adventuresinml / adventures-in-ml-code Goto Github PK

if target_network is None:
        updates[valid_idxs] += GAMMA * np.amax(prim_qtp1.numpy()[valid_idxs, :], axis=1)
    else:
        prim_action_tp1 = np.argmax(prim_qtp1.numpy(), axis=1)
        q_from_target = target_network(next_states)
        updates[valid_idxs] += GAMMA * q_from_target.numpy()[batch_idxs[valid_idxs], prim_action_tp1[valid_idxs]]

if target_network is None:
        updates[valid_idxs] += GAMMA * np.amax(prim_qtp1.numpy()[valid_idxs, :], axis=1)
    else:
        q_from_target = target_network(next_states)
        updates[valid_idxs] += GAMMA * np.amax(q_from_target.numpy()[valid_idxs, :], axis=1)        

/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/training.py:571 train_function  *
    outputs = self.distribute_strategy.run(
/usr/local/lib/python3.6/dist-packages/tensorflow/python/distribute/distribute_lib.py:951 run  **
    return self._extended.call_for_each_replica(fn, args=args, kwargs=kwargs)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/distribute/distribute_lib.py:2290 call_for_each_replica
    return self._call_for_each_replica(fn, args, kwargs)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/distribute/distribute_lib.py:2649 _call_for_each_replica
    return fn(*args, **kwargs)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/training.py:533 train_step  **
    y, y_pred, sample_weight, regularization_losses=self.losses)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/compile_utils.py:204 __call__
    loss_value = loss_obj(y_t, y_p, sample_weight=sw)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/losses.py:143 __call__
    losses = self.call(y_true, y_pred)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/losses.py:246 call
    return self.fn(y_true, y_pred, **self._fn_kwargs)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/losses.py:1527 categorical_crossentropy
    return K.categorical_crossentropy(y_true, y_pred, from_logits=from_logits)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/backend.py:4561 categorical_crossentropy
    target.shape.assert_is_compatible_with(output.shape)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/tensor_shape.py:1117 assert_is_compatible_with
    raise ValueError("Shapes %s and %s are incompatible" % (self, other))

ValueError: Shapes (24, 1) and (24, 2) are incompatible

ValueError                                Traceback (most recent call last)
<ipython-input-8-000aec5df542> in <module>()
     29 
     30     if steps > DELAY_TRAINING:
---> 31       loss = train(primary_network, memory, target_network)
     32       update_network(primary_network, target_network)
     33       _, error = get_per_error(tf.reshape(old_state_stack, (1, POST_PROCESS_IMAGE_SIZE[0], 

1 frames
<ipython-input-5-920194395f77> in get_per_error(states, actions, rewards, next_states, terminal, primary_network, target_network)
     10     # the q value for the prim_action_tp1 from the target network
     11     q_from_target = target_network(next_states)
---> 12     updates = rewards + (1 - terminal) * GAMMA * q_from_target.numpy()[:, prim_action_tp1]
     13     target_q[:, actions] = updates
     14     # calculate the loss / error to update priorites

ValueError: operands could not be broadcast together with shapes (31,) (32,32) 

""" This is the jupyter notebook version of the tutorial with some small fixes.
    Instead of running it in command line like "python keras_lstm.py 1" with 
    runopt parameter here = 1, you need to assign run_opt parameter in the code!

    To run this code, you'll need to first download and extract the text dataset
    from here: http://www.fit.vutbr.cz/~imikolov/rnnlm/simple-examples.tgz. Change the
    data_path variable below to your local extraction path and assign run_opt
    """

data_path = "C:\\Users\\Andy\\Documents\\simple-examples\\data\\"

# 'An integer: 1 to train, 2 to test'
run_opt = 1

####

# all from https://adventuresinmachinelearning.com/keras-lstm-tutorial/
# and originally full code from https://github.com/adventuresinML/adventures-in-ml-code/blob/master/keras_lstm.py
# issue of running it as jupyter nb: https://github.com/adventuresinML/adventures-in-ml-code/issues/12
# (solved here)

from __future__ import print_function
import collections
import os
import tensorflow as tf
from keras.models import Sequential, load_model
from keras.layers import Dense, Activation, Embedding, Dropout, TimeDistributed
from keras.layers import LSTM
from keras.optimizers import Adam
from keras.utils import to_categorical
from keras.callbacks import ModelCheckpoint
import numpy as np
import pandas as pd

####

def build_vocab(filename):
    data = pd.read_csv(filename, encoding='utf-8', header = None)
    data = ''.join([str(x) for x in data[0]]).split()

    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_id

def file_to_word_ids(filename, word_to_id):
    data = pd.read_csv(filename, encoding='utf-8', header = None)
    data = ''.join([str(x) for x in data[0]]).split()

    return [word_to_id[word] for word in data if word in word_to_id]


def load_data():
    # get the data paths
    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")

    # build the complete vocabulary, then convert text data to list of integers
    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)
    reversed_dictionary = dict(zip(word_to_id.values(), word_to_id.keys()))

    print(train_data[:5])
    print(list(word_to_id.items())[:10])
    print(vocabulary)
    print(" ".join([reversed_dictionary[x] for x in train_data[:10]]))
    return train_data, valid_data, test_data, vocabulary, reversed_dictionary

####

train_data, valid_data, test_data, vocabulary, reversed_dictionary = load_data()

####

class KerasBatchGenerator(object):

    def __init__(self, data, num_steps, batch_size, vocabulary, skip_step=5):
        self.data = data
        self.num_steps = num_steps
        self.batch_size = batch_size
        self.vocabulary = vocabulary
        # this will track the progress of the batches sequentially through the
        # data set - once the data reaches the end of the data set it will reset
        # back to zero
        self.current_idx = 0
        # skip_step is the number of words which will be skipped before the next
        # batch is skimmed from the data set
        self.skip_step = skip_step

    def generate(self):
        x = np.zeros((self.batch_size, self.num_steps))
        y = np.zeros((self.batch_size, self.num_steps, self.vocabulary))
        while True:
            for i in range(self.batch_size):
                if self.current_idx + self.num_steps >= len(self.data):
                    # reset the index back to the start of the data set
                    self.current_idx = 0
                x[i, :] = self.data[self.current_idx:self.current_idx + self.num_steps]
                temp_y = self.data[self.current_idx + 1:self.current_idx + self.num_steps + 1]
                # convert all of temp_y into a one hot representation
                y[i, :, :] = to_categorical(temp_y, num_classes=self.vocabulary)
                self.current_idx += self.skip_step
            yield x, y

####

num_steps = 30
batch_size = 20
train_data_generator = KerasBatchGenerator(train_data, num_steps, batch_size, vocabulary,
                                           skip_step=num_steps)
valid_data_generator = KerasBatchGenerator(valid_data, num_steps, batch_size, vocabulary,
                                           skip_step=num_steps)

####

hidden_size = 500
use_dropout=True
model = Sequential()
model.add(Embedding(vocabulary, hidden_size, input_length=num_steps))
model.add(LSTM(hidden_size, return_sequences=True))
model.add(LSTM(hidden_size, return_sequences=True))
if use_dropout:
    model.add(Dropout(0.5))
model.add(TimeDistributed(Dense(vocabulary)))
model.add(Activation('softmax'))

####

optimizer = Adam()
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['categorical_accuracy'])

print(model.summary())

####

checkpointer = ModelCheckpoint(filepath=data_path + '/model-{epoch:02d}.hdf5', verbose=1)
num_epochs = 1 #50
if run_opt == 1:
    model.fit_generator(train_data_generator.generate(), len(train_data)//(batch_size*num_steps), num_epochs,
                        validation_data=valid_data_generator.generate(),
                        validation_steps=len(valid_data)//(batch_size*num_steps), callbacks=[checkpointer])
    # model.fit_generator(train_data_generator.generate(), 2000, num_epochs,
    #                     validation_data=valid_data_generator.generate(),
    #                     validation_steps=10)
    model.save(data_path + "final_model.hdf5")
elif run_opt == 2:
    model = load_model(data_path + "\model-40.hdf5")
    dummy_iters = 40
    example_training_generator = KerasBatchGenerator(train_data, num_steps, 1, vocabulary,
                                                     skip_step=1)
    print("Training data:")
    for i in range(dummy_iters):
        dummy = next(example_training_generator.generate())
    num_predict = 10
    true_print_out = "Actual words: "
    pred_print_out = "Predicted words: "
    for i in range(num_predict):
        data = next(example_training_generator.generate())
        prediction = model.predict(data[0])
        predict_word = np.argmax(prediction[:, num_steps-1, :])
        true_print_out += reversed_dictionary[train_data[num_steps + dummy_iters + i]] + " "
        pred_print_out += reversed_dictionary[predict_word] + " "
    print(true_print_out)
    print(pred_print_out)
    # test data set
    dummy_iters = 40
    example_test_generator = KerasBatchGenerator(test_data, num_steps, 1, vocabulary,
                                                     skip_step=1)
    print("Test data:")
    for i in range(dummy_iters):
        dummy = next(example_test_generator.generate())
    num_predict = 10
    true_print_out = "Actual words: "
    pred_print_out = "Predicted words: "
    for i in range(num_predict):
        data = next(example_test_generator.generate())
        prediction = model.predict(data[0])
        predict_word = np.argmax(prediction[:, num_steps - 1, :])
        true_print_out += reversed_dictionary[test_data[num_steps + dummy_iters + i]] + " "
        pred_print_out += reversed_dictionary[predict_word] + " "
    print(true_print_out)
    print(pred_print_out)

# # start in commandline: python keras_lstm.py [-h] [--data_path DATA PATH] runopt
# 'An integer: 1 to train, 2 to test'
# i.e.: python keras_lstm.py 1
# or: python keras_lstm.py 2

from __future__ import print_function
import collections
import os
import tensorflow as tf
from keras.models import Sequential, load_model
from keras.layers import Dense, Activation, Embedding, Dropout, TimeDistributed
from keras.layers import LSTM
from keras.optimizers import Adam
from keras.utils import to_categorical
from keras.callbacks import ModelCheckpoint
import numpy as np
import argparse
import pandas as pd

"""To run this code, you'll need to first download and extract the text dataset
    from here: http://www.fit.vutbr.cz/~imikolov/rnnlm/simple-examples.tgz. Change the
    data_path variable below to your local exraction path"""

# data_path = "C:\\Users\Andy\Documents\simple-examples\data"

parser = argparse.ArgumentParser()
parser.add_argument('run_opt', type=int, default=1, help='An integer: 1 to train, 2 to test')
parser.add_argument('--data_path', type=str, default=data_path, help='The full path of the training data')
args = parser.parse_args()
if args.data_path:
    data_path = args.data_path

def read_words(filename):
    # Changed from tf.gfile to tf.io.gfile
    # https://github.com/tensorflow/tensorflow/issues/31315
    # with tf.io.gfile.GFile(filename, "r") as f:
    # https://github.com/tensorflow/tensorflow/issues/33563
    #     return f.read().decode("utf-8").replace("\n", "<eos>").split()
    data = pd.read_csv(filename, encoding='utf-8', header = None)
    data = ''.join([str(x) for x in data[0]]).split()
    return data

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_id


def 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 load_data():
    # get the data paths
    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")

    # build the complete vocabulary, then convert text data to list of integers
    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)
    reversed_dictionary = dict(zip(word_to_id.values(), word_to_id.keys()))

    print(train_data[:5])
    print(word_to_id)
    print(vocabulary)
    print(" ".join([reversed_dictionary[x] for x in train_data[:10]]))
    return train_data, valid_data, test_data, vocabulary, reversed_dictionary

train_data, valid_data, test_data, vocabulary, reversed_dictionary = load_data()


class KerasBatchGenerator(object):

    def __init__(self, data, num_steps, batch_size, vocabulary, skip_step=5):
        self.data = data
        self.num_steps = num_steps
        self.batch_size = batch_size
        self.vocabulary = vocabulary
        # this will track the progress of the batches sequentially through the
        # data set - once the data reaches the end of the data set it will reset
        # back to zero
        self.current_idx = 0
        # skip_step is the number of words which will be skipped before the next
        # batch is skimmed from the data set
        self.skip_step = skip_step

    def generate(self):
        x = np.zeros((self.batch_size, self.num_steps))
        y = np.zeros((self.batch_size, self.num_steps, self.vocabulary))
        while True:
            for i in range(self.batch_size):
                if self.current_idx + self.num_steps >= len(self.data):
                    # reset the index back to the start of the data set
                    self.current_idx = 0
                x[i, :] = self.data[self.current_idx:self.current_idx + self.num_steps]
                temp_y = self.data[self.current_idx + 1:self.current_idx + self.num_steps + 1]
                # convert all of temp_y into a one hot representation
                y[i, :, :] = to_categorical(temp_y, num_classes=self.vocabulary)
                self.current_idx += self.skip_step
            yield x, y

num_steps = 30
batch_size = 20
train_data_generator = KerasBatchGenerator(train_data, num_steps, batch_size, vocabulary,
                                           skip_step=num_steps)
valid_data_generator = KerasBatchGenerator(valid_data, num_steps, batch_size, vocabulary,
                                           skip_step=num_steps)

hidden_size = 500
use_dropout=True
model = Sequential()
model.add(Embedding(vocabulary, hidden_size, input_length=num_steps))
model.add(LSTM(hidden_size, return_sequences=True))
model.add(LSTM(hidden_size, return_sequences=True))
if use_dropout:
    model.add(Dropout(0.5))
model.add(TimeDistributed(Dense(vocabulary)))
model.add(Activation('softmax'))

optimizer = Adam()
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['categorical_accuracy'])

print(model.summary())
checkpointer = ModelCheckpoint(filepath=data_path + '/model-{epoch:02d}.hdf5', verbose=1)
num_epochs = 50
if args.run_opt == 1:
    model.fit_generator(train_data_generator.generate(), len(train_data)//(batch_size*num_steps), num_epochs,
                        validation_data=valid_data_generator.generate(),
                        validation_steps=len(valid_data)//(batch_size*num_steps), callbacks=[checkpointer])
    # model.fit_generator(train_data_generator.generate(), 2000, num_epochs,
    #                     validation_data=valid_data_generator.generate(),
    #                     validation_steps=10)
    model.save(data_path + "final_model.hdf5")
elif args.run_opt == 2:
    model = load_model(data_path + "\model-40.hdf5")
    dummy_iters = 40
    example_training_generator = KerasBatchGenerator(train_data, num_steps, 1, vocabulary,
                                                     skip_step=1)
    print("Training data:")
    for i in range(dummy_iters):
        dummy = next(example_training_generator.generate())
    num_predict = 10
    true_print_out = "Actual words: "
    pred_print_out = "Predicted words: "
    for i in range(num_predict):
        data = next(example_training_generator.generate())
        prediction = model.predict(data[0])
        predict_word = np.argmax(prediction[:, num_steps-1, :])
        true_print_out += reversed_dictionary[train_data[num_steps + dummy_iters + i]] + " "
        pred_print_out += reversed_dictionary[predict_word] + " "
    print(true_print_out)
    print(pred_print_out)
    # test data set
    dummy_iters = 40
    example_test_generator = KerasBatchGenerator(test_data, num_steps, 1, vocabulary,
                                                     skip_step=1)
    print("Test data:")
    for i in range(dummy_iters):
        dummy = next(example_test_generator.generate())
    num_predict = 10
    true_print_out = "Actual words: "
    pred_print_out = "Predicted words: "
    for i in range(num_predict):
        data = next(example_test_generator.generate())
        prediction = model.predict(data[0])
        predict_word = np.argmax(prediction[:, num_steps - 1, :])
        true_print_out += reversed_dictionary[test_data[num_steps + dummy_iters + i]] + " "
        pred_print_out += reversed_dictionary[predict_word] + " "
    print(true_print_out)
    print(pred_print_out)

loss = network.train_on_batch(states, discounted_rewards)