def train():
'''
$1 path to config file
'''
start = datetime.now()
logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)
# loading configs;
config = get_config(sys.argv[2])
k = config['k']
nthread = config['nthread']
asyn = config['asyn']
mm_path=config['mm_path']
var_path = config['var_path']
minibatch = config['minibatch']
corpus = _mCorpus.get_corpus(mm_path)
V = corpus.num_terms
if asyn:
eta = _masynf.asyn_framework(corpus,k,V,nthread,minibatch,var_path)
else:
eta = _msynf.syn_framework(corpus,k,V,nthread,minibatch,var_path,True)
fn = 'eta.final.pickle'
path = os.path.join(var_path,fn)
_mea.write_eta(eta,path)
end = datetime.now()
print end-start
def train():
'''
$1 path to config file
'''
start = datetime.now()
logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s',
level=logging.INFO)
# loading configs;
config = get_config(sys.argv[2])
k = config['k']
nthread = config['nthread']
asyn = config['asyn']
mm_path = config['mm_path']
var_path = config['var_path']
minibatch = config['minibatch']
corpus = _mCorpus.get_corpus(mm_path)
V = corpus.num_terms
if asyn:
eta = _masynf.asyn_framework(corpus, k, V, nthread, minibatch,
var_path)
else:
eta = _msynf.syn_framework(corpus, k, V, nthread, minibatch, var_path,
True)
fn = 'eta.final.pickle'
path = os.path.join(var_path, fn)
_mea.write_eta(eta, path)
end = datetime.now()
print end - start
def test():
'''
$1 path to config file
'''
logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s',
level=logging.INFO)
# loading configs;
config = get_config(sys.argv[2])
k = config['k']
test_path = config['test_path']
test_train_path = config['test_train']
test_test_path = config['test_test']
eta_path = config['eta_path']
gensim = config['gensim']
print eta_path
corpus = _mCorpus.get_corpus(test_path)
V = corpus.num_terms
voc_set = set()
for doc in corpus:
for wid, count in doc:
voc_set.add(wid)
etaTest, etaSum = None, None
if gensim:
etaTest, etaSum = _mea.get_gensim_eta_etaSum(eta_path, voc_set)
else:
eta = _mea.load_eta(eta_path)
etaTest = _mea.get_eta(k, eta, voc_set)
etaSum = _mea.get_eta_sum(eta, k, V)
test_test = _mCorpus.get_corpus(test_test_path)
test_train = _mCorpus.get_corpus(test_train_path)
alpha = _mea.get_alpha(k)
perplexity = _mper.perplexity(test_train, test_test, alpha, etaTest,
etaSum)
print perplexity
def test():
'''
$1 path to config file
'''
logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)
# loading configs;
config = get_config(sys.argv[2])
k = config['k']
test_path = config['test_path']
test_train_path = config['test_train']
test_test_path = config['test_test']
eta_path = config['eta_path']
gensim= config['gensim']
print eta_path
corpus = _mCorpus.get_corpus(test_path)
V = corpus.num_terms
voc_set = set()
for doc in corpus:
for wid,count in doc:
voc_set.add(wid)
etaTest, etaSum = None,None
if gensim:
etaTest, etaSum = _mea.get_gensim_eta_etaSum(eta_path,voc_set)
else:
eta = _mea.load_eta(eta_path)
etaTest = _mea.get_eta(k,eta,voc_set)
etaSum = _mea.get_eta_sum(eta,k,V)
test_test = _mCorpus.get_corpus(test_test_path)
test_train = _mCorpus.get_corpus(test_train_path)
alpha = _mea.get_alpha(k)
perplexity = _mper.perplexity(test_train,test_test,alpha,etaTest,etaSum)
print perplexity
def main():
# Initializations and preliminaries
comm = MPI.COMM_WORLD # get MPI communicator object
size = comm.size # total number of processes
rank = comm.rank # rank of this process
status = MPI.Status() # get MPI status object
tags = enum('READY', 'DONE', 'EXIT', 'START')
if rank == 0:
# Master process
'''
$1 path to config file
'''
start = datetime.now()
logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s',
level=logging.INFO)
# loading configs;
config = get_config(sys.argv[2])
k = config['k']
nthread = config['nthread']
asyn = config['asyn'] # here, the value should be 'mpi'
mm_path = config['mm_path']
var_path = config['var_path']
minibatch = config['minibatch']
corpus = _mCorpus.get_corpus(mm_path)
V = corpus.num_terms
eta = master_process(comm, status, tags, corpus, k, V, nthread,
minibatch, var_path)
# store the final pickle
fn = 'eta.final.pickle'
path = os.path.join(var_path, fn)
_mea.write_eta(eta, path)
end = datetime.now()
print end - start
else:
# Worker process
name = MPI.Get_processor_name()
worker_process(comm, status, tags, name)
def main():
# Initializations and preliminaries
comm = MPI.COMM_WORLD # get MPI communicator object
size = comm.size # total number of processes
rank = comm.rank # rank of this process
status = MPI.Status() # get MPI status object
tags = enum('READY', 'DONE', 'EXIT', 'START')
if rank == 0:
# Master process
'''
$1 path to config file
'''
start = datetime.now()
logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)
# loading configs;
config = get_config(sys.argv[2])
k = config['k']
nthread = config['nthread']
asyn = config['asyn'] # here, the value should be 'mpi'
mm_path=config['mm_path']
var_path = config['var_path']
minibatch = config['minibatch']
corpus = _mCorpus.get_corpus(mm_path)
V = corpus.num_terms
eta = master_process(comm,status,tags,corpus,k,V,nthread,minibatch,var_path)
# store the final pickle
fn = 'eta.final.pickle'
path = os.path.join(var_path,fn)
_mea.write_eta(eta,path)
end = datetime.now()
print end-start
else:
# Worker process
name = MPI.Get_processor_name()
worker_process(comm,status,tags,name)