Skip to content
体验新版
项目
组织
正在加载...
登录
切换导航
打开侧边栏
PaddlePaddle
models
提交
34601005
M
models
项目概览
PaddlePaddle
/
models
大约 1 年 前同步成功
通知
222
Star
6828
Fork
2962
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
602
列表
看板
标记
里程碑
合并请求
255
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
M
models
项目概览
项目概览
详情
发布
仓库
仓库
文件
提交
分支
标签
贡献者
分支图
比较
Issue
602
Issue
602
列表
看板
标记
里程碑
合并请求
255
合并请求
255
Pages
分析
分析
仓库分析
DevOps
Wiki
0
Wiki
成员
成员
收起侧边栏
关闭侧边栏
动态
分支图
创建新Issue
提交
Issue看板
提交
34601005
编写于
6月 19, 2018
作者:
G
guosheng
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
Refine Transformer for wmt14_en-de
上级
f629ce07
变更
4
隐藏空白更改
内联
并排
Showing
4 changed file
with
117 addition
and
102 deletion
+117
-102
fluid/neural_machine_translation/transformer/config.py
fluid/neural_machine_translation/transformer/config.py
+15
-10
fluid/neural_machine_translation/transformer/infer.py
fluid/neural_machine_translation/transformer/infer.py
+1
-0
fluid/neural_machine_translation/transformer/model.py
fluid/neural_machine_translation/transformer/model.py
+38
-33
fluid/neural_machine_translation/transformer/train.py
fluid/neural_machine_translation/transformer/train.py
+63
-59
未找到文件。
fluid/neural_machine_translation/transformer/config.py
浏览文件 @
34601005
class
TrainTaskConfig
(
object
):
class
TrainTaskConfig
(
object
):
# only support GPU currently
use_gpu
=
True
use_gpu
=
True
# the epoch number to train.
# the epoch number to train.
pass_num
=
30
pass_num
=
30
# the number of sequences contained in a mini-batch.
# the number of sequences contained in a mini-batch.
# deprecated, set batch_size in args.
batch_size
=
32
batch_size
=
32
# the hyper parameters for Adam optimizer.
# the hyper parameters for Adam optimizer.
# This static learning_rate will be multiplied to the LearningRateScheduler
# This static learning_rate will be multiplied to the LearningRateScheduler
...
@@ -13,8 +15,6 @@ class TrainTaskConfig(object):
...
@@ -13,8 +15,6 @@ class TrainTaskConfig(object):
eps
=
1e-9
eps
=
1e-9
# the parameters for learning rate scheduling.
# the parameters for learning rate scheduling.
warmup_steps
=
4000
warmup_steps
=
4000
# the flag indicating to use average loss or sum loss when training.
use_avg_cost
=
True
# the weight used to mix up the ground-truth distribution and the fixed
# the weight used to mix up the ground-truth distribution and the fixed
# uniform distribution in label smoothing when training.
# uniform distribution in label smoothing when training.
# Set this as zero if label smoothing is not wanted.
# Set this as zero if label smoothing is not wanted.
...
@@ -38,22 +38,20 @@ class InferTaskConfig(object):
...
@@ -38,22 +38,20 @@ class InferTaskConfig(object):
batch_size
=
10
batch_size
=
10
# the parameters for beam search.
# the parameters for beam search.
beam_size
=
5
beam_size
=
5
max_length
=
30
max_length
=
256
# the number of decoded sentences to output.
# the number of decoded sentences to output.
n_best
=
1
n_best
=
1
# the flags indicating whether to output the special tokens.
# the flags indicating whether to output the special tokens.
output_bos
=
False
output_bos
=
False
output_eos
=
False
output_eos
=
False
output_unk
=
Fals
e
output_unk
=
Tru
e
# the directory for loading the trained model.
# the directory for loading the trained model.
model_path
=
"trained_models/pass_1.infer.model"
model_path
=
"trained_models/pass_1.infer.model"
class
ModelHyperParams
(
object
):
class
ModelHyperParams
(
object
):
# This model directly uses paddle.dataset.wmt16 in which <bos>, <eos> and
# These following five vocabularies related configurations will be set
# <unk> token has alreay been added. As for the <pad> token, any token
# automatically according to the passed vocabulary path and special tokens.
# included in dict can be used to pad, since the paddings' loss will be
# masked out and make no effect on parameter gradients.
# size of source word dictionary.
# size of source word dictionary.
src_vocab_size
=
10000
src_vocab_size
=
10000
# size of target word dictionay
# size of target word dictionay
...
@@ -68,13 +66,13 @@ class ModelHyperParams(object):
...
@@ -68,13 +66,13 @@ class ModelHyperParams(object):
# The size of position encoding table should at least plus 1, since the
# The size of position encoding table should at least plus 1, since the
# sinusoid position encoding starts from 1 and 0 can be used as the padding
# sinusoid position encoding starts from 1 and 0 can be used as the padding
# token for position encoding.
# token for position encoding.
max_length
=
50
max_length
=
256
# the dimension for word embeddings, which is also the last dimension of
# the dimension for word embeddings, which is also the last dimension of
# the input and output of multi-head attention, position-wise feed-forward
# the input and output of multi-head attention, position-wise feed-forward
# networks, encoder and decoder.
# networks, encoder and decoder.
d_model
=
512
d_model
=
512
# size of the hidden layer in position-wise feed-forward networks.
# size of the hidden layer in position-wise feed-forward networks.
d_inner_hid
=
1024
d_inner_hid
=
2048
# the dimension that keys are projected to for dot-product attention.
# the dimension that keys are projected to for dot-product attention.
d_key
=
64
d_key
=
64
# the dimension that values are projected to for dot-product attention.
# the dimension that values are projected to for dot-product attention.
...
@@ -85,6 +83,9 @@ class ModelHyperParams(object):
...
@@ -85,6 +83,9 @@ class ModelHyperParams(object):
n_layer
=
6
n_layer
=
6
# dropout rate used by all dropout layers.
# dropout rate used by all dropout layers.
dropout
=
0.1
dropout
=
0.1
# the flag indicating whether to share embedding and softmax weights.
# vocabularies in source and target should be same for weight sharing.
weight_sharing
=
True
def
merge_cfg_from_list
(
cfg_list
,
g_cfgs
):
def
merge_cfg_from_list
(
cfg_list
,
g_cfgs
):
...
@@ -172,6 +173,10 @@ input_descs = {
...
@@ -172,6 +173,10 @@ input_descs = {
"lbl_weight"
:
[(
1
*
(
ModelHyperParams
.
max_length
+
1
),
1L
),
"float32"
],
"lbl_weight"
:
[(
1
*
(
ModelHyperParams
.
max_length
+
1
),
1L
),
"float32"
],
}
}
# Names of word embedding table which might be reused for weight sharing.
word_emb_param_names
=
(
"src_word_emb_table"
,
"trg_word_emb_table"
,
)
# Names of position encoding table which will be initialized externally.
# Names of position encoding table which will be initialized externally.
pos_enc_param_names
=
(
pos_enc_param_names
=
(
"src_pos_enc_table"
,
"src_pos_enc_table"
,
...
...
fluid/neural_machine_translation/transformer/infer.py
浏览文件 @
34601005
...
@@ -359,6 +359,7 @@ def infer(args):
...
@@ -359,6 +359,7 @@ def infer(args):
start_mark
=
args
.
special_token
[
0
],
start_mark
=
args
.
special_token
[
0
],
end_mark
=
args
.
special_token
[
1
],
end_mark
=
args
.
special_token
[
1
],
unk_mark
=
args
.
special_token
[
2
],
unk_mark
=
args
.
special_token
[
2
],
max_length
=
ModelHyperParams
.
max_length
,
clip_last_batch
=
False
)
clip_last_batch
=
False
)
trg_idx2word
=
test_data
.
load_dict
(
trg_idx2word
=
test_data
.
load_dict
(
...
...
fluid/neural_machine_translation/transformer/model.py
浏览文件 @
34601005
...
@@ -46,26 +46,14 @@ def multi_head_attention(queries,
...
@@ -46,26 +46,14 @@ def multi_head_attention(queries,
"""
"""
q
=
layers
.
fc
(
input
=
queries
,
q
=
layers
.
fc
(
input
=
queries
,
size
=
d_key
*
n_head
,
size
=
d_key
*
n_head
,
param_attr
=
fluid
.
initializer
.
Xavier
(
uniform
=
False
,
fan_in
=
d_model
*
d_key
,
fan_out
=
n_head
*
d_key
),
bias_attr
=
False
,
bias_attr
=
False
,
num_flatten_dims
=
2
)
num_flatten_dims
=
2
)
k
=
layers
.
fc
(
input
=
keys
,
k
=
layers
.
fc
(
input
=
keys
,
size
=
d_key
*
n_head
,
size
=
d_key
*
n_head
,
param_attr
=
fluid
.
initializer
.
Xavier
(
uniform
=
False
,
fan_in
=
d_model
*
d_key
,
fan_out
=
n_head
*
d_key
),
bias_attr
=
False
,
bias_attr
=
False
,
num_flatten_dims
=
2
)
num_flatten_dims
=
2
)
v
=
layers
.
fc
(
input
=
values
,
v
=
layers
.
fc
(
input
=
values
,
size
=
d_value
*
n_head
,
size
=
d_value
*
n_head
,
param_attr
=
fluid
.
initializer
.
Xavier
(
uniform
=
False
,
fan_in
=
d_model
*
d_value
,
fan_out
=
n_head
*
d_value
),
bias_attr
=
False
,
bias_attr
=
False
,
num_flatten_dims
=
2
)
num_flatten_dims
=
2
)
return
q
,
k
,
v
return
q
,
k
,
v
...
@@ -84,7 +72,7 @@ def multi_head_attention(queries,
...
@@ -84,7 +72,7 @@ def multi_head_attention(queries,
# The value 0 in shape attr means copying the corresponding dimension
# The value 0 in shape attr means copying the corresponding dimension
# size of the input as the output dimension size.
# size of the input as the output dimension size.
reshaped
=
layers
.
reshape
(
reshaped
=
layers
.
reshape
(
x
=
x
,
shape
=
[
0
,
-
1
,
n_head
,
hidden_size
//
n_head
])
x
=
x
,
shape
=
[
0
,
0
,
n_head
,
hidden_size
//
n_head
])
# permuate the dimensions into:
# permuate the dimensions into:
# [batch_size, n_head, max_sequence_len, hidden_size_per_head]
# [batch_size, n_head, max_sequence_len, hidden_size_per_head]
...
@@ -104,7 +92,7 @@ def multi_head_attention(queries,
...
@@ -104,7 +92,7 @@ def multi_head_attention(queries,
# size of the input as the output dimension size.
# size of the input as the output dimension size.
return
layers
.
reshape
(
return
layers
.
reshape
(
x
=
trans_x
,
x
=
trans_x
,
shape
=
map
(
int
,
[
0
,
-
1
,
trans_x
.
shape
[
2
]
*
trans_x
.
shape
[
3
]]))
shape
=
map
(
int
,
[
0
,
0
,
trans_x
.
shape
[
2
]
*
trans_x
.
shape
[
3
]]))
def
scaled_dot_product_attention
(
q
,
k
,
v
,
attn_bias
,
d_model
,
dropout_rate
):
def
scaled_dot_product_attention
(
q
,
k
,
v
,
attn_bias
,
d_model
,
dropout_rate
):
"""
"""
...
@@ -140,7 +128,6 @@ def multi_head_attention(queries,
...
@@ -140,7 +128,6 @@ def multi_head_attention(queries,
# Project back to the model size.
# Project back to the model size.
proj_out
=
layers
.
fc
(
input
=
out
,
proj_out
=
layers
.
fc
(
input
=
out
,
size
=
d_model
,
size
=
d_model
,
param_attr
=
fluid
.
initializer
.
Xavier
(
uniform
=
False
),
bias_attr
=
False
,
bias_attr
=
False
,
num_flatten_dims
=
2
)
num_flatten_dims
=
2
)
return
proj_out
return
proj_out
...
@@ -155,14 +142,8 @@ def positionwise_feed_forward(x, d_inner_hid, d_hid):
...
@@ -155,14 +142,8 @@ def positionwise_feed_forward(x, d_inner_hid, d_hid):
hidden
=
layers
.
fc
(
input
=
x
,
hidden
=
layers
.
fc
(
input
=
x
,
size
=
d_inner_hid
,
size
=
d_inner_hid
,
num_flatten_dims
=
2
,
num_flatten_dims
=
2
,
param_attr
=
fluid
.
initializer
.
Uniform
(
low
=-
(
d_hid
**-
0.5
),
high
=
(
d_hid
**-
0.5
)),
act
=
"relu"
)
act
=
"relu"
)
out
=
layers
.
fc
(
input
=
hidden
,
out
=
layers
.
fc
(
input
=
hidden
,
size
=
d_hid
,
num_flatten_dims
=
2
)
size
=
d_hid
,
num_flatten_dims
=
2
,
param_attr
=
fluid
.
initializer
.
Uniform
(
low
=-
(
d_inner_hid
**-
0.5
),
high
=
(
d_inner_hid
**-
0.5
)))
return
out
return
out
...
@@ -200,6 +181,7 @@ def prepare_encoder(src_word,
...
@@ -200,6 +181,7 @@ def prepare_encoder(src_word,
src_max_len
,
src_max_len
,
dropout_rate
=
0.
,
dropout_rate
=
0.
,
src_data_shape
=
None
,
src_data_shape
=
None
,
word_emb_param_name
=
None
,
pos_enc_param_name
=
None
):
pos_enc_param_name
=
None
):
"""Add word embeddings and position encodings.
"""Add word embeddings and position encodings.
The output tensor has a shape of:
The output tensor has a shape of:
...
@@ -209,7 +191,10 @@ def prepare_encoder(src_word,
...
@@ -209,7 +191,10 @@ def prepare_encoder(src_word,
src_word_emb
=
layers
.
embedding
(
src_word_emb
=
layers
.
embedding
(
src_word
,
src_word
,
size
=
[
src_vocab_size
,
src_emb_dim
],
size
=
[
src_vocab_size
,
src_emb_dim
],
param_attr
=
fluid
.
initializer
.
Normal
(
0.
,
1.
))
param_attr
=
fluid
.
ParamAttr
(
name
=
word_emb_param_name
,
initializer
=
fluid
.
initializer
.
Normal
(
0.
,
src_emb_dim
**-
0.5
)))
src_word_emb
=
layers
.
scale
(
x
=
src_word_emb
,
scale
=
src_emb_dim
**
0.5
)
src_pos_enc
=
layers
.
embedding
(
src_pos_enc
=
layers
.
embedding
(
src_pos
,
src_pos
,
size
=
[
src_max_len
,
src_emb_dim
],
size
=
[
src_max_len
,
src_emb_dim
],
...
@@ -415,7 +400,12 @@ def transformer(
...
@@ -415,7 +400,12 @@ def transformer(
d_model
,
d_model
,
d_inner_hid
,
d_inner_hid
,
dropout_rate
,
dropout_rate
,
weight_sharing
,
label_smooth_eps
,
):
label_smooth_eps
,
):
if
weight_sharing
:
assert
src_vocab_size
==
src_vocab_size
,
(
"Vocabularies in source and target should be same for weight sharing."
)
enc_inputs
=
make_all_inputs
(
encoder_data_input_fields
+
enc_inputs
=
make_all_inputs
(
encoder_data_input_fields
+
encoder_util_input_fields
)
encoder_util_input_fields
)
...
@@ -429,6 +419,7 @@ def transformer(
...
@@ -429,6 +419,7 @@ def transformer(
d_model
,
d_model
,
d_inner_hid
,
d_inner_hid
,
dropout_rate
,
dropout_rate
,
weight_sharing
,
enc_inputs
,
)
enc_inputs
,
)
dec_inputs
=
make_all_inputs
(
decoder_data_input_fields
[:
-
1
]
+
dec_inputs
=
make_all_inputs
(
decoder_data_input_fields
[:
-
1
]
+
...
@@ -444,6 +435,7 @@ def transformer(
...
@@ -444,6 +435,7 @@ def transformer(
d_model
,
d_model
,
d_inner_hid
,
d_inner_hid
,
dropout_rate
,
dropout_rate
,
weight_sharing
,
dec_inputs
,
dec_inputs
,
enc_output
,
)
enc_output
,
)
...
@@ -459,7 +451,6 @@ def transformer(
...
@@ -459,7 +451,6 @@ def transformer(
logits
=
predict
,
logits
=
predict
,
label
=
label
,
label
=
label
,
soft_label
=
True
if
label_smooth_eps
else
False
)
soft_label
=
True
if
label_smooth_eps
else
False
)
# cost = layers.softmax_with_cross_entropy(logits=predict, label=gold)
weighted_cost
=
cost
*
weights
weighted_cost
=
cost
*
weights
sum_cost
=
layers
.
reduce_sum
(
weighted_cost
)
sum_cost
=
layers
.
reduce_sum
(
weighted_cost
)
token_num
=
layers
.
reduce_sum
(
weights
)
token_num
=
layers
.
reduce_sum
(
weights
)
...
@@ -476,6 +467,7 @@ def wrap_encoder(src_vocab_size,
...
@@ -476,6 +467,7 @@ def wrap_encoder(src_vocab_size,
d_model
,
d_model
,
d_inner_hid
,
d_inner_hid
,
dropout_rate
,
dropout_rate
,
weight_sharing
,
enc_inputs
=
None
):
enc_inputs
=
None
):
"""
"""
The wrapper assembles together all needed layers for the encoder.
The wrapper assembles together all needed layers for the encoder.
...
@@ -497,7 +489,8 @@ def wrap_encoder(src_vocab_size,
...
@@ -497,7 +489,8 @@ def wrap_encoder(src_vocab_size,
d_model
,
d_model
,
max_length
,
max_length
,
dropout_rate
,
dropout_rate
,
src_data_shape
,
)
src_data_shape
,
word_emb_param_name
=
word_emb_param_names
[
0
])
enc_output
=
encoder
(
enc_output
=
encoder
(
enc_input
,
enc_input
,
src_slf_attn_bias
,
src_slf_attn_bias
,
...
@@ -522,6 +515,7 @@ def wrap_decoder(trg_vocab_size,
...
@@ -522,6 +515,7 @@ def wrap_decoder(trg_vocab_size,
d_model
,
d_model
,
d_inner_hid
,
d_inner_hid
,
dropout_rate
,
dropout_rate
,
weight_sharing
,
dec_inputs
=
None
,
dec_inputs
=
None
,
enc_output
=
None
):
enc_output
=
None
):
"""
"""
...
@@ -547,7 +541,9 @@ def wrap_decoder(trg_vocab_size,
...
@@ -547,7 +541,9 @@ def wrap_decoder(trg_vocab_size,
d_model
,
d_model
,
max_length
,
max_length
,
dropout_rate
,
dropout_rate
,
trg_data_shape
,
)
trg_data_shape
,
word_emb_param_name
=
word_emb_param_names
[
0
]
if
weight_sharing
else
word_emb_param_names
[
1
])
dec_output
=
decoder
(
dec_output
=
decoder
(
dec_input
,
dec_input
,
enc_output
,
enc_output
,
...
@@ -565,11 +561,20 @@ def wrap_decoder(trg_vocab_size,
...
@@ -565,11 +561,20 @@ def wrap_decoder(trg_vocab_size,
src_attn_pre_softmax_shape
,
src_attn_pre_softmax_shape
,
src_attn_post_softmax_shape
,
)
src_attn_post_softmax_shape
,
)
# Return logits for training and probs for inference.
# Return logits for training and probs for inference.
predict
=
layers
.
reshape
(
if
weight_sharing
:
x
=
layers
.
fc
(
input
=
dec_output
,
predict
=
layers
.
reshape
(
size
=
trg_vocab_size
,
x
=
layers
.
matmul
(
bias_attr
=
False
,
x
=
dec_output
,
num_flatten_dims
=
2
),
y
=
fluid
.
get_var
(
word_emb_param_names
[
0
]),
shape
=
[
-
1
,
trg_vocab_size
],
transpose_y
=
True
),
act
=
"softmax"
if
dec_inputs
is
None
else
None
)
shape
=
[
-
1
,
trg_vocab_size
],
act
=
"softmax"
if
dec_inputs
is
None
else
None
)
else
:
predict
=
layers
.
reshape
(
x
=
layers
.
fc
(
input
=
dec_output
,
size
=
trg_vocab_size
,
bias_attr
=
False
,
num_flatten_dims
=
2
),
shape
=
[
-
1
,
trg_vocab_size
],
act
=
"softmax"
if
dec_inputs
is
None
else
None
)
return
predict
return
predict
fluid/neural_machine_translation/transformer/train.py
浏览文件 @
34601005
...
@@ -203,50 +203,50 @@ def prepare_batch_input(insts, data_input_names, util_input_names, src_pad_idx,
...
@@ -203,50 +203,50 @@ def prepare_batch_input(insts, data_input_names, util_input_names, src_pad_idx,
[
num_token
],
dtype
=
"float32"
)
[
num_token
],
dtype
=
"float32"
)
def
train
(
args
):
def
read_multiple
(
reader
,
count
,
clip_last
=
True
):
dev_count
=
fluid
.
core
.
get_cuda_device_count
()
"""
Stack data from reader for multi-devices.
"""
def
read_multiple
(
reader
,
def
__impl__
():
count
=
dev_count
if
args
.
use_token_batch
else
1
,
res
=
[]
clip_last
=
True
):
for
item
in
reader
():
"""
res
.
append
(
item
)
Stack data from reader for multi-devices.
"""
def
__impl__
():
res
=
[]
for
item
in
reader
():
res
.
append
(
item
)
if
len
(
res
)
==
count
:
yield
res
res
=
[]
if
len
(
res
)
==
count
:
if
len
(
res
)
==
count
:
yield
res
yield
res
elif
not
clip_last
:
res
=
[]
data
=
[]
if
len
(
res
)
==
count
:
for
item
in
res
:
yield
res
data
+=
item
elif
not
clip_last
:
if
len
(
data
)
>
count
:
data
=
[]
inst_num_per_part
=
len
(
data
)
//
count
for
item
in
res
:
yield
[
data
+=
item
data
[
inst_num_per_part
*
i
:
inst_num_per_part
*
(
i
+
1
)]
if
len
(
data
)
>
count
:
for
i
in
range
(
count
)
inst_num_per_part
=
len
(
data
)
//
count
]
yield
[
data
[
inst_num_per_part
*
i
:
inst_num_per_part
*
(
i
+
1
)]
return
__impl__
for
i
in
range
(
count
)
]
def
split_data
(
data
,
num_part
=
dev_count
):
"""
return
__impl__
Split data for each device.
"""
if
len
(
data
)
==
num_part
:
def
split_data
(
data
,
num_part
):
return
data
"""
data
=
data
[
0
]
Split data for each device.
inst_num_per_part
=
len
(
data
)
//
num_part
"""
return
[
if
len
(
data
)
==
num_part
:
data
[
inst_num_per_part
*
i
:
inst_num_per_part
*
(
i
+
1
)]
return
data
for
i
in
range
(
num_part
)
data
=
data
[
0
]
]
inst_num_per_part
=
len
(
data
)
//
num_part
return
[
data
[
inst_num_per_part
*
i
:
inst_num_per_part
*
(
i
+
1
)]
for
i
in
range
(
num_part
)
]
def
train
(
args
):
dev_count
=
fluid
.
core
.
get_cuda_device_count
()
sum_cost
,
avg_cost
,
predict
,
token_num
=
transformer
(
sum_cost
,
avg_cost
,
predict
,
token_num
=
transformer
(
ModelHyperParams
.
src_vocab_size
,
ModelHyperParams
.
trg_vocab_size
,
ModelHyperParams
.
src_vocab_size
,
ModelHyperParams
.
trg_vocab_size
,
...
@@ -254,7 +254,7 @@ def train(args):
...
@@ -254,7 +254,7 @@ def train(args):
ModelHyperParams
.
n_head
,
ModelHyperParams
.
d_key
,
ModelHyperParams
.
n_head
,
ModelHyperParams
.
d_key
,
ModelHyperParams
.
d_value
,
ModelHyperParams
.
d_model
,
ModelHyperParams
.
d_value
,
ModelHyperParams
.
d_model
,
ModelHyperParams
.
d_inner_hid
,
ModelHyperParams
.
dropout
,
ModelHyperParams
.
d_inner_hid
,
ModelHyperParams
.
dropout
,
TrainTaskConfig
.
label_smooth_eps
)
ModelHyperParams
.
weight_sharing
,
TrainTaskConfig
.
label_smooth_eps
)
lr_scheduler
=
LearningRateScheduler
(
ModelHyperParams
.
d_model
,
lr_scheduler
=
LearningRateScheduler
(
ModelHyperParams
.
d_model
,
TrainTaskConfig
.
warmup_steps
,
TrainTaskConfig
.
warmup_steps
,
...
@@ -288,9 +288,12 @@ def train(args):
...
@@ -288,9 +288,12 @@ def train(args):
start_mark
=
args
.
special_token
[
0
],
start_mark
=
args
.
special_token
[
0
],
end_mark
=
args
.
special_token
[
1
],
end_mark
=
args
.
special_token
[
1
],
unk_mark
=
args
.
special_token
[
2
],
unk_mark
=
args
.
special_token
[
2
],
max_length
=
ModelHyperParams
.
max_length
,
clip_last_batch
=
False
)
clip_last_batch
=
False
)
train_data
=
read_multiple
(
reader
=
train_data
.
batch_generator
,
count
=
dev_count
if
args
.
use_token_batch
else
1
)
train_data
=
read_multiple
(
reader
=
train_data
.
batch_generator
)
build_strategy
=
fluid
.
BuildStrategy
()
build_strategy
=
fluid
.
BuildStrategy
()
# Since the token number differs among devices, customize gradient scale to
# Since the token number differs among devices, customize gradient scale to
# use token average cost among multi-devices. and the gradient scale is
# use token average cost among multi-devices. and the gradient scale is
...
@@ -303,9 +306,11 @@ def train(args):
...
@@ -303,9 +306,11 @@ def train(args):
def
test_context
():
def
test_context
():
# Context to do validation.
# Context to do validation.
test_program
=
fluid
.
default_main_program
().
clone
()
test_program
=
fluid
.
default_main_program
().
clone
(
for_test
=
True
)
with
fluid
.
program_guard
(
test_program
):
test_exe
=
fluid
.
ParallelExecutor
(
test_program
=
fluid
.
io
.
get_inference_program
([
avg_cost
])
use_cuda
=
TrainTaskConfig
.
use_gpu
,
main_program
=
test_program
,
share_vars_from
=
train_exe
)
val_data
=
reader
.
DataReader
(
val_data
=
reader
.
DataReader
(
src_vocab_fpath
=
args
.
src_vocab_fpath
,
src_vocab_fpath
=
args
.
src_vocab_fpath
,
...
@@ -319,22 +324,22 @@ def train(args):
...
@@ -319,22 +324,22 @@ def train(args):
start_mark
=
args
.
special_token
[
0
],
start_mark
=
args
.
special_token
[
0
],
end_mark
=
args
.
special_token
[
1
],
end_mark
=
args
.
special_token
[
1
],
unk_mark
=
args
.
special_token
[
2
],
unk_mark
=
args
.
special_token
[
2
],
max_length
=
ModelHyperParams
.
max_length
,
clip_last_batch
=
False
,
clip_last_batch
=
False
,
shuffle
=
False
,
shuffle
=
False
,
shuffle_batch
=
False
)
shuffle_batch
=
False
)
test_exe
=
fluid
.
ParallelExecutor
(
use_cuda
=
TrainTaskConfig
.
use_gpu
,
main_program
=
test_program
,
share_vars_from
=
train_exe
)
def
test
(
exe
=
test_exe
):
def
test
(
exe
=
test_exe
):
test_total_cost
=
0
test_total_cost
=
0
test_total_token
=
0
test_total_token
=
0
test_data
=
read_multiple
(
reader
=
val_data
.
batch_generator
)
test_data
=
read_multiple
(
reader
=
val_data
.
batch_generator
,
count
=
dev_count
if
args
.
use_token_batch
else
1
)
for
batch_id
,
data
in
enumerate
(
test_data
()):
for
batch_id
,
data
in
enumerate
(
test_data
()):
feed_list
=
[]
feed_list
=
[]
for
place_id
,
data_buffer
in
enumerate
(
split_data
(
data
)):
for
place_id
,
data_buffer
in
enumerate
(
split_data
(
data
,
num_part
=
dev_count
)):
data_input_dict
,
util_input_dict
,
_
=
prepare_batch_input
(
data_input_dict
,
util_input_dict
,
_
=
prepare_batch_input
(
data_buffer
,
data_input_names
,
util_input_names
,
data_buffer
,
data_input_names
,
util_input_names
,
ModelHyperParams
.
eos_idx
,
ModelHyperParams
.
eos_idx
,
ModelHyperParams
.
eos_idx
,
ModelHyperParams
.
eos_idx
,
...
@@ -367,7 +372,9 @@ def train(args):
...
@@ -367,7 +372,9 @@ def train(args):
feed_list
=
[]
feed_list
=
[]
total_num_token
=
0
total_num_token
=
0
lr_rate
=
lr_scheduler
.
update_learning_rate
()
lr_rate
=
lr_scheduler
.
update_learning_rate
()
for
place_id
,
data_buffer
in
enumerate
(
split_data
(
data
)):
for
place_id
,
data_buffer
in
enumerate
(
split_data
(
data
,
num_part
=
dev_count
)):
data_input_dict
,
util_input_dict
,
num_token
=
prepare_batch_input
(
data_input_dict
,
util_input_dict
,
num_token
=
prepare_batch_input
(
data_buffer
,
data_input_names
,
util_input_names
,
data_buffer
,
data_input_names
,
util_input_names
,
ModelHyperParams
.
eos_idx
,
ModelHyperParams
.
eos_idx
,
ModelHyperParams
.
eos_idx
,
ModelHyperParams
.
eos_idx
,
...
@@ -377,17 +384,14 @@ def train(args):
...
@@ -377,17 +384,14 @@ def train(args):
dict
(
data_input_dict
.
items
()
+
util_input_dict
.
items
()
+
dict
(
data_input_dict
.
items
()
+
util_input_dict
.
items
()
+
{
lr_scheduler
.
learning_rate
.
name
:
lr_rate
}.
items
()))
{
lr_scheduler
.
learning_rate
.
name
:
lr_rate
}.
items
()))
if
not
init
:
if
not
init
:
# init the position encoding table
for
pos_enc_param_name
in
pos_enc_param_names
:
for
pos_enc_param_name
in
pos_enc_param_names
:
pos_enc
=
position_encoding_init
(
pos_enc
=
position_encoding_init
(
ModelHyperParams
.
max_length
+
1
,
ModelHyperParams
.
max_length
+
1
,
ModelHyperParams
.
d_model
)
ModelHyperParams
.
d_model
)
feed_list
[
place_id
][
pos_enc_param_name
]
=
pos_enc
feed_list
[
place_id
][
pos_enc_param_name
]
=
pos_enc
for
feed_dict
in
feed_list
:
for
feed_dict
in
feed_list
:
feed_dict
[
feed_dict
[
sum_cost
.
name
+
"@GRAD"
]
=
1.
/
total_num_token
sum_cost
.
name
+
"@GRAD"
]
=
1.
/
total_num_token
if
TrainTaskConfig
.
use_avg_cost
else
np
.
asarray
(
[
1.
],
dtype
=
"float32"
)
outs
=
train_exe
.
run
(
fetch_list
=
[
sum_cost
.
name
,
token_num
.
name
],
outs
=
train_exe
.
run
(
fetch_list
=
[
sum_cost
.
name
,
token_num
.
name
],
feed
=
feed_list
)
feed
=
feed_list
)
sum_cost_val
,
token_num_val
=
np
.
array
(
outs
[
0
]),
np
.
array
(
outs
[
1
])
sum_cost_val
,
token_num_val
=
np
.
array
(
outs
[
0
]),
np
.
array
(
outs
[
1
])
...
...
编辑
预览
Markdown
is supported
0%
请重试
或
添加新附件
.
添加附件
取消
You are about to add
0
people
to the discussion. Proceed with caution.
先完成此消息的编辑!
取消
想要评论请
注册
或
登录