def build_word2vec_model(model_name):
''' retrieve the model from persistent files.
Args:
model_name (string): the model name, assuming models
are located in the standard folder.
Return:
model (word2vec object): the retrieved model
'''
NUM_THREADS = 2 * multiprocessing.cpu_count() - 1
COMMON_PATH = os.path.join(os.path.expanduser("~"),
'local_tensorflow_content')
model_config = {}
model_config['model_name'] = model_name
model_config['restore_model'] = True
model_config['eval_mode'] = True
use_gpu = False
if use_gpu:
model_config['sess_config'] = tf.ConfigProto(
log_device_placement=False,
gpu_options=tf.GPUOptions(per_process_gpu_memory_fraction=0.5))
else:
os.environ[
'CUDA_VISIBLE_DEVICES'] = '-1' # the only way to completely not use GPU
model_config['sess_config'] = tf.ConfigProto(
intra_op_parallelism_threads=NUM_THREADS)
model_config['model_path'] = create_local_model_path(
COMMON_PATH, model_config['model_name'])
model_config['log_path'] = create_local_log_path(
COMMON_PATH, model_config['model_name'])
model = word2vec(**model_config)
return model
def predict_with_word2vec():
model_config = {}
model_config['model_name'] = 'word2vec'
model_config['restore_model'] = True
model_config['eval_mode'] = True
use_gpu = False
if use_gpu:
model_config['sess_config'] = tf.ConfigProto(
log_device_placement=False,
gpu_options=tf.GPUOptions(per_process_gpu_memory_fraction=0.5))
else:
os.environ[
'CUDA_VISIBLE_DEVICES'] = '-1' # the only way to completely not use GPU
model_config['sess_config'] = tf.ConfigProto(
intra_op_parallelism_threads=NUM_THREADS)
model_config['model_path'] = create_local_model_path(
COMMON_PATH, model_config['model_name'])
model_config['log_path'] = create_local_log_path(
COMMON_PATH, model_config['model_name'])
model = word2vec(**model_config)
max_X, min_X, mean_X = model.predict([2, 4, 5])
print max_X
print min_X
def main():
NUM_THREADS = multiprocessing.cpu_count()
COMMON_PATH = os.path.join(os.path.expanduser("~"),
'local_tensorflow_content')
pickle_file = 'titles_CBOW_data.pkl'
pickle_file_path = os.path.join(os.path.expanduser("~"), pickle_file)
dataGen = DataGenerator(pickle_file_path)
model_config, training_config = {}, {}
model_config['vocab_size'] = dataGen.vocab_size
model_config['batch_size'] = 32
model_config['context_window'] = 2
model_config['embedding_size'] = 128
model_config['neg_sample_size'] = 2
model_config['learning_rate'] = 0.0005
model_config['model_name'] = 'word2vec'
batches = dataGen.generate_sequence(model_config['batch_size'])
model = word2vec(**model_config)
use_gpu = False
if use_gpu:
training_config['sess_config'] = tf.ConfigProto(
log_device_placement=False,
gpu_options=tf.GPUOptions(per_process_gpu_memory_fraction=0.5))
else:
os.environ[
'CUDA_VISIBLE_DEVICES'] = '-1' # the only way to completely not use GPU
training_config['sess_config'] = tf.ConfigProto(
intra_op_parallelism_threads=NUM_THREADS)
training_config['model_path'] = create_local_model_path(
COMMON_PATH, model_config['model_name'])
training_config['log_path'] = create_local_log_path(
COMMON_PATH, model_config['model_name'])
generate_tensorboard_script(
training_config['log_path']
) # create the script to start a tensorboard session
training_config['epoch_num'] = 20000
training_config['display_steps'] = 1000
training_config['saving_steps'] = 1 * training_config['display_steps']
training_config['num_batches'] = int(dataGen.data_size *
training_config['epoch_num'] /
model_config['batch_size'])
print 'total #batches: {}, vocab_size: {}'.format(
training_config['num_batches'], model_config['vocab_size'])
model.train(batches, training_config, restore_model=False)
def model_train():
pickle_file = 'lemmatized_only_skip_gram_window_2_skips_2.pkl'
pickle_file_path = os.path.join(os.path.expanduser("~"), pickle_file)
dataGen = DataGenerator(pickle_file_path)
model_config, training_config = {}, {}
model_config['vocab_size'] = dataGen.vocab_size
model_config['model_name'] = 'word2vec_skip_gram_lemmatized_only_window_2_skips_2'
model_config['model_type'] = "SKIP_GRAM"
model_config['batch_size'] = 32
model_config['context_window'] = 2
model_config['embedding_size'] = 128
model_config['neg_sample_size'] = 10
model_config['learning_rate'] = 0.001
model_config['saving_steps'] = 20000
#model_config['model_name'] = 'word2vec_model'
model_config['restore_model'] = False
model_config['eval_mode'] = False
batches = dataGen.generate_sequence(model_config['batch_size'])
use_gpu = False
if use_gpu:
model_config['sess_config'] = tf.ConfigProto(log_device_placement=False,
gpu_options=tf.GPUOptions(per_process_gpu_memory_fraction=0.5))
else:
os.environ['CUDA_VISIBLE_DEVICES'] = '-1' # the only way to completely not use GPU
model_config['sess_config'] = tf.ConfigProto(intra_op_parallelism_threads=NUM_THREADS)
model_config['model_path'] = create_local_model_path(COMMON_PATH, model_config['model_name'])
model_config['log_path'] = create_local_log_path(COMMON_PATH, model_config['model_name'])
generate_tensorboard_script(model_config['log_path']) # create the script to start a tensorboard session
model = word2vec(**model_config)
epoch_num = 20000
training_config['batches'] = batches
training_config['display_steps'] = 10000
training_config['saving_steps'] = model_config['saving_steps']
training_config['num_batches'] = int(dataGen.data_size * epoch_num / model_config['batch_size'])
print 'total #batches: {}, vocab_size: {}'.format(training_config['num_batches'], model_config['vocab_size'])
model.train(**training_config)
def main():
model_config = {}
model_config['batch_size'] = 32
model_config['context_window'] = 2
model_config['embedding_size'] = 128
model_config['neg_sample_size'] = 10
model_config['learning_rate'] = 0.001
model_config['model_name'] = 'word2vec_test'
model_config['restore_model'] = False
model_config['eval_mode'] = False
model_config['model_type'] = 'SKIP_GRAM'
model_config['model_path'] = create_local_model_path(
COMMON_PATH, model_config['model_name'])
model_config['log_path'] = create_local_log_path(
COMMON_PATH, model_config['model_name'])
use_gpu = False
if use_gpu:
model_config['sess_config'] = tf.ConfigProto(
log_device_placement=False,
gpu_options=tf.GPUOptions(per_process_gpu_memory_fraction=0.5))
else:
os.environ[
'CUDA_VISIBLE_DEVICES'] = '-1' # the only way to completely not use GPU
model_config['sess_config'] = tf.ConfigProto(
intra_op_parallelism_threads=NUM_THREADS)
model = word2vec(**model_config)
with model.graph.as_default():
input_X_, input_y = skip_gram_batch_content()
print "input_X_: ", input_X_
print "input_y: ", input_y
_, loss = model.sess.run([model.train_op, model.loss],
feed_dict={
model.X: input_X_,
model.y: input_y
})
print "the model loss: {}".format(loss)