semantic_args['n_hidden'] = args['semantic_tensor_n_hidden']
_semantic_model = semantic_class(**semantic_args)
combined_args = {
'w_trainer':_syntactic_model,
'v_trainer':_semantic_model,
'vocab_size':vocab_size,
'indices_in_intersection':list(indices_in_intersection),
'dimensions':args['dimensions'],
'w_loss_multiplier':args['w_loss_multiplier'],
'other_params':args,
'mode':args['mode']
}
if args['simple_joint']:
model = Joint(**combined_args)
else:
combined_args['rho'] = args['rho']
model = ADMM(**combined_args)
def save_model(filename=None):
if filename is None:
filename = 'model-%d.pkl.gz' % model.k
fname = os.path.join(args['base_dir'], filename)
sys.stdout.write('dumping model to %s' % fname)
sys.stdout.flush()
with gzip.open(fname, 'wb') as f:
cPickle.dump(model, f)
sys.stdout.write('\r')
sys.stdout.flush()
_semantic_model = SimilarityNN(**semantic_args)
combined_args = {
'w_trainer':_syntactic_model,
'v_trainer':_semantic_model,
'vocab_size':args['vocab_size'],
'indices_in_intersection':list(indices_in_intersection),
'dimensions':args['dimensions'],
'w_loss_multiplier':args['w_loss_multiplier'],
'other_params':args,
'mode':args['mode']
}
if args['simple_joint']:
model = Joint(**combined_args)
else:
combined_args['rho'] = args['rho']
model = ADMM(**combined_args)
def save_model():
fname = os.path.join(args['base_dir'], 'model-%d.pkl.gz' % model.k)
sys.stdout.write('dumping model to %s' % fname)
sys.stdout.flush()
with gzip.open(fname, 'wb') as f:
cPickle.dump(model, f)
sys.stdout.write('\r')
sys.stdout.flush()
# save the initial state
if not model_loaded:
_semantic_model = SimilarityNN(**semantic_args)
combined_args = {
'w_trainer': _syntactic_model,
'v_trainer': _semantic_model,
'vocab_size': args['vocab_size'],
'indices_in_intersection': list(indices_in_intersection),
'dimensions': args['dimensions'],
'w_loss_multiplier': args['w_loss_multiplier'],
'other_params': args,
'mode': args['mode']
}
if args['simple_joint']:
model = Joint(**combined_args)
else:
combined_args['rho'] = args['rho']
model = ADMM(**combined_args)
def save_model():
fname = os.path.join(args['base_dir'], 'model-%d.pkl.gz' % model.k)
sys.stdout.write('dumping model to %s' % fname)
sys.stdout.flush()
with gzip.open(fname, 'wb') as f:
cPickle.dump(model, f)
sys.stdout.write('\r')
sys.stdout.flush()
# save the initial state
if not model_loaded:
semantic_args['n_hidden'] = args['semantic_tensor_n_hidden']
_semantic_model = semantic_class(**semantic_args)
combined_args = {
'w_trainer': _syntactic_model,
'v_trainer': _semantic_model,
'vocab_size': vocab_size,
'indices_in_intersection': list(indices_in_intersection),
'dimensions': args['dimensions'],
'w_loss_multiplier': args['w_loss_multiplier'],
'other_params': args,
'mode': args['mode']
}
if args['simple_joint']:
model = Joint(**combined_args)
else:
combined_args['rho'] = args['rho']
model = ADMM(**combined_args)
def save_model(filename=None):
if filename is None:
filename = 'model-%d.pkl.gz' % model.k
fname = os.path.join(args['base_dir'], filename)
sys.stdout.write('dumping model to %s' % fname)
sys.stdout.flush()
with gzip.open(fname, 'wb') as f:
cPickle.dump(model, f)
sys.stdout.write('\r')
sys.stdout.flush()
