def apply(self, x): # (batsize, outerseqlen, innerseqlen)
y = x
xm = T.neq(x, self.maskid) if self.maskid is not None else None
for enc in self.encoders:
y = EncLastDim(enc)(y, mask=xm)
xm = T.sum(xm, axis=-1) > 0 if self.maskid is not None else None
return y
def apply(self, x):
if self.l2emb is not None:
l1tensor = x[:, :, 1:]
l1encs = EncLastDim(self.l1enc)(l1tensor)
l2mat = x[:, :, 0]
assert (l2mat.ndim == 2)
l2embs = self.l2emb(l2mat)
l2vecs = T.concatenate([l1encs, l2embs], axis=2)
wmask = T.neq(l2mat,
self.maskid) if self.maskid is not None else None
else:
l2vecs = EncLastDim(self.l1enc)(x)
wmask = T.gt(T.sum(T.eq(x, self.maskid), axis=2), 0)
l2vecs.mask = wmask
fenc = self.l2enc(l2vecs)
return fenc #, wmask #mask for debug
def apply(self, x):
# word vectors and mask
charten = x[:, :, 1:]
charencs = EncLastDim(self.charenc)(charten)
wordmat = x[:, :, 0]
wordembs = self.wordemb(wordmat)
wordvecs = T.concatenate([charencs, wordembs], axis=2)
wordmask = T.neq(wordmat, self.maskid)
wordvecs.mask = wordmask
# do outerpolation
weights, mask = self.outerpol(wordvecs)
leftenco = self.leftenc(wordvecs,
weights=weights).dimshuffle(0, 'x', 1)
rightenco = self.rightenc(wordvecs,
weights=(1 - weights)).dimshuffle(0, 'x', 1)
ret = T.concatenate([self.leftlin(leftenco),
self.rightlin(rightenco)],
axis=1)
return ret # (batsize, 2, decdim)
def _autogenerate_mask(self, seq, seqemb):
assert(seqemb.ndim == 3)
print "automasking in SeqEncoder (%s)" % __file__
axes = range(2, seq.ndim) # mask must be 2D
if "int" in seq.dtype: # ==> indexes # mask must be 2D
if seq.ndim < 2:
raise AttributeError("CAN NOT GENERATE MASK FOR NON-SEQUENCE")
elif seq.ndim == 2:
seqformask = seq
else:
print "generating default mask for non-standard seq shape (SeqEncoder, %s)" % __file__
seqformask = seq[(slice(None, None, None),) * 2 + (0,) * (seq.ndim-2)]
#if self._maskconfig.maskid != 0:
# raise AttributeError("CAN NOT CREATE MASK USING CUSTOM MASKID %d BECAUSE OF NON-STANDARD SEQ (%d dims, %s)" % (self._maskconfig.maskid, seq.ndim, str(seq.dtype)))
#mask = T.gt(seq.sum(axis=axes), 0) # 0 is TERMINUS
assert(seqformask.ndim == 2)
mask = T.neq(seqformask, self._maskconfig.maskid)
else:
#TODO raise AttributeError("CAN NOT GENERATE MASK FOR NON-INT SEQ")
mask = T.gt(seq.norm(2, axis=axes), 0)
return mask
def _maskfrom(self, ret, x):
if self.maskid is not None:
mask = T.neq(x, self.maskid)
else:
mask = None
ret.mask = mask