i+=1
#print("Index {} overeenkomstig label {}".format(index,targets[index]));sys.stdout.flush()
else:
indices.append(index)
if shuffle:
np.random.shuffle(indices)
for start_idx in range(0, len(indices) - batch_size + 1, batch_size):
excerpt = indices[start_idx:start_idx + batch_size]
yield inputs[excerpt], targets[excerpt]
import load_class
load = load_class.load(data_ratio)
import convnet
convnet = convnet.convnet(20)
try:
i = 0
path = "/home/jasper/oneshot-gestures/"
while os.path.exists("{}output/acc-cost_{}.csv".format(path,i)):
i += 1
fo1 = open("{}output/acc-cost_{}.csv".format(path,i), "w")
fo1.write("training_loss;validation_loss;validation_accuracy;epoch_time\n")
except IOError as e:
print("I/O error({0}): {1}".format(e.errno, e.strerror))
raise
except:
print("unexpected error")
raise
save_param_path = "{}model_parameters/param_model_19_2".format(path)
def main(unused_argv):
if len(unused_argv
) != 1: # prints a message if you've entered flags incorrectly
raise Exception("Problem with flags: %s" % unused_argv)
# Loading the external information first
extra_info = {}
if os.path.exists(FLAGS.external_config):
external_params = xml_parser.parse(FLAGS.external_config, flat=False)
if 'sent2vec_params' in external_params:
sent2vec_params = external_params['sent2vec_params']
convnet_params = sent2vec_params['convnet_params']
convnet_model2load = sent2vec_params['model2load']
gamma = 0.2 if not 'gamma' in sent2vec_params else sent2vec_params[
'gamma']
my_convnet = convnet.convnet(convnet_params)
my_convnet.train_validate_test_init()
my_convnet.load_params(file2load=convnet_model2load)
fixed_vars = tf.global_variables()
fixed_vars.remove(my_convnet.embedding_matrix)
extra_info['sent2vec'] = {'gamma': gamma, 'network': my_convnet}
extra_info['fixed_vars'] = fixed_vars
if 'key_phrases' in external_params:
# TODO: phrase some parameters to import the results of key-phrase extracted or \
# parameters for online key-phrase extraction
extra_info['key_phrases'] = {}
raise NotImplementedError(
'Key phrases part has not been implemented yet')
tf.logging.set_verbosity(
tf.logging.INFO) # choose what level of logging you want
tf.logging.info('Starting seq2seq_attention in %s mode...', (FLAGS.mode))
# Change log_root to FLAGS.log_root/FLAGS.exp_name and create the dir if necessary
FLAGS.log_root = os.path.join(FLAGS.log_root, FLAGS.exp_name)
if not os.path.exists(FLAGS.log_root):
if FLAGS.mode == "train":
os.makedirs(FLAGS.log_root)
else:
raise Exception(
"Logdir %s doesn't exist. Run in train mode to create it." %
(FLAGS.log_root))
vocab = Vocab(FLAGS.vocab_path, FLAGS.vocab_size) # create a vocabulary
# If in decode mode, set batch_size = beam_size
# Reason: in decode mode, we decode one example at a time.
# On each step, we have beam_size-many hypotheses in the beam, so we need to make a batch of these hypotheses.
if FLAGS.mode == 'decode':
FLAGS.batch_size = FLAGS.beam_size
# Make a namedtuple hps, containing the values of the hyperparameters that the model needs
hparam_list = [
'mode', 'lr', 'adagrad_init_acc', 'rand_unif_init_mag',
'trunc_norm_init_std', 'max_grad_norm', 'hidden_dim', 'emb_dim',
'batch_size', 'max_dec_steps', 'max_enc_steps', 'coverage',
'cov_loss_wt', 'pointer_gen'
]
hps_dict = {}
for key, val in FLAGS.__flags.iteritems(): # for each flag
if key in hparam_list: # if it's in the list
hps_dict[key] = val # add it to the dict
hps = namedtuple("HParams", hps_dict.keys())(**hps_dict)
# Create a batcher object that will create minibatches of data
batcher = Batcher(FLAGS.data_path,
vocab,
hps,
single_pass=FLAGS.single_pass)
tf.set_random_seed(111) # a seed value for randomness
if hps.mode == 'train':
print "creating model..."
model = SummarizationModel(hps, vocab, extra_info)
setup_training(model, batcher)
elif hps.mode == 'eval':
model = SummarizationModel(hps, vocab, extra_info)
run_eval(model, batcher, vocab)
elif hps.mode == 'decode':
decode_model_hps = hps # This will be the hyperparameters for the decoder model
decode_model_hps = hps._replace(
max_dec_steps=1
) # The model is configured with max_dec_steps=1 because we only ever run one step of the decoder at a time (to do beam search). Note that the batcher is initialized with max_dec_steps equal to e.g. 100 because the batches need to contain the full summaries
model = SummarizationModel(decode_model_hps, vocab, extra_info)
decoder = BeamSearchDecoder(model, batcher, vocab)
decoder.decode(
) # decode indefinitely (unless single_pass=True, in which case deocde the dataset exactly once)
else:
raise ValueError("The 'mode' flag must be one of train/eval/decode")
embedding_loader_params = embedding_params['embedding_loader_params']
source = embedding_params['source']
format = embedding_params['format']
my_embedding_loader = embedding_loader.embedding_loader(
embedding_loader_params)
assert convnet_model_params['embedding_dim']==my_embedding_loader.embedding_dim, \
'dimensions of word embeddings do not match, convnet=%d, embedding_loader=%d'%(
convnet_model_params['embedding_dim'], my_embedding_loader.embedding_dim)
my_embedding_loader.load_embedding(source=source, format=format)
embedding_matrix = my_embedding_loader.gen_embedding_matrix(
generator=my_data_generator)
convnet_model_params['embedding_matrix'] = embedding_matrix
my_network = convnet.convnet(convnet_model_params)
if not os.path.exists(model_saved_folder):
os.makedirs(model_saved_folder)
my_network.train_validate_test_init()
if 'model2load' in network_params:
try:
my_network.load_params(network_params['model2load'])
except:
print('ERROR: Failed to load checkpoint: %s' %
network_params['model2load'])
train_loss_dict = {}
validate_loss_dict = {}
train_loss = []
best_validation_loss = 1e8 # best validation loss
for start_idx in range(0, len(indices) - batch_size + 1, batch_size):
excerpt = indices[start_idx:start_idx + batch_size]
yield inputs[excerpt], targets[excerpt]
base_dir_path = "/home/jasper/oneshot-gestures/"
test_accuracies = []
load = load_class.load(size_ratio=1.0)
# Load data
x_validate, labels_validate, indices_validate = load.load_validation_set()
x_train, labels_train, indices_train = load.load_training_set()
x_test, labels_test, indices_test = load.load_testing_set()
convnet = convnet.convnet(num_output_units=20)
convnet.save_param_values("{}/default_param".format(base_dir_path))
for oneshot_class in xrange(20
):
print("Learning gestures excluding class {}".format(oneshot_class))
save_param_patch = "{}convnet_params/param-excl-class-{}".format(base_dir_path,oneshot_class)
convnet.load_param_values(save_param_patch)
try:
fo1 = open("{}output_19cl/excl-class-{}.csv".format(base_dir_path,oneshot_class),"w")
fo1.write("training_loss;validation_loss;validation_accuracy;epoch_time\n")
except IOError as e:
print("I/O error({0}): {1}".format(e.errno, e.strerror))
except:
from keras.layers import Input, Lambda
from keras.models import Model
from encode_data import encode_data
from show_neighbors import show_neighbors
from sklearn.neighbors import NearestNeighbors
from sklearn.manifold import TSNE
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train = x_train / 255
x_test = x_test / 255
in_dim = (28, 28, 1)
out_dim = 128
encoder = convnet(in_dim, out_dim)
in_a = Input(shape=in_dim)
in_p = Input(shape=in_dim)
in_n = Input(shape=in_dim)
emb_a = encoder(in_a)
emb_p = encoder(in_p)
emb_n = encoder(in_n)
positive_dist = Lambda(euclidean_distance, name='pos_dist')([emb_a, emb_p])
negative_dist = Lambda(euclidean_distance, name='neg_dist')([emb_a, emb_n])
tertiary_dist = Lambda(euclidean_distance, name='ter_dist')([emb_p, emb_n])
stacked_dists = Lambda(lambda vects: K.stack(vects, axis=1),
name='stacked_dists')(
if output_file.split('.')[-1] in allowed_format:
pure_name='.'.join(output_file.split('.')[:-1])
mark=pure_name.split('_')[0]
suffix=pure_name.split('_')[-1]
if suffix==output_suffix:
output_name2file[mark]=output_folder+os.sep+output_file
print('In the decode folder, there are %d detected files'%len(output_name2file.keys()))
name2files={}
for key in refer_name2file:
if key in output_name2file:
name2files[key]=(refer_name2file[key],output_name2file[key])
print('There are %d reference-decode paire detected'%len(name2files.keys()))
# Load model
my_convnet=convnet.convnet(convnet_params)
my_convnet.train_validate_test_init()
my_convnet.load_params(file2dump=model2load)
word2idx=cPickle.load(open(word2idx_file, 'r'))
tosave={'summary':{'cosine_average':0.0, 'dist_average':0.0, 'valid_documents':0},'details':[]}
r=rouge.Rouge()
for idx,key in enumerate(name2files):
sys.stdout.write('loading documents %d/%d=%.1f%%\r'%((idx+1),len(name2files.keys()),
float(idx+1)/float(len(name2files.keys()))*100))
sys.stdout.flush()
refer_file,output_file=name2files[key]
try:
refer_sequence=open(refer_file, 'r').readlines()
def main(unused_argv):
if len(unused_argv) != 1:
raise Exception('Problem with flags: %s' % str(unused_argv))
try:
assert (FLAGS.mode == 'decode')
except:
raise ValueError('mode much be "decode" but now it is %s' %
str(FLAGS.mode))
FLAGS.log_root = os.path.join(FLAGS.log_root, FLAGS.exp_name)
try:
assert (os.path.exists(FLAGS.log_root))
except:
raise ValueError('Invalid log_root: %s' % str(FLAGS.log_root))
FLAGS.batch_size = FLAGS.beam_size
data_manager = BinaryDataManager(binary_file=FLAGS.data_path,
single_pass=True)
data_manager.load_data()
# Loading the external information first
extra_info = {}
if os.path.exists(FLAGS.external_config):
external_params = xml_parser.parse(FLAGS.external_config, flat=False)
if 'sent2vec_params' in external_params:
sent2vec_params = external_params['sent2vec_params']
convnet_params = sent2vec_params['convnet_params']
convnet_model2load = sent2vec_params['model2load']
gamma = 0.2 if not 'gamma' in sent2vec_params else sent2vec_params[
'gamma']
my_convnet = convnet.convnet(convnet_params)
my_convnet.train_validate_test_init()
my_convnet.load_params(file2load=convnet_model2load)
fixed_vars = tf.global_variables()
fixed_vars.remove(my_convnet.embedding_matrix)
extra_info['sent2vec'] = {'gamma': gamma, 'network': my_convnet}
extra_info['fixed_vars'] = fixed_vars
if 'key_phrases' in external_params:
# TODO: phrase some parameters to import the results of key-phrase extracted or \
# parameters for online key-phrase extraction
extra_info['key_phrases'] = {}
raise NotImplementedError(
'Key phrases part has not been implemented yet')
model_hp_list = [
'mode', 'lr', 'adagrad_init_acc', 'rand_unif_init_mag',
'trunc_norm_init_std', 'max_grad_norm', 'hidden_dim', 'emb_dim',
'batch_size', 'max_dec_steps', 'max_enc_steps', 'coverage',
'cov_loss_wt', 'pointer_gen'
]
model_hp_dict = {}
for key, value in FLAGS.__flags.iteritems():
if key in model_hp_list:
model_hp_dict[key] = value
model_settings = namedtuple('HParams',
model_hp_dict.keys())(**model_hp_dict)
model_settings = model_settings._replace(max_dec_steps=1)
for folder in [
FLAGS.article_folder, FLAGS.refer_folder, FLAGS.output_folder
]:
if not os.path.exists(folder):
os.makedirs(folder)
solver = RunTimeWrapper(hp=FLAGS,
model_settings=model_settings,
extra_info=extra_info)
solver.start()
result2write = ''
for idx, (article, abstract) in enumerate(data_manager.text_abstract_pair):
sys.stdout.write(
'Analysizing the documents %d/%d = %.1f%% \r' %
(idx + 1, len(data_manager.text_abstract_pair), float(idx + 1) /
float(len(data_manager.text_abstract_pair)) * 100))
sys.stdout.flush()
_, summary = solver.run(query=article)
abstract = '\n'.join(abstract)
# Reference and compare
with open(FLAGS.article_folder + os.sep + '%04d_article.txt' % idx,
'w') as fopen:
fopen.write(article)
with open(FLAGS.refer_folder + os.sep + '%04d_reference.txt' % idx,
'w') as fopen:
fopen.write(abstract)
with open(FLAGS.output_folder + os.sep + '%04d_decode.txt' % idx,
'w') as fopen:
fopen.write(summary)
result2write += '\n\n===\n%s\n\n>>>refer:\n%s\n\n>>>output:\n%s\n' % (
article, abstract, summary)
if (idx + 1) % 100 == 0:
with open('results.txt', 'w') as fopen:
fopen.write(result2write)
solver.end()
data_dir = arguments['data_dir']
output_dir = arguments['output_dir']
with tf.Graph().as_default():
tf.set_random_seed(1234)
image_data, label_data = read.preprocess(file = os.path.join(data_dir, 'train.json'))
batch = read.batch(images = image_data, labels = label_data)
#training
X = tf.placeholder(dtype = tf.float32, shape=(100, 75, 75, 2))
Y = tf.placeholder(dtype = tf.int32, shape=(100, 1))
output = convnet.convnet(X)
loss = convnet.loss(output, Y)
loss_summary = tf.summary.scalar(name = 'loss_summary', tensor = loss)
optimizer = tf.train.GradientDescentOptimizer(learning_rate = 0.01)
grads = optimizer.compute_gradients(loss)
train_op = optimizer.apply_gradients(grads)
saver = tf.train.Saver(save_relative_paths= True)
#eval_ops
eval_images = tf.placeholder(dtype = tf.float32, name= 'eval_images')
eval_op = convnet.convnet(eval_images)
