def forward(self, input_embeds_list):
"""
Args:
input_embeds_list (list[SequenceBatchElement]): where each element is of shape (batch_size, input_dim)
Returns:
hidden_states_list (list[SequenceBatchElement]) where each element is (batch_size, hidden_dim)
"""
batch_size = input_embeds_list[0].values.size()[0]
h = tile_state(self.h0, batch_size) # (batch_size, hidden_dim)
c = tile_state(self.c0, batch_size) # (batch_size, hidden_dim)
hidden_states_list = []
for t, x in enumerate(input_embeds_list):
# x.values has shape (batch_size, input_dim)
# x.mask has shape (batch_size, 1)
h_new, c_new = self.rnn_cell(x.values, (h, c))
h = gated_update(h, h_new, x.mask)
c = gated_update(c, c_new, x.mask)
hidden_states_list.append(
SequenceBatchElement(self.dropout(h), x.mask))
return hidden_states_list
def initialize(self, batch_size):
h = tile_state(self.h0, batch_size)
c = tile_state(self.c0, batch_size)
# no initial weights, context is just zero vector
init_attn = lambda attention: AttentionOutput(
None, GPUVariable(torch.zeros(batch_size, attention.memory_dim)))
return AttentionRNNState([h] * self.num_layers, [c] * self.num_layers,
init_attn(self.source_attention),
init_attn(self.insert_attention),
init_attn(self.delete_attention))
def initialize(self, batch_size):
h = tile_state(self.h0, batch_size)
c = tile_state(self.c0, batch_size)
# no initial weights, context is just zero vector
init_attn = lambda attention: AttentionOutput(
weights=None,
context=GPUVariable(torch.zeros(batch_size, attention.memory_dim)),
logits=None)
return AttentionRNNState(
[h] * self.num_layers, [c] * self.num_layers,
[init_attn(attn) for attn in self.input_attentions])
def initialize(self, batch_size):
h = tile_state(self.h0, batch_size)
c = tile_state(self.c0, batch_size)
return MultilayeredRNNState([h] * self.num_layers,
[c] * self.num_layers)
def initialize(self, batch_size):
h = tile_state(self.h0, batch_size)
c = tile_state(self.c0, batch_size)
return SimpleRNNState(h, c)
def _encoder_output(self, batch_size):
return tile_state(self.agenda, batch_size)
def test_tile_state():
h = GPUVariable(torch.FloatTensor([1, 2, 3]))
h_tiled = tile_state(h, 3)
assert_tensor_equal(h_tiled, [[1, 2, 3], [1, 2, 3], [1, 2, 3]])