def __init__(self, config, **kwargs):
super(Model, self).__init__(**kwargs)
self.config = config
self.pre_context_embedder = ContextEmbedder(
config.pre_embedder, name='pre_context_embedder')
self.post_context_embedder = ContextEmbedder(
config.post_embedder, name='post_context_embedder')
in1 = 2 + sum(x[2] for x in config.pre_embedder.dim_embeddings)
self.input_to_rec = MLP(activations=[Tanh()],
dims=[in1, config.hidden_state_dim],
name='input_to_rec')
self.rec = LSTM(dim=config.hidden_state_dim, name='recurrent')
in2 = config.hidden_state_dim + sum(
x[2] for x in config.post_embedder.dim_embeddings)
self.rec_to_output = MLP(activations=[Tanh()],
dims=[in2, 2],
name='rec_to_output')
self.sequences = ['latitude', 'latitude_mask', 'longitude']
self.context = self.pre_context_embedder.inputs + self.post_context_embedder.inputs
self.inputs = self.sequences + self.context
self.children = [
self.pre_context_embedder, self.post_context_embedder,
self.input_to_rec, self.rec, self.rec_to_output
]
self.initial_state_ = shared_floatx_zeros((config.hidden_state_dim, ),
name="initial_state")
self.initial_cells = shared_floatx_zeros((config.hidden_state_dim, ),
name="initial_cells")
def __init__(self, config, **kwargs):
super(Model, self).__init__(**kwargs)
self.config = config
self.context_embedder = ContextEmbedder(config)
self.mlp = MLP(activations=[Rectifier()
for _ in config.dim_hidden] + [Identity()],
dims=[config.dim_input] + config.dim_hidden +
[config.dim_output])
self.inputs = self.context_embedder.inputs # + self.extremities.keys()
self.children = [self.context_embedder, self.mlp]
def __init__(self, config, output_layer=None, **kwargs):
super(FFMLP, self).__init__(**kwargs)
self.config = config
self.context_embedder = ContextEmbedder(config)
output_activation = [] if output_layer is None else [output_layer()]
output_dim = [] if output_layer is None else [config.dim_output]
self.mlp = MLP(activations=[Rectifier() for _ in config.dim_hidden] + output_activation,
dims=[config.dim_input] + config.dim_hidden + output_dim)
self.extremities = {'%s_k_%s' % (side, ['latitude', 'longitude'][axis]): axis for side in ['first', 'last'] for axis in [0, 1]}
self.inputs = self.context_embedder.inputs + self.extremities.keys()
self.children = [ self.context_embedder, self.mlp ]
def __init__(self, config, output_dim, activation, **kwargs):
super(MLPEncoder, self).__init__(**kwargs)
self.config = config
self.context_embedder = ContextEmbedder(self.config)
self.encoder_mlp = MLP(activations=[Rectifier() for _ in config.dim_hidden]
+ [activation()],
dims=[config.dim_input]
+ config.dim_hidden
+ [output_dim],
name='encoder')
self.extremities = {'%s_k_%s' % (side, ['latitude', 'longitude'][axis]): axis
for side in ['first', 'last'] for axis in [0, 1]}
self.children = [ self.context_embedder,
self.encoder_mlp ]
def __init__(self, config, output_dim=2, **kwargs):
super(BidiRNN, self).__init__(**kwargs)
self.config = config
self.context_embedder = ContextEmbedder(config)
act = config.rec_activation() if hasattr(config,
'rec_activation') else None
self.rec = SegregatedBidirectional(
LSTM(dim=config.hidden_state_dim, activation=act,
name='recurrent'))
self.fwd_fork = Fork([
name
for name in self.rec.prototype.apply.sequences if name != 'mask'
],
prototype=Linear(),
name='fwd_fork')
self.bkwd_fork = Fork([
name
for name in self.rec.prototype.apply.sequences if name != 'mask'
],
prototype=Linear(),
name='bkwd_fork')
rto_in = config.hidden_state_dim * 2 + sum(
x[2] for x in config.dim_embeddings)
self.rec_to_output = MLP(
activations=[Rectifier()
for _ in config.dim_hidden] + [Identity()],
dims=[rto_in] + config.dim_hidden + [output_dim])
self.sequences = ['latitude', 'latitude_mask', 'longitude']
self.inputs = self.sequences + self.context_embedder.inputs
self.children = [
self.context_embedder, self.fwd_fork, self.bkwd_fork, self.rec,
self.rec_to_output
]
def __init__(self, config, **kwargs):
super(Model, self).__init__(**kwargs)
self.config = config
self.context_embedder = ContextEmbedder(config)
self.prefix_encoder = MLP(activations=[
Rectifier() for _ in config.prefix_encoder.dim_hidden
] + [config.representation_activation()],
dims=[config.prefix_encoder.dim_input] +
config.prefix_encoder.dim_hidden +
[config.representation_size],
name='prefix_encoder')
self.candidate_encoder = MLP(
activations=[
Rectifier() for _ in config.candidate_encoder.dim_hidden
] + [config.representation_activation()],
dims=[config.candidate_encoder.dim_input] +
config.candidate_encoder.dim_hidden + [config.representation_size],
name='candidate_encoder')
self.softmax = Softmax()
self.prefix_extremities = {
'%s_k_%s' % (side, ['latitude', 'longitude'][axis]): axis
for side in ['first', 'last'] for axis in [0, 1]
}
self.candidate_extremities = {
'candidate_%s_k_%s' % (side, ['latitude', 'longitude'][axis]): axis
for side in ['first', 'last'] for axis in [0, 1]
}
self.inputs = self.context_embedder.inputs + [
'candidate_%s' % k for k in self.context_embedder.inputs
] + self.prefix_extremities.keys() + self.candidate_extremities.keys()
self.children = [
self.context_embedder, self.prefix_encoder, self.candidate_encoder,
self.softmax
]
