def init_params(options, preemb=None):
"""
Initialize all parameters
"""
params = OrderedDict()
# Word embedding
if preemb == None:
params['Wemb'] = norm_weight(options['n_words'], options['dim_word'])
else:
print('using preemb')
params['Wemb'] = preemb
# Encoder
params = get_layer(options['encoder'])[0](options, params, prefix='encoder',
nin=options['dim_word'], dim=options['dim'])
# Decoder: next sentence
params = get_layer(options['decoder'])[0](options, params, prefix='decoder_f',
nin=options['dim_word'], dim=options['dim'])
# Decoder: previous sentence
params = get_layer(options['decoder'])[0](options, params, prefix='decoder_b',
nin=options['dim_word'], dim=options['dim'])
# Output layer
params = get_layer('ff')[0](options, params, prefix='ff_logit', nin=options['dim'], nout=options['n_words'])
return params
def param_init_fflayer(options, params, prefix='ff', nin=None, nout=None, ortho=True):
"""
Affine transformation + point-wise nonlinearity
"""
if nin == None:
nin = options['dim_proj']
if nout == None:
nout = options['dim_proj']
params[_p(prefix,'W')] = norm_weight(nin, nout, ortho=ortho)
params[_p(prefix,'b')] = numpy.zeros((nout,)).astype('float32')
return params
def param_init_gru(options, params, prefix='gru', nin=None, dim=None):
"""
Gated Recurrent Unit (GRU)
"""
if nin == None:
nin = options['dim_proj']
if dim == None:
dim = options['dim_proj']
W = numpy.concatenate([norm_weight(nin,dim),
norm_weight(nin,dim)], axis=1)
params[_p(prefix,'W')] = W
params[_p(prefix,'b')] = numpy.zeros((2 * dim,)).astype('float32')
U = numpy.concatenate([ortho_weight(dim),
ortho_weight(dim)], axis=1)
params[_p(prefix,'U')] = U
Wx = norm_weight(nin, dim)
params[_p(prefix,'Wx')] = Wx
Ux = ortho_weight(dim)
params[_p(prefix,'Ux')] = Ux
params[_p(prefix,'bx')] = numpy.zeros((dim,)).astype('float32')
return params
def init_params(options, preemb=None):
"""
Initialize all parameters
"""
params = OrderedDict()
# Word embedding
if preemb == None:
params['Wemb'] = norm_weight(options['n_words'], options['dim_word'])
else:
print('using preemb')
params['Wemb'] = preemb
# Encoder
params = get_layer(options['encoder'])[0](options,
params,
prefix='encoder',
nin=options['dim_word'],
dim=options['dim'])
# Decoder: next sentence
params = get_layer(options['decoder'])[0](options,
params,
prefix='decoder_f',
nin=options['dim_word'],
dim=options['dim'])
# Decoder: previous sentence
params = get_layer(options['decoder'])[0](options,
params,
prefix='decoder_b',
nin=options['dim_word'],
dim=options['dim'])
# Output layer
params = get_layer('ff')[0](options,
params,
prefix='ff_logit',
nin=options['dim'],
nout=options['n_words'])
return params