Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Dev/update ptb #4083

Merged
merged 6 commits into from
Dec 18, 2019
Merged

Dev/update ptb #4083

Changes from 1 commit
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
9ea1546
update ptb model to use auto generated op functions, test=develop
zhiqiu Oct 21, 2019
dbad8d5
Merge branch 'develop' of https://github.com/PaddlePaddle/models into…
zhiqiu Dec 4, 2019
40d8ca4
refine model code, test=develop
zhiqiu Dec 4, 2019
79e0d49
refine model code, test=develop
zhiqiu Dec 4, 2019
62416c4
update ptb model, test=develop
zhiqiu Dec 17, 2019
9981501
test=develop, Merge branch 'develop' of https://github.com/PaddlePadd…
zhiqiu Dec 18, 2019
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Next Next commit
update ptb model to use auto generated op functions, test=develop
  • Loading branch information
@zhiqiu
zhiqiu committed Oct 21, 2019
commit 9ea154657dc761f0053c319fa7a54b64936587a3
68 changes: 47 additions & 21 deletions dygraph/ptb_lm/ptb_dy.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,7 +37,7 @@
reload(sys)
sys.setdefaultencoding("utf-8")


origin = 0
class SimpleLSTMRNN(fluid.Layer):
def __init__(self,
name_scope,
Expand Down Expand Up @@ -70,15 +70,21 @@ def __init__(self,
dtype="float32",
default_initializer=fluid.initializer.UniformInitializer(
low=-self._init_scale, high=self._init_scale))
self.weight_1_arr.append(self.add_parameter('w_%d' % i, weight_1))
if origin:
self.weight_1_arr.append(self.add_parameter('w_%d' % i, weight_1))
else:
self.weight_1_arr.append(self.add_parameter('w_%d' % i, weight_1)._ivar)
bias_1 = self.create_parameter(
attr=fluid.ParamAttr(
initializer=fluid.initializer.UniformInitializer(
low=-self._init_scale, high=self._init_scale)),
shape=[self._hidden_size * 4],
dtype="float32",
default_initializer=fluid.initializer.Constant(0.0))
self.bias_arr.append(self.add_parameter('b_%d' % i, bias_1))
if origin:
self.bias_arr.append(self.add_parameter('b_%d' % i, bias_1))
else:
self.bias_arr.append(self.add_parameter('b_%d' % i, bias_1)._ivar)

def forward(self, input_embedding, init_hidden=None, init_cell=None):
self.cell_array = []
Expand Down Expand Up @@ -114,9 +120,13 @@ def forward(self, input_embedding, init_hidden=None, init_cell=None):
gate_input = fluid.layers.elementwise_add(gate_input, bias)
i, j, f, o = fluid.layers.split(
gate_input, num_or_sections=4, dim=-1)
c = pre_cell * fluid.layers.sigmoid(f) + fluid.layers.sigmoid(
i) * fluid.layers.tanh(j)
m = fluid.layers.tanh(c) * fluid.layers.sigmoid(o)
# c = pre_cell * fluid.layers.sigmoid(f) + fluid.layers.sigmoid(
# i) * fluid.layers.tanh(j)
# m = fluid.layers.tanh(c) * fluid.layers.sigmoid(o)
t1 = fluid.layers.elementwise_mul(pre_cell, fluid.layers.sigmoid(f))
t2 = fluid.layers.elementwise_mul(fluid.layers.sigmoid(i), fluid.layers.tanh(j))
c = fluid.layers.elementwise_add(t1, t2)
m = fluid.layers.elementwise_mul(fluid.layers.tanh(c), fluid.layers.sigmoid(o))
self.hidden_array[k] = m
self.cell_array[k] = c
self._input = m
Expand Down Expand Up @@ -180,12 +190,16 @@ def __init__(self,
dtype="float32",
default_initializer=fluid.initializer.UniformInitializer(
low=-self.init_scale, high=self.init_scale))
if not origin:
self.softmax_weight = self.softmax_weight._ivar
self.softmax_bias = self.create_parameter(
attr=fluid.ParamAttr(),
shape=[self.vocab_size],
dtype="float32",
default_initializer=fluid.initializer.UniformInitializer(
low=-self.init_scale, high=self.init_scale))
if not origin:
self.softmax_bias = self.softmax_bias._ivar

def build_once(self, input, label, init_hidden, init_cell):
pass
Expand Down Expand Up @@ -223,7 +237,7 @@ def forward(self, input, label, init_hidden, init_cell):
loss = fluid.layers.reshape(loss, shape=[-1, self.num_steps])
loss = fluid.layers.reduce_mean(loss, dim=[0])
loss = fluid.layers.reduce_sum(loss)
loss.permissions = True
#loss.persistable = True

return loss, last_hidden, last_cell

Expand Down Expand Up @@ -322,7 +336,10 @@ def train_ptb_lm():

batch_len = len(train_data) // batch_size
total_batch_size = (batch_len - 1) // num_steps
print(total_batch_size)

log_interval = total_batch_size // 20
print(log_interval)

bd = []
lr_arr = [1.0]
Expand Down Expand Up @@ -357,10 +374,13 @@ def eval(model, data):
dy_loss, last_hidden, last_cell = ptb_model(x, y, init_hidden,
init_cell)

out_loss = dy_loss.numpy()
#out_loss = dy_loss.numpy()
out_loss = framework._var_base_to_np(dy_loss)

init_hidden_data = last_hidden.numpy()
init_cell_data = last_cell.numpy()
# init_hidden_data = last_hidden.numpy()
# init_cell_data = last_cell.numpy()
init_hidden_data = framework._var_base_to_np(last_hidden)
init_cell_data = framework._var_base_to_np(last_cell)

total_loss += out_loss
iters += num_steps
Expand Down Expand Up @@ -396,30 +416,36 @@ def eval(model, data):
dy_loss, last_hidden, last_cell = ptb_model(x, y, init_hidden,
init_cell)

out_loss = dy_loss.numpy()
if origin:
out_loss = dy_loss.numpy()
init_hidden_data = last_hidden.numpy()
init_cell_data = last_cell.numpy()
else:
out_loss = framework._var_base_to_np(dy_loss)
init_hidden_data = framework._var_base_to_np(last_hidden)
init_cell_data = framework._var_base_to_np(last_cell)

init_hidden_data = last_hidden.numpy()
init_cell_data = last_cell.numpy()
dy_loss.backward()
sgd.minimize(dy_loss, grad_clip=grad_clip)



ptb_model.clear_gradients()
total_loss += out_loss
iters += num_steps

if batch_id > 0 and batch_id % log_interval == 0:
ppl = np.exp(total_loss / iters)
print(epoch_id, "ppl ", batch_id, ppl[0],
sgd._global_learning_rate().numpy())
print(epoch_id, "ppl ", batch_id, ppl[0], sgd._global_learning_rate().numpy())

print("one ecpoh finished", epoch_id)
print("time cost ", time.time() - start_time)
ppl = np.exp(total_loss / iters)
print("ppl ", epoch_id, ppl[0])
if args.ce:
print("kpis\ttrain_ppl\t%0.3f" % ppl[0])
break
# ppl = np.exp(total_loss / iters)
# print("ppl ", epoch_id, ppl[0])
# if args.ce:
# print("kpis\ttrain_ppl\t%0.3f" % ppl[0])

eval(ptb_model, test_data)
#eval(ptb_model, test_data)


train_ptb_lm()