基于tensorflow的LSTM实现PTB预测

#%%
# Copyright 2016 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================import time
import numpy as np
import tensorflow as tf
import reader#flags = tf.flags
#logging = tf.logging#flags.DEFINE_string("save_path", None,
#                    "Model output directory.")
#flags.DEFINE_bool("use_fp16", False,
#                  "Train using 16-bit floats instead of 32bit floats")#FLAGS = flags.FLAGS#def data_type():
#  return tf.float16 if FLAGS.use_fp16 else tf.float32class PTBInput(object):"""The input data."""def __init__(self, config, data, name=None):self.batch_size = batch_size = config.batch_sizeself.num_steps = num_steps = config.num_stepsself.epoch_size = ((len(data) // batch_size) - 1) // num_stepsself.input_data, self.targets = reader.ptb_producer(data, batch_size, num_steps, name=name)class PTBModel(object):"""The PTB model."""def __init__(self, is_training, config, input_):self._input = input_batch_size = input_.batch_sizenum_steps = input_.num_stepssize = config.hidden_sizevocab_size = config.vocab_size# Slightly better results can be obtained with forget gate biases# initialized to 1 but the hyperparameters of the model would need to be# different than reported in the paper.def lstm_cell():return tf.contrib.rnn.BasicLSTMCell(size, forget_bias=0.0, state_is_tuple=True)attn_cell = lstm_cellif is_training and config.keep_prob < 1:def attn_cell():return tf.contrib.rnn.DropoutWrapper(lstm_cell(), output_keep_prob=config.keep_prob)cell = tf.contrib.rnn.MultiRNNCell([attn_cell() for _ in range(config.num_layers)], state_is_tuple=True)self._initial_state = cell.zero_state(batch_size, tf.float32)with tf.device("/cpu:0"):embedding = tf.get_variable("embedding", [vocab_size, size], dtype=tf.float32)inputs = tf.nn.embedding_lookup(embedding, input_.input_data)if is_training and config.keep_prob < 1:inputs = tf.nn.dropout(inputs, config.keep_prob)# Simplified version of models/tutorials/rnn/rnn.py's rnn().# This builds an unrolled LSTM for tutorial purposes only.# In general, use the rnn() or state_saving_rnn() from rnn.py.## The alternative version of the code below is:## inputs = tf.unstack(inputs, num=num_steps, axis=1)# outputs, state = tf.nn.rnn(cell, inputs,#                            initial_state=self._initial_state)outputs = []state = self._initial_statewith tf.variable_scope("RNN"):for time_step in range(num_steps):if time_step > 0: tf.get_variable_scope().reuse_variables()(cell_output, state) = cell(inputs[:, time_step, :], state)outputs.append(cell_output)output = tf.reshape(tf.concat(outputs, 1), [-1, size])self.output=outputsoftmax_w = tf.get_variable("softmax_w", [size, vocab_size], dtype=tf.float32)softmax_b = tf.get_variable("softmax_b", [vocab_size], dtype=tf.float32)logits = tf.matmul(output, softmax_w) + softmax_bloss = tf.contrib.legacy_seq2seq.sequence_loss_by_example([logits],[tf.reshape(input_.targets, [-1])],[tf.ones([batch_size * num_steps], dtype=tf.float32)])self._cost = cost = tf.reduce_sum(loss) / batch_sizeself._final_state = stateif not is_training:returnself._lr = tf.Variable(0.0, trainable=False)tvars = tf.trainable_variables()grads, _ = tf.clip_by_global_norm(tf.gradients(cost, tvars),config.max_grad_norm)optimizer = tf.train.GradientDescentOptimizer(self._lr)self._train_op = optimizer.apply_gradients(zip(grads, tvars),global_step=tf.contrib.framework.get_or_create_global_step())self._new_lr = tf.placeholder(tf.float32, shape=[], name="new_learning_rate")self._lr_update = tf.assign(self._lr, self._new_lr)def assign_lr(self, session, lr_value):session.run(self._lr_update, feed_dict={self._new_lr: lr_value})@propertydef input(self):return self._input@propertydef initial_state(self):return self._initial_state@propertydef cost(self):return self._cost@propertydef final_state(self):return self._final_state@propertydef lr(self):return self._lr@propertydef train_op(self):return self._train_opclass SmallConfig(object):"""Small config."""init_scale = 0.1learning_rate = 1.0max_grad_norm = 5num_layers = 2num_steps = 20hidden_size = 200max_epoch = 4max_max_epoch = 13keep_prob = 1.0lr_decay = 0.5batch_size = 20vocab_size = 10000class MediumConfig(object):"""Medium config."""init_scale = 0.05learning_rate = 1.0max_grad_norm = 5num_layers = 2num_steps = 35hidden_size = 650max_epoch = 6max_max_epoch = 39keep_prob = 0.5lr_decay = 0.8batch_size = 20vocab_size = 10000class LargeConfig(object):"""Large config."""init_scale = 0.04learning_rate = 1.0max_grad_norm = 10num_layers = 2num_steps = 35hidden_size = 1500max_epoch = 14max_max_epoch = 55keep_prob = 0.35lr_decay = 1 / 1.15batch_size = 20vocab_size = 10000class TestConfig(object):"""Tiny config, for testing."""init_scale = 0.1learning_rate = 1.0max_grad_norm = 1num_layers = 1num_steps = 2hidden_size = 2max_epoch = 1max_max_epoch = 1keep_prob = 1.0lr_decay = 0.5batch_size = 20vocab_size = 10000def run_epoch(session, model, eval_op=None, verbose=False):"""Runs the model on the given data."""start_time = time.time()costs = 0.0iters = 0state = session.run(model.initial_state)fetches = {"cost": model.cost,"final_state": model.final_state,}if eval_op is not None:fetches["eval_op"] = eval_opfor step in range(model.input.epoch_size):feed_dict = {}for i, (c, h) in enumerate(model.initial_state):feed_dict[c] = state[i].cfeed_dict[h] = state[i].hvals = session.run(fetches, feed_dict)cost = vals["cost"]state = vals["final_state"]costs += costiters += model.input.num_stepsif verbose and step % (model.input.epoch_size // 10) == 10:print("%.3f perplexity: %.3f speed: %.0f wps" %(step * 1.0 / model.input.epoch_size, np.exp(costs / iters),iters * model.input.batch_size / (time.time() - start_time)))return np.exp(costs / iters)raw_data = reader.ptb_raw_data('simple-examples/data/')
train_data, valid_data, test_data, _ = raw_dataconfig = SmallConfig()
eval_config = SmallConfig()
eval_config.batch_size = 1
eval_config.num_steps = 1with tf.Graph().as_default():initializer = tf.random_uniform_initializer(-config.init_scale,config.init_scale)with tf.name_scope("Train"):train_input = PTBInput(config=config, data=train_data, name="TrainInput")with tf.variable_scope("Model", reuse=None, initializer=initializer):m = PTBModel(is_training=True, config=config, input_=train_input)#tf.scalar_summary("Training Loss", m.cost)#tf.scalar_summary("Learning Rate", m.lr)with tf.name_scope("Valid"):valid_input = PTBInput(config=config, data=valid_data, name="ValidInput")with tf.variable_scope("Model", reuse=True, initializer=initializer):mvalid = PTBModel(is_training=False, config=config, input_=valid_input)#tf.scalar_summary("Validation Loss", mvalid.cost)with tf.name_scope("Test"):test_input = PTBInput(config=eval_config, data=test_data, name="TestInput")with tf.variable_scope("Model", reuse=True, initializer=initializer):mtest = PTBModel(is_training=False, config=eval_config,input_=test_input)sv = tf.train.Supervisor()with sv.managed_session() as session:for i in range(config.max_max_epoch):lr_decay = config.lr_decay ** max(i + 1 - config.max_epoch, 0.0)m.assign_lr(session, config.learning_rate * lr_decay)print("Epoch: %d Learning rate: %.3f" % (i + 1, session.run(m.lr)))train_perplexity = run_epoch(session, m, eval_op=m.train_op,verbose=True)print("Epoch: %d Train Perplexity: %.3f" % (i + 1, train_perplexity))valid_perplexity = run_epoch(session, mvalid)print("Epoch: %d Valid Perplexity: %.3f" % (i + 1, valid_perplexity))test_perplexity = run_epoch(session, mtest)print("Test Perplexity: %.3f" % test_perplexity)# if FLAGS.save_path:#   print("Saving model to %s." % FLAGS.save_path)#   sv.saver.save(session, FLAGS.save_path, global_step=sv.global_step)#if __name__ == "__main__":
#  tf.app.run()

将reader.py放到代码相同目录下

# Copyright 2015 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# =============================================================================="""Utilities for parsing PTB text files."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_functionimport collections
import os
import sysimport tensorflow as tfPy3 = sys.version_info[0] == 3def _read_words(filename):with tf.gfile.GFile(filename, "r") as f:if Py3:return f.read().replace("\n", "<eos>").split()else:return f.read().decode("utf-8").replace("\n", "<eos>").split()def _build_vocab(filename):data = _read_words(filename)counter = collections.Counter(data)count_pairs = sorted(counter.items(), key=lambda x: (-x[1], x[0]))words, _ = list(zip(*count_pairs))word_to_id = dict(zip(words, range(len(words))))return word_to_iddef _file_to_word_ids(filename, word_to_id):data = _read_words(filename)return [word_to_id[word] for word in data if word in word_to_id]def ptb_raw_data(data_path=None):"""Load PTB raw data from data directory "data_path".Reads PTB text files, converts strings to integer ids,and performs mini-batching of the inputs.The PTB dataset comes from Tomas Mikolov's webpage:http://www.fit.vutbr.cz/~imikolov/rnnlm/simple-examples.tgzArgs:data_path: string path to the directory where simple-examples.tgz hasbeen extracted.Returns:tuple (train_data, valid_data, test_data, vocabulary)where each of the data objects can be passed to PTBIterator."""train_path = os.path.join(data_path, "ptb.train.txt")valid_path = os.path.join(data_path, "ptb.valid.txt")test_path = os.path.join(data_path, "ptb.test.txt")word_to_id = _build_vocab(train_path)train_data = _file_to_word_ids(train_path, word_to_id)valid_data = _file_to_word_ids(valid_path, word_to_id)test_data = _file_to_word_ids(test_path, word_to_id)vocabulary = len(word_to_id)return train_data, valid_data, test_data, vocabularydef ptb_producer(raw_data, batch_size, num_steps, name=None):"""Iterate on the raw PTB data.This chunks up raw_data into batches of examples and returns Tensors thatare drawn from these batches.Args:raw_data: one of the raw data outputs from ptb_raw_data.batch_size: int, the batch size.num_steps: int, the number of unrolls.name: the name of this operation (optional).Returns:A pair of Tensors, each shaped [batch_size, num_steps]. The second elementof the tuple is the same data time-shifted to the right by one.Raises:tf.errors.InvalidArgumentError: if batch_size or num_steps are too high."""with tf.name_scope(name, "PTBProducer", [raw_data, batch_size, num_steps]):raw_data = tf.convert_to_tensor(raw_data, name="raw_data", dtype=tf.int32)data_len = tf.size(raw_data)batch_len = data_len // batch_sizedata = tf.reshape(raw_data[0 : batch_size * batch_len],[batch_size, batch_len])epoch_size = (batch_len - 1) // num_stepsassertion = tf.assert_positive(epoch_size,message="epoch_size == 0, decrease batch_size or num_steps")with tf.control_dependencies([assertion]):epoch_size = tf.identity(epoch_size, name="epoch_size")i = tf.train.range_input_producer(epoch_size, shuffle=False).dequeue()x = tf.strided_slice(data, [0, i * num_steps],[batch_size, (i + 1) * num_steps])x.set_shape([batch_size, num_steps])y = tf.strided_slice(data, [0, i * num_steps + 1],[batch_size, (i + 1) * num_steps + 1])y.set_shape([batch_size, num_steps])return x, y

将simple-examples.tgz下载解压，将其中的simple-examples文件夹放到代码相同目录，然后运行，

C:\Users\Administrator\AppData\Local\Programs\Python\Python36\python.exe D:/Deeplearning/projects/tensorflow_projects/7_2_LSTM.py
C:\Users\Administrator\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\dtypes.py:469: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'._np_qint8 = np.dtype([("qint8", np.int8, 1)])
C:\Users\Administrator\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\dtypes.py:470: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'._np_quint8 = np.dtype([("quint8", np.uint8, 1)])
C:\Users\Administrator\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\dtypes.py:471: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'._np_qint16 = np.dtype([("qint16", np.int16, 1)])
C:\Users\Administrator\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\dtypes.py:472: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'._np_quint16 = np.dtype([("quint16", np.uint16, 1)])
C:\Users\Administrator\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\dtypes.py:473: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'._np_qint32 = np.dtype([("qint32", np.int32, 1)])
C:\Users\Administrator\AppData\Local\Programs\Python\Python36\lib\site-packages\tensorflow\python\framework\dtypes.py:476: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.np_resource = np.dtype([("resource", np.ubyte, 1)])
WARNING:tensorflow:From D:/Deeplearning/projects/tensorflow_projects/7_2_LSTM.py:126: get_or_create_global_step (from tensorflow.contrib.framework.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Please switch to tf.train.get_or_create_global_step
2020-05-12 12:06:21.995000: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\platform\cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX AVX2
2020-05-12 12:06:22.100000: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1030] Found device 0 with properties:
name: GeForce GTX 1060 6GB major: 6 minor: 1 memoryClockRate(GHz): 1.7085
pciBusID: 0000:01:00.0
totalMemory: 6.00GiB freeMemory: 5.79GiB
2020-05-12 12:06:22.100000: I C:\tf_jenkins\home\workspace\rel-win\M\windows-gpu\PY\36\tensorflow\core\common_runtime\gpu\gpu_device.cc:1120] Creating TensorFlow device (/device:GPU:0) -> (device: 0, name: GeForce GTX 1060 6GB, pci bus id: 0000:01:00.0, compute capability: 6.1)
WARNING:tensorflow:Standard services need a 'logdir' passed to the SessionManager
Epoch: 1 Learning rate: 1.000
0.004 perplexity: 5851.399 speed: 6103 wps
0.104 perplexity: 838.206 speed: 10772 wps
0.204 perplexity: 625.268 speed: 10906 wps
0.304 perplexity: 505.829 speed: 10992 wps
0.404 perplexity: 436.651 speed: 11032 wps
0.504 perplexity: 391.628 speed: 11034 wps
0.604 perplexity: 352.968 speed: 11055 wps
0.703 perplexity: 326.436 speed: 11076 wps
0.803 perplexity: 305.440 speed: 11094 wps
0.903 perplexity: 286.137 speed: 11089 wps
Epoch: 1 Train Perplexity: 271.721
Epoch: 1 Valid Perplexity: 180.890
Epoch: 2 Learning rate: 1.000
0.004 perplexity: 214.130 speed: 10784 wps
0.104 perplexity: 152.788 speed: 11148 wps
0.204 perplexity: 160.312 speed: 11108 wps
0.304 perplexity: 155.377 speed: 11073 wps
0.404 perplexity: 152.412 speed: 11081 wps
0.504 perplexity: 150.029 speed: 11058 wps
0.604 perplexity: 145.367 speed: 11058 wps
0.703 perplexity: 143.269 speed: 11072 wps
0.803 perplexity: 141.295 speed: 11093 wps
0.903 perplexity: 137.624 speed: 11104 wps
Epoch: 2 Train Perplexity: 135.525
Epoch: 2 Valid Perplexity: 145.000
Epoch: 3 Learning rate: 1.000
0.004 perplexity: 147.102 speed: 10758 wps
0.104 perplexity: 106.609 speed: 11249 wps
0.204 perplexity: 115.864 speed: 11178 wps
0.304 perplexity: 113.114 speed: 11193 wps
0.404 perplexity: 111.991 speed: 11209 wps
0.504 perplexity: 111.132 speed: 11193 wps
0.604 perplexity: 108.429 speed: 11190 wps
0.703 perplexity: 107.791 speed: 11179 wps
0.803 perplexity: 107.106 speed: 11178 wps
0.903 perplexity: 104.823 speed: 11165 wps
Epoch: 3 Train Perplexity: 103.823
Epoch: 3 Valid Perplexity: 133.848
Epoch: 4 Learning rate: 1.000
0.004 perplexity: 117.925 speed: 10918 wps
0.104 perplexity: 86.010 speed: 11162 wps
0.204 perplexity: 94.614 speed: 11159 wps
0.304 perplexity: 92.540 speed: 11139 wps
0.404 perplexity: 92.043 speed: 11142 wps
0.504 perplexity: 91.642 speed: 11143 wps
0.604 perplexity: 89.705 speed: 11141 wps
0.703 perplexity: 89.507 speed: 11131 wps
0.803 perplexity: 89.214 speed: 11133 wps
0.903 perplexity: 87.536 speed: 11134 wps
Epoch: 4 Train Perplexity: 86.976
Epoch: 4 Valid Perplexity: 128.205
Epoch: 5 Learning rate: 0.500
0.004 perplexity: 99.482 speed: 11518 wps
0.104 perplexity: 71.572 speed: 11135 wps
0.204 perplexity: 77.783 speed: 11117 wps
0.304 perplexity: 75.010 speed: 11095 wps
0.404 perplexity: 73.936 speed: 11080 wps
0.504 perplexity: 72.969 speed: 11070 wps
0.604 perplexity: 70.799 speed: 11084 wps
0.703 perplexity: 70.063 speed: 11085 wps
0.803 perplexity: 69.277 speed: 11094 wps
0.903 perplexity: 67.383 speed: 11091 wps
Epoch: 5 Train Perplexity: 66.387
Epoch: 5 Valid Perplexity: 119.345
Epoch: 6 Learning rate: 0.250
0.004 perplexity: 80.949 speed: 11253 wps
0.104 perplexity: 59.010 speed: 11037 wps
0.204 perplexity: 64.474 speed: 11000 wps
0.304 perplexity: 62.112 speed: 10997 wps
0.404 perplexity: 61.141 speed: 11029 wps
0.504 perplexity: 60.229 speed: 11038 wps
0.604 perplexity: 58.343 speed: 11016 wps
0.703 perplexity: 57.627 speed: 11033 wps
0.803 perplexity: 56.841 speed: 11044 wps
0.903 perplexity: 55.117 speed: 11046 wps
Epoch: 6 Train Perplexity: 54.139
Epoch: 6 Valid Perplexity: 118.480
Epoch: 7 Learning rate: 0.125
0.004 perplexity: 71.864 speed: 11168 wps
0.104 perplexity: 52.413 speed: 11129 wps
0.204 perplexity: 57.320 speed: 11180 wps
0.304 perplexity: 55.204 speed: 11151 wps
0.404 perplexity: 54.303 speed: 11095 wps
0.504 perplexity: 53.469 speed: 11104 wps
0.604 perplexity: 51.751 speed: 11066 wps
0.703 perplexity: 51.083 speed: 11001 wps
0.803 perplexity: 50.330 speed: 10912 wps
0.903 perplexity: 48.729 speed: 10782 wps
Epoch: 7 Train Perplexity: 47.803
Epoch: 7 Valid Perplexity: 119.730
Epoch: 8 Learning rate: 0.062
0.004 perplexity: 67.433 speed: 11311 wps
0.104 perplexity: 49.194 speed: 10756 wps
0.204 perplexity: 53.766 speed: 10837 wps
0.304 perplexity: 51.729 speed: 10921 wps
0.404 perplexity: 50.863 speed: 10930 wps
0.504 perplexity: 50.067 speed: 10937 wps
0.604 perplexity: 48.428 speed: 10921 wps
0.703 perplexity: 47.775 speed: 10935 wps
0.803 perplexity: 47.038 speed: 10924 wps
0.903 perplexity: 45.496 speed: 10902 wps
Epoch: 8 Train Perplexity: 44.603
Epoch: 8 Valid Perplexity: 120.403
Epoch: 9 Learning rate: 0.031
0.004 perplexity: 64.861 speed: 11055 wps
0.104 perplexity: 47.546 speed: 11043 wps
0.204 perplexity: 51.967 speed: 10913 wps
0.304 perplexity: 49.953 speed: 10559 wps
0.404 perplexity: 49.107 speed: 10547 wps
0.504 perplexity: 48.320 speed: 10557 wps
0.604 perplexity: 46.724 speed: 10610 wps
0.703 perplexity: 46.066 speed: 10670 wps
0.803 perplexity: 45.328 speed: 10714 wps
0.903 perplexity: 43.817 speed: 10738 wps
Epoch: 9 Train Perplexity: 42.936
Epoch: 9 Valid Perplexity: 120.310
Epoch: 10 Learning rate: 0.016
0.004 perplexity: 63.397 speed: 10577 wps
0.104 perplexity: 46.576 speed: 10926 wps
0.204 perplexity: 50.966 speed: 10927 wps
0.304 perplexity: 48.964 speed: 10886 wps
0.404 perplexity: 48.134 speed: 10857 wps
0.504 perplexity: 47.355 speed: 10895 wps
0.604 perplexity: 45.795 speed: 10895 wps
0.703 perplexity: 45.138 speed: 10849 wps
0.803 perplexity: 44.399 speed: 10865 wps
0.903 perplexity: 42.907 speed: 10879 wps
Epoch: 10 Train Perplexity: 42.034
Epoch: 10 Valid Perplexity: 119.880
Epoch: 11 Learning rate: 0.008
0.004 perplexity: 62.544 speed: 11253 wps
0.104 perplexity: 45.981 speed: 10952 wps
0.204 perplexity: 50.366 speed: 10979 wps
0.304 perplexity: 48.381 speed: 11025 wps
0.404 perplexity: 47.563 speed: 11005 wps
0.504 perplexity: 46.790 speed: 11007 wps
0.604 perplexity: 45.252 speed: 11011 wps
0.703 perplexity: 44.602 speed: 11022 wps
0.803 perplexity: 43.871 speed: 11013 wps
0.903 perplexity: 42.392 speed: 11001 wps
Epoch: 11 Train Perplexity: 41.527
Epoch: 11 Valid Perplexity: 119.427
Epoch: 12 Learning rate: 0.004
0.004 perplexity: 62.054 speed: 10973 wps
0.104 perplexity: 45.632 speed: 10424 wps
0.204 perplexity: 50.013 speed: 10452 wps
0.304 perplexity: 48.048 speed: 10299 wps
0.404 perplexity: 47.240 speed: 10300 wps
0.504 perplexity: 46.472 speed: 10436 wps
0.604 perplexity: 44.947 speed: 10523 wps
0.703 perplexity: 44.301 speed: 10582 wps
0.803 perplexity: 43.579 speed: 10621 wps
0.903 perplexity: 42.108 speed: 10657 wps
Epoch: 12 Train Perplexity: 41.247
Epoch: 12 Valid Perplexity: 119.145
Epoch: 13 Learning rate: 0.002
0.004 perplexity: 61.794 speed: 10945 wps
0.104 perplexity: 45.438 speed: 10817 wps
0.204 perplexity: 49.814 speed: 10845 wps
0.304 perplexity: 47.862 speed: 10831 wps
0.404 perplexity: 47.062 speed: 10767 wps
0.504 perplexity: 46.300 speed: 10803 wps
0.604 perplexity: 44.782 speed: 10825 wps
0.703 perplexity: 44.139 speed: 10831 wps
0.803 perplexity: 43.422 speed: 10864 wps
0.903 perplexity: 41.956 speed: 10874 wps
Epoch: 13 Train Perplexity: 41.097
Epoch: 13 Valid Perplexity: 119.008
Test Perplexity: 114.313Process finished with exit code 0

基于tensorflow的LSTM实现PTB预测相关推荐

TensorFlow搭建LSTM实现时间序列预测（负荷预测）
目录 I. 前言 II. 数据处理 III. 模型 IV. 训练/测试 V. 源码及数据 I. 前言前面已经写过不少时间序列预测的文章: 深入理解PyTorch中LSTM的输入和输出(从input输 ...
基于Keras的LSTM多变量时间序列预测（北京PM2.5数据集pollution.csv）
基于Keras的LSTM多变量时间序列预测传统的线性模型难以解决多变量或多输入问题,而神经网络如LSTM则擅长于处理多个变量的问题,该特性使 ...
【基于MATLAB实现LSTM光伏输出功率预测】
基于MATLAB实现LSTM(长短记忆网络)光伏输出功率预测背景近年来,光伏市场在全世界的规模迅速增长.2020 年中国宣布采取新政策来提高国家的贡献力,目标是在2030 年前实现碳达峰,在206 ...
情感分析之电影评论分析-基于Tensorflow的LSTM
1. 深度学习在自然语言处理中的应用自然语言处理是教会机器如何去处理或者读懂人类语言的系统,目前比较热门的方向,包括如下几类: 对话系统 - 比较著名的案例有:Siri,Alexa 和 Cortan ...
基于Keras的LSTM多变量时间序列预测
LSTM是一种时间递归神经网络,它出现的原因是为了解决RNN的一个致命的缺陷.原生的RNN会遇到一个很大的问题,叫做The vanishing gradient problem for RNNs,也就 ...
如何基于TensorFlow使用LSTM和CNN实现时序分类任务
https://www.jiqizhixin.com/articles/2017-09-12-5 By 蒋思源2017年9月12日 09:54 时序数据经常出现在很多领域中,如金融.信号处理.语音识别 ...
【人工智能笔记】第五节：基于TensorFlow 2.0进行股票预测(JIT与Eager双模式实现)
该模型是典型的数据预测模型.实现多参数输入含时序,预测多个结果数据. 输入维度(batch_size,历史数据长度,输入参数数量),这里取15个维度,包含:年.月.日.上证指数.深证指数与目标股票(开 ...
基于Keras的LSTM多变量时间序列预测，包括单步和多步（北京PM2.5数据集pollution.csv）
直接简单点转自其他博客大神 Jason Brownlee的杰作:https://machinelearningmastery.com/multivariate-time-series-forecas ...
基于keras 搭建LSTM GRU模型预测共享单车使用情况完整代码+数据数据分析计算机毕设
项目运行教程:https://www.bilibili.com/video/BV1nT411k7dT/?spm_id_from=333.999.0.0 附完整代码数据:
手把手教你：基于LSTM的股票预测系统
系列文章第七章.手把手教你:基于深度残差网络(ResNet)的水果分类识别系统第六章.手把手教你:人脸识别的视频打码第五章.手把手教你:基于深度学习的滚动轴承故障诊断目录系列文章一.项目简 ...

基于tensorflow的LSTM实现PTB预测

基于tensorflow的LSTM实现PTB预测相关推荐

最新文章

热门文章