0 0]
predicted > [[ 3 3 3]
[ 7 7 7]
[ 6 5 6]
[ 5 6 5]
[ 9 7 7]
[ 7 9 9]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [6 8 8 0 0 0 0 0]
predicted > [[ 8 6 8]
[ 6 8 6]
[ 6 8 5]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [7 6 6 7 2 6 3 2]
predicted > [[ 6 6 6]
[ 7 7 7]
[ 7 7 6]
[ 7 7 7]
[ 6 6 7]
[ 2 2 2]
[ 2 5 8]
[ 6 8 4]
[ 1 1 1]
[-1 -1 -1]]
batch 2400
minibatch loss: 0.16510817408561707
sample 1:
input > [4 3 9 2 4 8 9]
predicted > [[4 4 4]
[3 3 3]
[5 5 9]
[9 9 2]
[7 9 4]
[8 7 8]
[4 2 6]
[2 8 5]
[1 1 1]]
sample 2:
input > [4 2 5 0 0 0 0]
predicted > [[ 2 4 4]
[ 4 2 2]
[ 5 5 2]
[ 1 1 6]
[-1 -1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [2 6 2 9 0 0 0]
predicted > [[ 2 2 2]
[ 6 6 8]
[ 2 2 5]
[ 9 7 6]
[ 1 5 1]
[-1 1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 2600
minibatch loss: 0.18280677497386932
sample 1:
input > [9 4 9 0 0 0 0 0]
predicted > [[ 9 9 6]
[ 4 9 9]
[ 9 4 9]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [3 5 2 6 4 6 0 0]
predicted > [[ 3 3 5]
[ 5 5 3]
[ 6 6 7]
[ 2 2 2]
[ 4 2 6]
[ 2 4 8]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [8 5 9 4 2 0 0 0]
predicted > [[ 8 8 8]
[ 5 5 9]
[ 9 9 5]
[ 4 4 4]
[ 2 8 3]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 2800
minibatch loss: 0.32199400663375854
sample 1:
input > [7 2 6 3 6 0 0 0]
predicted > [[ 7 7 7]
[ 2 2 3]
[ 6 6 8]
[ 3 6 4]
[ 6 3 4]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [6 7 3 9 0 0 0 0]
predicted > [[ 7 6 7]
[ 6 7 6]
[ 1 3 8]
[-1 9 3]
[-1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [4 3 8 7 7 3 0 0]
predicted > [[ 4 4 4]
[ 3 3 3]
[ 8 8 8]
[ 7 9 7]
[ 7 7 3]
[ 3 3 7]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]]
batch 3000
minibatch loss: 0.48668527603149414
sample 1:
input > [3 7 2 2 4 0 0 0]
predicted > [[ 3 2 2]
[ 7 7 7]
[ 2 3 3]
[ 2 3 3]
[ 4 6 8]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [2 7 4 4 2 3 9 0]
predicted > [[ 2 2 2]
[ 7 7 4]
[ 4 4 7]
[ 4 4 2]
[ 2 2 7]
[ 3 3 4]
[ 9 4 6]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [5 5 8 4 3 7 6 0]
predicted > [[ 5 5 5]
[ 5 5 5]
[ 8 8 8]
[ 4 4 4]
[ 3 3 7]
[ 7 9 3]
[ 6 6 8]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]]
batch 3200
minibatch loss: 0.2466438114643097
sample 1:
input > [4 9 6 9 8 6 0]
predicted > [[ 4 4 4]
[ 9 9 9]
[ 6 8 6]
[ 9 5 8]
[ 8 6 9]
[ 6 6 9]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [7 4 8 9 5 2 6]
predicted > [[ 7 7 7]
[ 4 4 4]
[ 8 8 8]
[ 9 9 9]
[ 5 5 5]
[ 2 5 5]
[ 6 2 4]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [8 3 4 4 8 5 0]
predicted > [[ 8 8 8]
[ 3 3 7]
[ 4 4 5]
[ 4 4 3]
[ 8 9 4]
[ 5 6 8]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 3400
minibatch loss: 0.2961788773536682
sample 1:
input > [4 3 8 0 0 0 0 0]
predicted > [[ 4 4 4]
[ 3 3 7]
[ 8 2 8]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [4 6 2 0 0 0 0 0]
predicted > [[ 4 6 4]
[ 6 4 4]
[ 2 2 6]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [7 6 5 9 7 8 3 0]
predicted > [[ 7 7 7]
[ 6 6 6]
[ 5 5 5]
[ 8 9 8]
[ 7 7 7]
[ 9 8 9]
[ 9 3 7]
[ 1 1 1]
[-1 -1 -1]]
batch 3600
minibatch loss: 0.3043099641799927
sample 1:
input > [7 7 2 3 4 4 3 9]
predicted > [[7 7 7]
[7 7 7]
[2 2 2]
[4 4 4]
[3 3 3]
[3 5 9]
[9 7 3]
[4 8 4]
[1 1 1]]
sample 2:
input > [7 2 5 9 0 0 0 0]
predicted > [[ 7 7 7]
[ 2 5 8]
[ 5 2 5]
[ 9 9 4]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [6 3 5 0 0 0 0 0]
predicted > [[ 6 6 6]
[ 3 3 3]
[ 5 9 1]
[ 1 1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 3800
minibatch loss: 0.26044222712516785
sample 1:
input > [8 8 5 0 0 0 0 0]
predicted > [[ 8 8 2]
[ 8 8 8]
[ 5 2 7]
[ 1 1 8]
[-1 -1 6]
[-1 -1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [6 6 8 2 9 0 0 0]
predicted > [[ 6 6 6]
[ 6 6 8]
[ 8 8 6]
[ 2 8 4]
[ 9 5 9]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [4 5 7 2 4 0 0 0]
predicted > [[ 4 4 4]
[ 5 5 5]
[ 7 7 7]
[ 2 2 2]
[ 4 2 2]
[ 1 1 9]
[-1 -1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 4000
minibatch loss: 0.3985794484615326
sample 1:
input > [9 5 9 0 0 0 0 0]
predicted > [[ 5 9 9]
[ 9 9 5]
[ 9 5 9]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [8 5 3 6 8 6 0 0]
predicted > [[ 5 5 5]
[ 8 8 8]
[ 8 8 8]
[ 3 3 3]
[ 6 6 6]
[ 7 7 6]
[ 1 3 1]
[-1 1 -1]
[-1 -1 -1]]
sample 3:
input > [2 5 4 0 0 0 0 0]
predicted > [[ 2 2 2]
[ 5 2 6]
[ 4 6 4]
[ 1 1 5]
[-1 -1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 4200
minibatch loss: 0.29730188846588135
sample 1:
input > [8 3 9 9 7 0 0]
predicted > [[ 8 8 8]
[ 3 3 3]
[ 9 9 9]
[ 9 4 9]
[ 7 9 9]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [5 6 7 6 5 8 2]
predicted > [[5 5 5]
[6 6 6]
[7 7 7]
[6 6 6]
[5 3 5]
[8 8 8]
[8 4 2]
[1 1 1]]
sample 3:
input > [7 7 3 2 8 0 0]
predicted > [[ 7 7 7]
[ 7 7 7]
[ 3 2 8]
[ 2 3 3]
[ 8 8 2]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]]
batch 4400
minibatch loss: 0.35038599371910095
sample 1:
input > [4 7 7 6 0 0 0 0]
predicted > [[ 4 4 7]
[ 7 7 4]
[ 7 6 4]
[ 6 7 3]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [5 5 7 2 6 6 0 0]
predicted > [[ 5 5 5]
[ 5 5 5]
[ 9 7 7]
[ 3 2 6]
[ 6 6 2]
[ 6 6 2]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [8 2 5 0 0 0 0 0]
predicted > [[ 8 8 8]
[ 2 3 5]
[ 5 2 2]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 4600
minibatch loss: 0.2776692807674408
sample 1:
input > [7 8 9 9 0 0 0 0]
predicted > [[ 7 7 7]
[ 8 8 8]
[ 9 9 9]
[ 9 8 6]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [3 5 9 5 2 6 5 0]
predicted > [[ 3 3 5]
[ 5 5 3]
[ 9 9 5]
[ 5 5 7]
[ 2 8 8]
[ 6 5 2]
[ 9 2 3]
[ 1 1 9]
[-1 -1 1]
[-1 -1 -1]]
sample 3:
input > [6 3 7 9 5 2 0 0]
predicted > [[ 6 7 6]
[ 3 6 3]
[ 7 5 7]
[ 9 3 9]
[ 5 8 5]
[ 2 3 3]
[ 1 1 9]
[-1 -1 1]
[-1 -1 -1]
[-1 -1 -1]]
batch 4800
minibatch loss: 0.24092251062393188
sample 1:
input > [2 5 9 2 4 6 2 7]
predicted > [[2 2 2]
[5 5 5]
[9 4 9]
[2 9 2]
[4 2 2]
[6 5 4]
[2 8 6]
[7 7 3]
[1 1 1]]
sample 2:
input > [8 3 2 9 5 4 0 0]
predicted > [[ 8 8 8]
[ 3 3 3]
[ 2 2 2]
[ 9 5 9]
[ 5 9 5]
[ 4 2 2]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [7 6 4 2 2 0 0 0]
predicted > [[ 7 7 7]
[ 6 4 8]
[ 4 6 6]
[ 2 2 4]
[ 2 8 5]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 5000
minibatch loss: 0.24987684190273285
sample 1:
input > [2 8 4 2 2 0 0 0]
predicted > [[ 2 2 2]
[ 2 8 8]
[ 4 4 2]
[ 8 2 4]
[ 8 2 2]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [6 9 4 8 9 8 2 0]
predicted > [[ 6 6 6]
[ 9 9 9]
[ 4 4 4]
[ 8 8 8]
[ 9 9 9]
[ 8 8 3]
[ 2 2 8]
[ 1 1 1]
[ 1 -1 -1]]
sample 3:
input > [6 6 2 5 5 6 0 0]
predicted > [[ 6 6 6]
[ 5 6 6]
[ 6 5 5]
[ 2 2 2]
[ 6 2 2]
[ 2 5 6]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]]
batch 5200
minibatch loss: 0.2918197810649872
sample 1:
input > [5 4 8 5 0 0 0 0]
predicted > [[ 5 3 5]
[ 4 4 2]
[ 8 5 6]
[ 5 8 4]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [6 8 6 3 8 5 0 0]
predicted > [[ 8 6 6]
[ 6 8 8]
[ 6 6 6]
[ 3 3 3]
[ 6 8 6]
[ 6 5 2]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [4 3 6 3 9 2 0 0]
predicted > [[ 4 4 4]
[ 3 3 3]
[ 6 6 6]
[ 3 5 3]
[ 9 7 9]
[ 2 3 4]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 5400
minibatch loss: 0.15758396685123444
sample 1:
input > [3 2 7 0 0 0 0 0]
predicted > [[ 3 3 3]
[ 2 2 3]
[ 7 3 4]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [2 7 6 3 0 0 0 0]
predicted > [[ 2 2 2]
[ 7 7 1]
[ 6 1 -1]
[ 3 -1 -1]
[ 1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [5 5 7 5 0 0 0 0]
predicted > [[ 5 5 5]
[ 5 5 5]
[ 7 7 9]
[ 5 3 7]
[ 1 1 3]
[-1 -1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 5600
minibatch loss: 0.306735634803772
sample 1:
input > [9 2 2 2 2 4 0 0]
predicted > [[ 4 4 4]
[ 9 9 5]
[ 2 2 2]
[ 2 2 8]
[ 2 5 9]
[ 8 2 1]
[ 1 8 -1]
[-1 1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 2:
input > [4 2 4 5 3 9 7 4]
predicted > [[ 4 4 4]
[ 2 2 2]
[ 4 4 9]
[ 5 5 2]
[ 7 3 6]
[ 3 7 3]
[ 9 9 4]
[ 4 4 5]
[ 1 1 1]
[-1 -1 -1]]
sample 3:
input > [6 3 2 3 6 0 0 0]
predicted > [[ 6 6 6]
[ 3 3 3]
[ 4 2 3]
[ 3 3 2]
[ 2 6 7]
[ 5 1 1]
[ 1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 5800
minibatch loss: 0.4090876281261444
sample 1:
input > [6 6 8 5 5 7 9 8]
predicted > [[ 6 6 6]
[ 6 5 5]
[ 8 6 6]
[ 5 8 8]
[ 5 8 8]
[ 9 7 7]
[ 7 7 7]
[ 8 3 1]
[ 1 1 -1]]
sample 2:
input > [5 2 3 7 4 0 0 0]
predicted > [[ 5 5 5]
[ 2 2 4]
[ 3 3 3]
[ 7 4 2]
[ 4 7 7]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
sample 3:
input > [2 8 4 4 8 0 0 0]
predicted > [[ 2 2 2]
[ 8 4 8]
[ 4 8 4]
[ 4 8 8]
[ 8 4 4]
[ 1 1 1]
[-1 -1 -1]
[-1 -1 -1]
[-1 -1 -1]]
batch 6000
minibatch loss: 0.2645653188228607
sample 1:
input > [6 2 8 4 2 2 8]
predicted > [[ 6 6 6]
[ 5 2 2]
[ 2 6 8]
[ 8 2 2]
[ 2 4 4]
[ 4 5 2]
[ 8 8 8]
[ 1 8 1]
[-1 1 -1]]
sample 2:
input > [4 7 9 8 2 3 4]
predicted > [[ 4 4 4]
[ 7 7 7]
[ 9 9 9]
[ 8 8 8]
[ 2 2 2]
[ 5 3 3]
[ 3 4 5]
[ 1 1 1]
[-1 -1 -1]]
sample 3:
input > [3 4 9 2 6 8 3]
predicted > [[ 3 3 3]
[ 4 2 4]
[ 9 9 9]
[ 2 4 8]
[ 6 6 2]
[ 8 6 6]
[ 3 3 5]
[ 1 1 1]
[-1 -1 -1]]