Rooba
/
spaCy
mirror of https://github.com/explosion/spaCy.git
1
1
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Fix GPU usage

This commit is contained in:
Matthew Honnibal 2017-10-31 02:33:16 +01:00
parent e98451b5f7
commit 7698903617
1 changed files with 9 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
spacy/syntax/nn_parser.pyx
View File

@ -89,11 +89,14 @@ cdef class precompute_hiddens:
cached = gpu_cached.get(stream=cuda_stream) cached = gpu_cached.get(stream=cuda_stream)
else: else:
cached = gpu_cached cached = gpu_cached
if not isinstance(lower_model.b, numpy.ndarray):
self.bias = lower_model.b.get()
else:
self.bias = lower_model.b
self.nF = cached.shape[1] self.nF = cached.shape[1]
self.nP = getattr(lower_model, 'nP', 1) self.nP = getattr(lower_model, 'nP', 1)
self.nO = cached.shape[2] self.nO = cached.shape[2]
self.ops = lower_model.ops self.ops = lower_model.ops
self.bias = lower_model.b
self._is_synchronized = False self._is_synchronized = False
self._cuda_stream = cuda_stream self._cuda_stream = cuda_stream
self._cached = cached self._cached = cached
@ -126,7 +129,8 @@ cdef class precompute_hiddens:
state_vector += self.bias state_vector += self.bias
state_vector, bp_nonlinearity = self._nonlinearity(state_vector) state_vector, bp_nonlinearity = self._nonlinearity(state_vector)
def backward(d_state_vector, sgd=None): def backward(d_state_vector_ids, sgd=None):
d_state_vector, token_ids = d_state_vector_ids
d_state_vector = bp_nonlinearity(d_state_vector, sgd) d_state_vector = bp_nonlinearity(d_state_vector, sgd)
# This will usually be on GPU # This will usually be on GPU
if not isinstance(d_state_vector, self.ops.xp.ndarray): if not isinstance(d_state_vector, self.ops.xp.ndarray):
@ -157,7 +161,8 @@ cdef void sum_state_features(float* output,
const float* cached, const int* token_ids, int B, int F, int O) nogil: const float* cached, const int* token_ids, int B, int F, int O) nogil:
cdef int idx, b, f, i cdef int idx, b, f, i
cdef const float* feature cdef const float* feature
padding = cached - (F * O) padding = cached
cached += F * O
for b in range(B): for b in range(B):
for f in range(F): for f in range(F):
if token_ids[f] < 0: if token_ids[f] < 0:
@ -657,7 +662,7 @@ cdef class Parser:
cuda_stream.synchronize() cuda_stream.synchronize()
xp = get_array_module(d_tokvecs) xp = get_array_module(d_tokvecs)
for ids, d_vector, bp_vector in backprops: for ids, d_vector, bp_vector in backprops:
d_state_features = bp_vector(d_vector, sgd=sgd) d_state_features = bp_vector((d_vector, ids), sgd=sgd)
ids = ids.flatten() ids = ids.flatten()
d_state_features = d_state_features.reshape( d_state_features = d_state_features.reshape(
(ids.size, d_state_features.shape[2])) (ids.size, d_state_features.shape[2]))