ner-few-shot
This commit is contained in:
lilei
GitHub
parent
275cd070a3
commit
f6f375d7cf
|
|
@ -0,0 +1,56 @@
|
|||
## 快速上手
|
||||
|
||||
### 环境依赖
|
||||
|
||||
> python == 3.8
|
||||
|
||||
- torch == 1.7
|
||||
- tensorboardX ==2.4
|
||||
- transformers == 3.4.0
|
||||
- deepke
|
||||
|
||||
### 克隆代码
|
||||
```
|
||||
git clone git@github.com:zjunlp/DeepKE.git
|
||||
```
|
||||
### 使用pip安装
|
||||
|
||||
首先创建python虚拟环境，再进入虚拟环境
|
||||
|
||||
- 安装依赖: ```pip install -r requirements.txt```
|
||||
|
||||
### 使用数据进行训练预测
|
||||
|
||||
- 存放数据：在 `data` 文件夹下存放训练数据。包含conll2003，mit-movie，mit-restaurant和atis等数据集。
|
||||
|
||||
- conll2003包含以下数据：
|
||||
|
||||
- `train.txt`：存放训练数据集
|
||||
|
||||
- `dev.txt`：存放验证数据集
|
||||
|
||||
- `test.txt`：存放测试数据集
|
||||
|
||||
- `indomain-train.txt`：存放indomain数据集
|
||||
|
||||
- mit-movie, mit-restaurant和atis包含以下数据：
|
||||
|
||||
- `k-shot-train.txt`：k=[10, 20, 50, 100, 200, 500]，存放训练数据集
|
||||
|
||||
- `test.txt`：存放测试数据集
|
||||
|
||||
|
||||
- 开始训练：模型加载和保存位置以及配置可以在shell脚本中修改
|
||||
|
||||
- 训练conll2003：` python run.py ` (训练所用到参数都在conf文件夹中，修改即可)
|
||||
|
||||
- 进行few-shot训练：` python run.py +train=few_shot ` (若要加载模型，修改few_shot.yaml中的load_path)
|
||||
|
||||
|
||||
- 每次训练的日志保存在 `logs` 文件夹内，模型结果保存目录可以自定义。
|
||||
|
||||
- 进行预测： ` python predict.py `
|
||||
|
||||
|
||||
## 模型内容
|
||||
LightNER
|
||||
|
|
@ -0,0 +1,4 @@
|
|||
cwd: ???
|
||||
|
||||
defaults:
|
||||
- train/conll
|
||||
|
|
@ -0,0 +1,29 @@
|
|||
cwd: ???
|
||||
|
||||
defaults:
|
||||
seed: 1
|
||||
|
||||
bart_name: "facebook/bart-large"
|
||||
dataset_name: conll2003
|
||||
device: cuda
|
||||
|
||||
num_epochs: 30
|
||||
batch_size: 16
|
||||
learning_rate: 2e-5
|
||||
warmup_ratio: 0.01
|
||||
eval_begin_epoch: 16
|
||||
src_seq_ratio: 0.6
|
||||
tgt_max_len: 10
|
||||
num_beams: 1
|
||||
length_penalty: 1
|
||||
|
||||
use_prompt: True
|
||||
prompt_len: 10
|
||||
prompt_dim: 800
|
||||
|
||||
freeze_plm: True
|
||||
learn_weights: True
|
||||
notes: ''
|
||||
save_path: null # 模型保存路径
|
||||
load_path: load_path # 模型加载路径，不能为空
|
||||
write_path: "data/conll2003/predict.txt"
|
||||
|
|
@ -0,0 +1,25 @@
|
|||
seed: 1
|
||||
|
||||
bart_name: "/data/lilei/project/BARTNER-AMAX/facebook/bart-large"
|
||||
dataset_name: conll2003
|
||||
device: cuda
|
||||
|
||||
num_epochs: 30
|
||||
batch_size: 16
|
||||
learning_rate: 2e-5
|
||||
warmup_ratio: 0.01
|
||||
eval_begin_epoch: 16
|
||||
src_seq_ratio: 0.6
|
||||
tgt_max_len: 10
|
||||
num_beams: 1
|
||||
length_penalty: 1
|
||||
|
||||
use_prompt: True
|
||||
prompt_len: 10
|
||||
prompt_dim: 800
|
||||
|
||||
freeze_plm: True
|
||||
learn_weights: True
|
||||
save_path: save path # 模型保存路径
|
||||
load_path: null
|
||||
notes: ''
|
||||
|
|
@ -0,0 +1,25 @@
|
|||
seed: 1
|
||||
|
||||
bart_name: "/data/lilei/project/BARTNER-AMAX/facebook/bart-large"
|
||||
dataset_name: mit-movie
|
||||
device: cuda
|
||||
|
||||
num_epochs: 30
|
||||
batch_size: 16
|
||||
learning_rate: 1e-4
|
||||
warmup_ratio: 0.01
|
||||
eval_begin_epoch: 16
|
||||
src_seq_ratio: 0.8
|
||||
tgt_max_len: 10
|
||||
num_beams: 1
|
||||
length_penalty: 1
|
||||
|
||||
use_prompt: True
|
||||
prompt_len: 10
|
||||
prompt_dim: 800
|
||||
|
||||
freeze_plm: True
|
||||
learn_weights: True
|
||||
save_path: null # 模型保存路径
|
||||
load_path: null # 模型加载路径，
|
||||
notes: ''
|
||||
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
|
|
@ -0,0 +1,558 @@
|
|||
-DOCSTART- O
|
||||
|
||||
please O
|
||||
find O
|
||||
a O
|
||||
movie O
|
||||
starring O
|
||||
al B-ACTOR
|
||||
pacino I-ACTOR
|
||||
as O
|
||||
a O
|
||||
blind B-CHARACTER
|
||||
veteran I-CHARACTER
|
||||
|
||||
who O
|
||||
is O
|
||||
in O
|
||||
justin B-TITLE
|
||||
time I-TITLE
|
||||
|
||||
what O
|
||||
was O
|
||||
the O
|
||||
title O
|
||||
of O
|
||||
the O
|
||||
movie O
|
||||
of O
|
||||
julia B-ACTOR
|
||||
roberts I-ACTOR
|
||||
academy B-REVIEW
|
||||
award I-REVIEW
|
||||
winning I-REVIEW
|
||||
perfrmance I-REVIEW
|
||||
|
||||
who O
|
||||
created O
|
||||
the O
|
||||
soundtrak B-SONG
|
||||
for O
|
||||
star B-TRAILER
|
||||
wars I-TRAILER
|
||||
|
||||
who O
|
||||
was O
|
||||
the O
|
||||
voice O
|
||||
of O
|
||||
ariel B-CHARACTER
|
||||
in O
|
||||
the O
|
||||
little B-TITLE
|
||||
mermaid I-TITLE
|
||||
|
||||
find O
|
||||
clips B-TRAILER
|
||||
of O
|
||||
men B-TITLE
|
||||
in I-TITLE
|
||||
black I-TITLE
|
||||
ii I-TITLE
|
||||
|
||||
what O
|
||||
is O
|
||||
the O
|
||||
plot O
|
||||
of O
|
||||
red B-TITLE
|
||||
tails I-TITLE
|
||||
|
||||
do O
|
||||
you O
|
||||
know O
|
||||
of O
|
||||
any O
|
||||
movie O
|
||||
that O
|
||||
had O
|
||||
come B-SONG
|
||||
what I-SONG
|
||||
may I-SONG
|
||||
in O
|
||||
its O
|
||||
soundtrack O
|
||||
|
||||
i O
|
||||
want O
|
||||
to O
|
||||
know O
|
||||
what O
|
||||
movies O
|
||||
have O
|
||||
the O
|
||||
song O
|
||||
come B-SONG
|
||||
what I-SONG
|
||||
may I-SONG
|
||||
|
||||
show O
|
||||
me O
|
||||
the O
|
||||
half B-SONG
|
||||
baked I-SONG
|
||||
soundtrack I-SONG
|
||||
|
||||
find O
|
||||
me O
|
||||
a O
|
||||
list O
|
||||
of O
|
||||
bette B-ACTOR
|
||||
midler I-ACTOR
|
||||
movies O
|
||||
from O
|
||||
the O
|
||||
1980s B-YEAR
|
||||
|
||||
get O
|
||||
a O
|
||||
yul B-ACTOR
|
||||
brenner I-ACTOR
|
||||
western B-GENRE
|
||||
|
||||
play O
|
||||
trailer B-TRAILER
|
||||
on B-TITLE
|
||||
stranger I-TITLE
|
||||
tides I-TITLE
|
||||
|
||||
who O
|
||||
played O
|
||||
james B-CHARACTER
|
||||
bond I-CHARACTER
|
||||
in O
|
||||
goldfinger B-TITLE
|
||||
|
||||
what O
|
||||
is O
|
||||
an O
|
||||
ok B-RATINGS_AVERAGE
|
||||
pg B-RATING
|
||||
13 I-RATING
|
||||
adventure B-GENRE
|
||||
directed O
|
||||
by O
|
||||
mac B-DIRECTOR
|
||||
carter I-DIRECTOR
|
||||
|
||||
who O
|
||||
is O
|
||||
the O
|
||||
main B-CHARACTER
|
||||
character O
|
||||
of O
|
||||
a B-TITLE
|
||||
bronx I-TITLE
|
||||
tale I-TITLE
|
||||
|
||||
find O
|
||||
a O
|
||||
preview B-TRAILER
|
||||
for O
|
||||
a O
|
||||
elizabeth B-ACTOR
|
||||
perkins I-ACTOR
|
||||
period B-PLOT
|
||||
piece I-PLOT
|
||||
movie O
|
||||
|
||||
looking O
|
||||
for O
|
||||
the O
|
||||
movie O
|
||||
with O
|
||||
broderick B-ACTOR
|
||||
crawford I-ACTOR
|
||||
as O
|
||||
the O
|
||||
governor B-CHARACTER
|
||||
of I-CHARACTER
|
||||
louisiana I-CHARACTER
|
||||
|
||||
find O
|
||||
me O
|
||||
a O
|
||||
movie O
|
||||
with O
|
||||
the O
|
||||
song O
|
||||
my B-SONG
|
||||
heart I-SONG
|
||||
will I-SONG
|
||||
go I-SONG
|
||||
on I-SONG
|
||||
|
||||
list O
|
||||
a O
|
||||
crime B-GENRE
|
||||
film O
|
||||
which O
|
||||
director O
|
||||
was O
|
||||
david B-DIRECTOR
|
||||
lean I-DIRECTOR
|
||||
|
||||
when O
|
||||
did O
|
||||
little B-TITLE
|
||||
man I-TITLE
|
||||
come O
|
||||
out O
|
||||
on O
|
||||
dvd O
|
||||
|
||||
what O
|
||||
are O
|
||||
the O
|
||||
best B-REVIEW
|
||||
horror B-GENRE
|
||||
movies O
|
||||
from O
|
||||
2004 B-YEAR
|
||||
|
||||
how O
|
||||
many O
|
||||
james B-CHARACTER
|
||||
bond I-CHARACTER
|
||||
movies O
|
||||
are O
|
||||
there O
|
||||
|
||||
do O
|
||||
you O
|
||||
think O
|
||||
project B-TITLE
|
||||
x I-TITLE
|
||||
is O
|
||||
better B-REVIEW
|
||||
than O
|
||||
hangover B-TITLE
|
||||
|
||||
who O
|
||||
directed O
|
||||
cello B-TITLE
|
||||
eng I-TITLE
|
||||
subs I-TITLE
|
||||
|
||||
run O
|
||||
twilight B-TITLE
|
||||
zone I-TITLE
|
||||
|
||||
name O
|
||||
the O
|
||||
thriller B-GENRE
|
||||
starring O
|
||||
jack B-ACTOR
|
||||
haley I-ACTOR
|
||||
from O
|
||||
1960 B-YEAR
|
||||
that O
|
||||
is O
|
||||
rated O
|
||||
pg B-RATING
|
||||
13 I-RATING
|
||||
with O
|
||||
a O
|
||||
ratings O
|
||||
average O
|
||||
of O
|
||||
four B-RATINGS_AVERAGE
|
||||
|
||||
find O
|
||||
a O
|
||||
five B-RATINGS_AVERAGE
|
||||
star I-RATINGS_AVERAGE
|
||||
karate B-GENRE
|
||||
movie O
|
||||
|
||||
name O
|
||||
the O
|
||||
the O
|
||||
sequel O
|
||||
of O
|
||||
the O
|
||||
movie O
|
||||
pitch B-TITLE
|
||||
black I-TITLE
|
||||
starring O
|
||||
vin O
|
||||
diesel O
|
||||
|
||||
has O
|
||||
ingmar B-DIRECTOR
|
||||
bergman I-DIRECTOR
|
||||
ever O
|
||||
directed O
|
||||
any O
|
||||
r B-RATING
|
||||
rated O
|
||||
thriller B-GENRE
|
||||
films O
|
||||
|
||||
find O
|
||||
me O
|
||||
fantasy B-GENRE
|
||||
films O
|
||||
starring O
|
||||
warwick B-ACTOR
|
||||
davis I-ACTOR
|
||||
|
||||
find O
|
||||
a O
|
||||
kids B-GENRE
|
||||
movie I-GENRE
|
||||
with O
|
||||
benji B-CHARACTER
|
||||
the I-CHARACTER
|
||||
dog I-CHARACTER
|
||||
|
||||
was O
|
||||
a B-TITLE
|
||||
beautiful I-TITLE
|
||||
mind I-TITLE
|
||||
an O
|
||||
original B-SONG
|
||||
score I-SONG
|
||||
|
||||
i O
|
||||
wish O
|
||||
to O
|
||||
see O
|
||||
reviews B-REVIEW
|
||||
from O
|
||||
common O
|
||||
people O
|
||||
who O
|
||||
watched O
|
||||
the O
|
||||
movie O
|
||||
already O
|
||||
|
||||
how O
|
||||
many O
|
||||
james B-CHARACTER
|
||||
bond I-CHARACTER
|
||||
movies O
|
||||
did O
|
||||
sean B-ACTOR
|
||||
connery I-ACTOR
|
||||
star O
|
||||
in O
|
||||
|
||||
in O
|
||||
the O
|
||||
past B-YEAR
|
||||
nine I-YEAR
|
||||
years I-YEAR
|
||||
there O
|
||||
was O
|
||||
a O
|
||||
sci B-GENRE
|
||||
fi I-GENRE
|
||||
movie O
|
||||
that O
|
||||
was O
|
||||
directed O
|
||||
by O
|
||||
paddy B-DIRECTOR
|
||||
breathnach I-DIRECTOR
|
||||
and O
|
||||
received O
|
||||
nine B-RATINGS_AVERAGE
|
||||
stars I-RATINGS_AVERAGE
|
||||
also O
|
||||
included O
|
||||
parasites B-PLOT
|
||||
what O
|
||||
was O
|
||||
it O
|
||||
|
||||
suggest O
|
||||
a O
|
||||
movie O
|
||||
for O
|
||||
family B-REVIEW
|
||||
night I-REVIEW
|
||||
|
||||
run O
|
||||
a O
|
||||
trailer B-TRAILER
|
||||
for O
|
||||
a O
|
||||
stanley B-DIRECTOR
|
||||
kubrik I-DIRECTOR
|
||||
epic B-GENRE
|
||||
|
||||
what O
|
||||
are O
|
||||
people B-REVIEW
|
||||
saying I-REVIEW
|
||||
about O
|
||||
willy B-TITLE
|
||||
wonka I-TITLE
|
||||
the I-TITLE
|
||||
chocolate I-TITLE
|
||||
factory I-TITLE
|
||||
|
||||
find O
|
||||
a O
|
||||
trailer B-TRAILER
|
||||
for O
|
||||
last B-TITLE
|
||||
of I-TITLE
|
||||
the I-TITLE
|
||||
mohicans I-TITLE
|
||||
|
||||
show O
|
||||
me O
|
||||
a O
|
||||
movie O
|
||||
with O
|
||||
the O
|
||||
song O
|
||||
some B-SONG
|
||||
day I-SONG
|
||||
my I-SONG
|
||||
prince I-SONG
|
||||
will I-SONG
|
||||
come I-SONG
|
||||
|
||||
could O
|
||||
you O
|
||||
tell O
|
||||
me O
|
||||
the O
|
||||
titles O
|
||||
of O
|
||||
the O
|
||||
pg B-RATING
|
||||
13 I-RATING
|
||||
family B-GENRE
|
||||
films O
|
||||
from O
|
||||
the O
|
||||
past B-YEAR
|
||||
three I-YEAR
|
||||
decades I-YEAR
|
||||
by O
|
||||
director O
|
||||
david B-DIRECTOR
|
||||
presley I-DIRECTOR
|
||||
that O
|
||||
have O
|
||||
a O
|
||||
ratings O
|
||||
average O
|
||||
of O
|
||||
highly B-RATINGS_AVERAGE
|
||||
recommended I-RATINGS_AVERAGE
|
||||
|
||||
what O
|
||||
would O
|
||||
be O
|
||||
the O
|
||||
best O
|
||||
short B-GENRE
|
||||
movie O
|
||||
to O
|
||||
watch O
|
||||
starring O
|
||||
rene B-ACTOR
|
||||
russo I-ACTOR
|
||||
that O
|
||||
was O
|
||||
made O
|
||||
in O
|
||||
the O
|
||||
last B-YEAR
|
||||
six I-YEAR
|
||||
years I-YEAR
|
||||
|
||||
did O
|
||||
barbra B-ACTOR
|
||||
streisand I-ACTOR
|
||||
do O
|
||||
any O
|
||||
comedies B-GENRE
|
||||
|
||||
what O
|
||||
g B-RATING
|
||||
rated O
|
||||
drama B-GENRE
|
||||
with O
|
||||
humphrey B-ACTOR
|
||||
bogart I-ACTOR
|
||||
did O
|
||||
ok B-RATINGS_AVERAGE
|
||||
|
||||
find O
|
||||
me O
|
||||
the O
|
||||
movie O
|
||||
with O
|
||||
the O
|
||||
song O
|
||||
blue B-SONG
|
||||
hawaii I-SONG
|
||||
|
||||
what O
|
||||
are O
|
||||
some O
|
||||
good B-REVIEW
|
||||
g B-RATING
|
||||
rated O
|
||||
movies O
|
||||
for O
|
||||
kids B-REVIEW
|
||||
|
||||
im O
|
||||
looking O
|
||||
for O
|
||||
im B-TITLE
|
||||
not I-TITLE
|
||||
there I-TITLE
|
||||
|
||||
what O
|
||||
horror B-GENRE
|
||||
movie O
|
||||
features O
|
||||
a O
|
||||
character B-PLOT
|
||||
coming I-PLOT
|
||||
out I-PLOT
|
||||
of I-PLOT
|
||||
a I-PLOT
|
||||
tv I-PLOT
|
||||
screen I-PLOT
|
||||
|
||||
get O
|
||||
me O
|
||||
some O
|
||||
razzie B-REVIEW
|
||||
award I-REVIEW
|
||||
winners I-REVIEW
|
||||
from O
|
||||
2010 B-YEAR
|
||||
|
||||
find O
|
||||
a O
|
||||
preview B-TRAILER
|
||||
for O
|
||||
a O
|
||||
elizabeth B-ACTOR
|
||||
taylor I-ACTOR
|
||||
period B-GENRE
|
||||
piece I-GENRE
|
||||
movie O
|
||||
|
||||
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
|
|
@ -0,0 +1,456 @@
|
|||
-DOCSTART- O
|
||||
|
||||
what O
|
||||
restaurant O
|
||||
serves O
|
||||
tapas B-Dish
|
||||
on O
|
||||
forbes B-Location
|
||||
road I-Location
|
||||
that O
|
||||
has O
|
||||
celebrities B-Amenity
|
||||
visiting I-Amenity
|
||||
often I-Amenity
|
||||
|
||||
can O
|
||||
you O
|
||||
find O
|
||||
a O
|
||||
restaurant O
|
||||
near B-Location
|
||||
downtown I-Location
|
||||
that O
|
||||
serves O
|
||||
french B-Cuisine
|
||||
cuisine O
|
||||
|
||||
navigate O
|
||||
me O
|
||||
to O
|
||||
a O
|
||||
good B-Rating
|
||||
thai B-Cuisine
|
||||
restaurant O
|
||||
|
||||
is O
|
||||
there O
|
||||
a O
|
||||
cafe B-Restaurant_Name
|
||||
rio I-Restaurant_Name
|
||||
within B-Location
|
||||
ten I-Location
|
||||
miles I-Location
|
||||
that O
|
||||
is O
|
||||
not B-Price
|
||||
expensive I-Price
|
||||
|
||||
do O
|
||||
you O
|
||||
know O
|
||||
if O
|
||||
the O
|
||||
chinese B-Cuisine
|
||||
restaurant O
|
||||
on O
|
||||
king B-Location
|
||||
street I-Location
|
||||
have O
|
||||
outdoor B-Amenity
|
||||
parking I-Amenity
|
||||
|
||||
is O
|
||||
kostas B-Restaurant_Name
|
||||
pizza I-Restaurant_Name
|
||||
and I-Restaurant_Name
|
||||
seafood I-Restaurant_Name
|
||||
cheap B-Price
|
||||
|
||||
can O
|
||||
you O
|
||||
find O
|
||||
a O
|
||||
well B-Rating
|
||||
rated I-Rating
|
||||
japanese B-Cuisine
|
||||
restaurant O
|
||||
with O
|
||||
parking B-Amenity
|
||||
|
||||
where O
|
||||
can O
|
||||
i O
|
||||
get O
|
||||
a O
|
||||
martini B-Dish
|
||||
after B-Hours
|
||||
1 I-Hours
|
||||
am I-Hours
|
||||
in B-Location
|
||||
faneuil I-Location
|
||||
hall I-Location
|
||||
square I-Location
|
||||
|
||||
i O
|
||||
want O
|
||||
to O
|
||||
take O
|
||||
my O
|
||||
date O
|
||||
to O
|
||||
a O
|
||||
restaurant O
|
||||
that O
|
||||
has O
|
||||
good B-Rating
|
||||
mushrooms B-Cuisine
|
||||
|
||||
find O
|
||||
me O
|
||||
the O
|
||||
phone O
|
||||
number O
|
||||
of O
|
||||
the O
|
||||
closest B-Location
|
||||
restaurant O
|
||||
|
||||
what O
|
||||
mcdonalds B-Restaurant_Name
|
||||
are O
|
||||
there O
|
||||
near B-Location
|
||||
hitown I-Location
|
||||
city I-Location
|
||||
that O
|
||||
are O
|
||||
still B-Hours
|
||||
open I-Hours
|
||||
|
||||
is O
|
||||
ngon B-Restaurant_Name
|
||||
bistro I-Restaurant_Name
|
||||
open B-Hours
|
||||
right I-Hours
|
||||
now I-Hours
|
||||
|
||||
can O
|
||||
i O
|
||||
get O
|
||||
to O
|
||||
a O
|
||||
place O
|
||||
that O
|
||||
is O
|
||||
within B-Location
|
||||
five I-Location
|
||||
minutes I-Location
|
||||
and O
|
||||
open B-Hours
|
||||
before I-Hours
|
||||
noon I-Hours
|
||||
|
||||
does O
|
||||
the O
|
||||
german B-Cuisine
|
||||
restaurant O
|
||||
within B-Location
|
||||
5 I-Location
|
||||
miles I-Location
|
||||
of I-Location
|
||||
here I-Location
|
||||
offer O
|
||||
carry B-Amenity
|
||||
out I-Amenity
|
||||
|
||||
i O
|
||||
would O
|
||||
like O
|
||||
a O
|
||||
listing O
|
||||
of O
|
||||
olive B-Restaurant_Name
|
||||
gardens I-Restaurant_Name
|
||||
in B-Location
|
||||
the I-Location
|
||||
area I-Location
|
||||
|
||||
where O
|
||||
can O
|
||||
i O
|
||||
get O
|
||||
some O
|
||||
good B-Rating
|
||||
home B-Dish
|
||||
made I-Dish
|
||||
soup I-Dish
|
||||
|
||||
is O
|
||||
there O
|
||||
place O
|
||||
five B-Location
|
||||
minutes I-Location
|
||||
from I-Location
|
||||
here I-Location
|
||||
with O
|
||||
a O
|
||||
good B-Price
|
||||
price O
|
||||
with O
|
||||
vegetable B-Dish
|
||||
dishes O
|
||||
|
||||
find O
|
||||
a O
|
||||
mcdonalds B-Restaurant_Name
|
||||
near B-Location
|
||||
here I-Location
|
||||
|
||||
where O
|
||||
can O
|
||||
i O
|
||||
get O
|
||||
carry B-Amenity
|
||||
out I-Amenity
|
||||
chinese B-Cuisine
|
||||
food O
|
||||
in O
|
||||
glen B-Location
|
||||
cove I-Location
|
||||
new I-Location
|
||||
york I-Location
|
||||
|
||||
does O
|
||||
the O
|
||||
nearest B-Location
|
||||
steakhouse B-Cuisine
|
||||
take O
|
||||
diners B-Amenity
|
||||
club I-Amenity
|
||||
|
||||
locate O
|
||||
any O
|
||||
fast B-Cuisine
|
||||
food I-Cuisine
|
||||
restaurant O
|
||||
other B-Amenity
|
||||
than I-Amenity
|
||||
kfc B-Restaurant_Name
|
||||
that O
|
||||
is O
|
||||
within B-Location
|
||||
3 I-Location
|
||||
miles I-Location
|
||||
|
||||
im O
|
||||
feeling O
|
||||
mexican B-Cuisine
|
||||
food O
|
||||
|
||||
i O
|
||||
need O
|
||||
a O
|
||||
cheap B-Price
|
||||
place O
|
||||
with O
|
||||
a O
|
||||
strong B-Rating
|
||||
rating I-Rating
|
||||
|
||||
is O
|
||||
there O
|
||||
an O
|
||||
italian B-Cuisine
|
||||
restaurant O
|
||||
along B-Location
|
||||
my I-Location
|
||||
route I-Location
|
||||
that O
|
||||
has O
|
||||
a O
|
||||
lot O
|
||||
of O
|
||||
parking B-Amenity
|
||||
space I-Amenity
|
||||
|
||||
whats O
|
||||
the O
|
||||
best B-Rating
|
||||
place O
|
||||
to O
|
||||
eat O
|
||||
in O
|
||||
a O
|
||||
downtown B-Location
|
||||
hotel I-Location
|
||||
|
||||
can O
|
||||
you O
|
||||
find O
|
||||
me O
|
||||
a O
|
||||
restaurant O
|
||||
that O
|
||||
serves O
|
||||
vegetarian B-Dish
|
||||
meatloaf I-Dish
|
||||
|
||||
is O
|
||||
the O
|
||||
jamaica B-Restaurant_Name
|
||||
plain I-Restaurant_Name
|
||||
house I-Restaurant_Name
|
||||
pizza I-Restaurant_Name
|
||||
restaurant O
|
||||
cheap O
|
||||
|
||||
i O
|
||||
need O
|
||||
to O
|
||||
make O
|
||||
reservations B-Amenity
|
||||
for O
|
||||
three O
|
||||
at O
|
||||
alfredos B-Restaurant_Name
|
||||
for O
|
||||
tomorrow B-Hours
|
||||
night I-Hours
|
||||
at I-Hours
|
||||
6 I-Hours
|
||||
pm I-Hours
|
||||
|
||||
im O
|
||||
wanting O
|
||||
to O
|
||||
drink O
|
||||
a O
|
||||
lot O
|
||||
of O
|
||||
wine B-Dish
|
||||
who O
|
||||
gives O
|
||||
you O
|
||||
a O
|
||||
lot B-Price
|
||||
of I-Price
|
||||
wine I-Price
|
||||
for I-Price
|
||||
the I-Price
|
||||
money I-Price
|
||||
thats O
|
||||
within B-Location
|
||||
5 I-Location
|
||||
miles I-Location
|
||||
|
||||
i O
|
||||
want O
|
||||
to O
|
||||
find O
|
||||
the O
|
||||
closest B-Location
|
||||
olive B-Restaurant_Name
|
||||
garden I-Restaurant_Name
|
||||
|
||||
where O
|
||||
are O
|
||||
the O
|
||||
very O
|
||||
highly B-Rating
|
||||
rated I-Rating
|
||||
and O
|
||||
reasonably B-Price
|
||||
priced O
|
||||
restaurants O
|
||||
|
||||
id O
|
||||
like O
|
||||
to O
|
||||
try O
|
||||
a O
|
||||
four B-Rating
|
||||
star I-Rating
|
||||
restaurant O
|
||||
that O
|
||||
has O
|
||||
a O
|
||||
lunch B-Hours
|
||||
menu O
|
||||
|
||||
find O
|
||||
me O
|
||||
someplace O
|
||||
with O
|
||||
awesome B-Rating
|
||||
barbecue B-Dish
|
||||
dishes O
|
||||
in O
|
||||
holbrook B-Location
|
||||
|
||||
where O
|
||||
can O
|
||||
i O
|
||||
get O
|
||||
the O
|
||||
best B-Rating
|
||||
cake B-Dish
|
||||
|
||||
im O
|
||||
craving O
|
||||
some O
|
||||
cheap B-Price
|
||||
sausages B-Dish
|
||||
who O
|
||||
has O
|
||||
that O
|
||||
and O
|
||||
opens B-Hours
|
||||
before I-Hours
|
||||
noon I-Hours
|
||||
|
||||
im O
|
||||
looking O
|
||||
for O
|
||||
a O
|
||||
not B-Price
|
||||
too I-Price
|
||||
espensive I-Price
|
||||
restaurant O
|
||||
that O
|
||||
offers O
|
||||
halal B-Cuisine
|
||||
food O
|
||||
|
||||
what O
|
||||
is O
|
||||
the O
|
||||
rating B-Rating
|
||||
on O
|
||||
subway B-Restaurant_Name
|
||||
|
||||
is O
|
||||
there O
|
||||
a O
|
||||
starbucks B-Restaurant_Name
|
||||
nearby B-Location
|
||||
|
||||
is O
|
||||
there O
|
||||
a O
|
||||
steak B-Cuisine
|
||||
house I-Cuisine
|
||||
still B-Hours
|
||||
open I-Hours
|
||||
in B-Location
|
||||
this I-Location
|
||||
city I-Location
|
||||
|
||||
wheres O
|
||||
the O
|
||||
closest B-Location
|
||||
restaurant O
|
||||
with O
|
||||
free B-Amenity
|
||||
wifi I-Amenity
|
||||
|
||||
File diff suppressed because it is too large
Load Diff
|
|
@ -0,0 +1,909 @@
|
|||
-DOCSTART- O
|
||||
|
||||
how O
|
||||
much O
|
||||
does O
|
||||
it O
|
||||
cost O
|
||||
to O
|
||||
eat O
|
||||
at O
|
||||
mickeys B-Restaurant_Name
|
||||
diner I-Restaurant_Name
|
||||
per O
|
||||
person O
|
||||
|
||||
is O
|
||||
there O
|
||||
a O
|
||||
good B-Rating
|
||||
cafe B-Cuisine
|
||||
near O
|
||||
dolores B-Location
|
||||
park I-Location
|
||||
|
||||
are O
|
||||
there O
|
||||
any O
|
||||
cafes B-Cuisine
|
||||
open B-Hours
|
||||
past I-Hours
|
||||
11 I-Hours
|
||||
pm I-Hours
|
||||
which O
|
||||
have O
|
||||
internet B-Amenity
|
||||
access I-Amenity
|
||||
|
||||
where O
|
||||
is O
|
||||
a O
|
||||
restaurant O
|
||||
that O
|
||||
has O
|
||||
a O
|
||||
smoking B-Amenity
|
||||
area I-Amenity
|
||||
|
||||
what O
|
||||
is O
|
||||
the O
|
||||
top B-Rating
|
||||
rated I-Rating
|
||||
restaurant O
|
||||
in O
|
||||
helena B-Location
|
||||
montana I-Location
|
||||
|
||||
what O
|
||||
restaurant O
|
||||
serves O
|
||||
the O
|
||||
largest B-Dish
|
||||
pizza I-Dish
|
||||
in B-Location
|
||||
town I-Location
|
||||
|
||||
a O
|
||||
sub B-Dish
|
||||
place O
|
||||
near B-Location
|
||||
independence I-Location
|
||||
boulevard I-Location
|
||||
|
||||
i O
|
||||
want O
|
||||
a O
|
||||
cold O
|
||||
draft B-Dish
|
||||
beer I-Dish
|
||||
in O
|
||||
the O
|
||||
shade B-Amenity
|
||||
|
||||
find O
|
||||
me O
|
||||
a O
|
||||
japanese B-Cuisine
|
||||
restaurant O
|
||||
serving O
|
||||
japanese B-Cuisine
|
||||
food O
|
||||
|
||||
who O
|
||||
has O
|
||||
some O
|
||||
good B-Rating
|
||||
chicken B-Dish
|
||||
soup I-Dish
|
||||
|
||||
im O
|
||||
looking O
|
||||
for O
|
||||
a O
|
||||
restaurant O
|
||||
with O
|
||||
a O
|
||||
romantic B-Amenity
|
||||
atmosphere I-Amenity
|
||||
but O
|
||||
isnt B-Price
|
||||
too I-Price
|
||||
expensive I-Price
|
||||
|
||||
name O
|
||||
a O
|
||||
price O
|
||||
place O
|
||||
with O
|
||||
hungarian B-Cuisine
|
||||
cuisine O
|
||||
|
||||
lets O
|
||||
find O
|
||||
a O
|
||||
coffee B-Cuisine
|
||||
place O
|
||||
within B-Location
|
||||
5 I-Location
|
||||
miles I-Location
|
||||
of I-Location
|
||||
here I-Location
|
||||
|
||||
where O
|
||||
can O
|
||||
i O
|
||||
get O
|
||||
not B-Price
|
||||
so I-Price
|
||||
cheap I-Price
|
||||
chocolate B-Dish
|
||||
cake I-Dish
|
||||
and O
|
||||
be O
|
||||
able O
|
||||
to O
|
||||
byob B-Amenity
|
||||
|
||||
is O
|
||||
there O
|
||||
a O
|
||||
burger B-Restaurant_Name
|
||||
king I-Restaurant_Name
|
||||
near B-Location
|
||||
my I-Location
|
||||
current I-Location
|
||||
location I-Location
|
||||
|
||||
find O
|
||||
a O
|
||||
big B-Dish
|
||||
mac I-Dish
|
||||
from O
|
||||
mcdonalds B-Restaurant_Name
|
||||
around B-Location
|
||||
me I-Location
|
||||
|
||||
take O
|
||||
me O
|
||||
somewhere O
|
||||
nice B-Amenity
|
||||
to O
|
||||
eat O
|
||||
|
||||
can O
|
||||
you O
|
||||
get O
|
||||
me O
|
||||
directions O
|
||||
to O
|
||||
the O
|
||||
closest B-Location
|
||||
seafood B-Cuisine
|
||||
restaurant O
|
||||
|
||||
what O
|
||||
fine B-Amenity
|
||||
dining I-Amenity
|
||||
restaurants O
|
||||
are O
|
||||
open O
|
||||
for O
|
||||
dinner B-Hours
|
||||
and O
|
||||
feature O
|
||||
lunch B-Amenity
|
||||
specials I-Amenity
|
||||
|
||||
is O
|
||||
there O
|
||||
any O
|
||||
restaurant O
|
||||
that O
|
||||
has O
|
||||
good B-Rating
|
||||
reviews I-Rating
|
||||
but O
|
||||
is O
|
||||
also O
|
||||
vegan B-Amenity
|
||||
|
||||
i O
|
||||
could O
|
||||
go O
|
||||
for O
|
||||
some O
|
||||
mexican B-Cuisine
|
||||
food O
|
||||
know O
|
||||
where O
|
||||
i O
|
||||
can O
|
||||
get O
|
||||
some O
|
||||
|
||||
can O
|
||||
you O
|
||||
show O
|
||||
me O
|
||||
where O
|
||||
to O
|
||||
go O
|
||||
for O
|
||||
spit B-Dish
|
||||
buffalo I-Dish
|
||||
wings I-Dish
|
||||
and O
|
||||
handcrafted B-Dish
|
||||
beer I-Dish
|
||||
near B-Location
|
||||
my O
|
||||
home B-Location
|
||||
|
||||
im O
|
||||
hungry O
|
||||
wheres O
|
||||
the O
|
||||
nearest B-Location
|
||||
restaurant O
|
||||
|
||||
is O
|
||||
there O
|
||||
a O
|
||||
puleos B-Restaurant_Name
|
||||
dairy I-Restaurant_Name
|
||||
in B-Location
|
||||
north I-Location
|
||||
square I-Location
|
||||
that O
|
||||
is O
|
||||
a O
|
||||
place O
|
||||
the O
|
||||
local B-Rating
|
||||
people I-Rating
|
||||
like I-Rating
|
||||
|
||||
search O
|
||||
for O
|
||||
a O
|
||||
moderately B-Price
|
||||
priced O
|
||||
italian B-Cuisine
|
||||
restaurant O
|
||||
near B-Location
|
||||
my I-Location
|
||||
current I-Location
|
||||
location I-Location
|
||||
|
||||
nearest B-Location
|
||||
red B-Restaurant_Name
|
||||
lobster I-Restaurant_Name
|
||||
in B-Location
|
||||
town I-Location
|
||||
|
||||
where O
|
||||
can O
|
||||
i O
|
||||
find O
|
||||
cheap B-Price
|
||||
chinese B-Cuisine
|
||||
food O
|
||||
|
||||
are O
|
||||
there O
|
||||
any O
|
||||
diners O
|
||||
open B-Hours
|
||||
before I-Hours
|
||||
5 I-Hours
|
||||
am I-Hours
|
||||
that O
|
||||
allow B-Amenity
|
||||
smoking I-Amenity
|
||||
|
||||
what O
|
||||
is O
|
||||
the O
|
||||
most O
|
||||
inexpensive B-Price
|
||||
restaurant O
|
||||
|
||||
hi O
|
||||
i O
|
||||
am O
|
||||
looking O
|
||||
for O
|
||||
a O
|
||||
great B-Rating
|
||||
mexican B-Cuisine
|
||||
resturant O
|
||||
with O
|
||||
really O
|
||||
fast B-Amenity
|
||||
service I-Amenity
|
||||
any O
|
||||
suggestions O
|
||||
|
||||
looking O
|
||||
for O
|
||||
cheap B-Price
|
||||
eats O
|
||||
around B-Location
|
||||
here I-Location
|
||||
|
||||
is O
|
||||
there O
|
||||
a O
|
||||
kid B-Amenity
|
||||
friendly I-Amenity
|
||||
hot B-Dish
|
||||
wings I-Dish
|
||||
place O
|
||||
in O
|
||||
dixie B-Location
|
||||
inn I-Location
|
||||
|
||||
get O
|
||||
me O
|
||||
to O
|
||||
the O
|
||||
nearest B-Location
|
||||
car O
|
||||
repair O
|
||||
shop O
|
||||
in O
|
||||
town O
|
||||
|
||||
take O
|
||||
me O
|
||||
to O
|
||||
the O
|
||||
cheapest B-Price
|
||||
pizza B-Cuisine
|
||||
place O
|
||||
that O
|
||||
has O
|
||||
anchovies B-Dish
|
||||
|
||||
can O
|
||||
i O
|
||||
get O
|
||||
the O
|
||||
phone O
|
||||
number O
|
||||
for O
|
||||
the O
|
||||
nearest B-Location
|
||||
five B-Rating
|
||||
star I-Rating
|
||||
restaurant O
|
||||
|
||||
can O
|
||||
i O
|
||||
get O
|
||||
the O
|
||||
phone O
|
||||
number O
|
||||
of O
|
||||
a O
|
||||
restaurant O
|
||||
for O
|
||||
carry B-Amenity
|
||||
out I-Amenity
|
||||
order O
|
||||
of O
|
||||
spicy B-Dish
|
||||
buffalo I-Dish
|
||||
wings I-Dish
|
||||
|
||||
i O
|
||||
would O
|
||||
like O
|
||||
a O
|
||||
dinner B-Cuisine
|
||||
in B-Location
|
||||
my I-Location
|
||||
area I-Location
|
||||
please I-Location
|
||||
|
||||
is O
|
||||
there O
|
||||
a O
|
||||
restaurant O
|
||||
near B-Location
|
||||
here I-Location
|
||||
that O
|
||||
offers O
|
||||
carry B-Amenity
|
||||
out I-Amenity
|
||||
service I-Amenity
|
||||
|
||||
what O
|
||||
is O
|
||||
the O
|
||||
name O
|
||||
of O
|
||||
the O
|
||||
upscale B-Amenity
|
||||
restaurant O
|
||||
inside O
|
||||
the O
|
||||
marriott B-Location
|
||||
hotel I-Location
|
||||
of I-Location
|
||||
the I-Location
|
||||
shoals I-Location
|
||||
|
||||
find O
|
||||
out O
|
||||
where O
|
||||
the O
|
||||
closest B-Location
|
||||
panda B-Restaurant_Name
|
||||
chinese I-Restaurant_Name
|
||||
restaurant O
|
||||
is O
|
||||
|
||||
where O
|
||||
can O
|
||||
i O
|
||||
find O
|
||||
garlic B-Dish
|
||||
eggplant I-Dish
|
||||
|
||||
i O
|
||||
want O
|
||||
to O
|
||||
find O
|
||||
a O
|
||||
black B-Amenity
|
||||
tie I-Amenity
|
||||
restaurant O
|
||||
that O
|
||||
would O
|
||||
be O
|
||||
good B-Rating
|
||||
to O
|
||||
take O
|
||||
someone B-Amenity
|
||||
on I-Amenity
|
||||
a I-Amenity
|
||||
date I-Amenity
|
||||
|
||||
id O
|
||||
like O
|
||||
to O
|
||||
find O
|
||||
a O
|
||||
decent B-Rating
|
||||
italian B-Cuisine
|
||||
restaurant O
|
||||
near B-Location
|
||||
by I-Location
|
||||
|
||||
what O
|
||||
time O
|
||||
does O
|
||||
applebees B-Restaurant_Name
|
||||
in O
|
||||
san B-Location
|
||||
leandro I-Location
|
||||
open B-Hours
|
||||
|
||||
do O
|
||||
you O
|
||||
know O
|
||||
the O
|
||||
health B-Rating
|
||||
ratings I-Rating
|
||||
of O
|
||||
restaurants O
|
||||
|
||||
whats O
|
||||
an O
|
||||
average B-Price
|
||||
priced O
|
||||
restaurant O
|
||||
thats O
|
||||
open B-Hours
|
||||
at I-Hours
|
||||
11 I-Hours
|
||||
am I-Hours
|
||||
|
||||
i O
|
||||
want O
|
||||
to O
|
||||
find O
|
||||
a O
|
||||
goos B-Restaurant_Name
|
||||
delivery B-Amenity
|
||||
pizza B-Cuisine
|
||||
place O
|
||||
|
||||
where O
|
||||
can O
|
||||
i O
|
||||
find O
|
||||
a O
|
||||
family B-Amenity
|
||||
style I-Amenity
|
||||
place O
|
||||
price B-Price
|
||||
doesnt I-Price
|
||||
matter I-Price
|
||||
|
||||
is O
|
||||
there O
|
||||
a O
|
||||
colombian B-Cuisine
|
||||
buffet B-Amenity
|
||||
close B-Location
|
||||
by I-Location
|
||||
|
||||
where O
|
||||
is O
|
||||
the O
|
||||
nearest O
|
||||
olive B-Restaurant_Name
|
||||
garden I-Restaurant_Name
|
||||
restaurant O
|
||||
to O
|
||||
zip B-Location
|
||||
code I-Location
|
||||
04032 I-Location
|
||||
|
||||
i O
|
||||
would O
|
||||
like O
|
||||
the O
|
||||
closest B-Location
|
||||
japanese B-Cuisine
|
||||
restaurant O
|
||||
with O
|
||||
free B-Amenity
|
||||
parking I-Amenity
|
||||
|
||||
find O
|
||||
me O
|
||||
a O
|
||||
place O
|
||||
where O
|
||||
i O
|
||||
can O
|
||||
just O
|
||||
get O
|
||||
a O
|
||||
light B-Dish
|
||||
lunch I-Dish
|
||||
that O
|
||||
wouldnt O
|
||||
be O
|
||||
bad O
|
||||
for B-Amenity
|
||||
someone I-Amenity
|
||||
on I-Amenity
|
||||
a I-Amenity
|
||||
diet I-Amenity
|
||||
|
||||
is O
|
||||
taco B-Restaurant_Name
|
||||
bell I-Restaurant_Name
|
||||
still B-Hours
|
||||
open I-Hours
|
||||
|
||||
find O
|
||||
me O
|
||||
a O
|
||||
cheap B-Price
|
||||
sushi B-Cuisine
|
||||
restaurant O
|
||||
|
||||
where O
|
||||
is O
|
||||
a O
|
||||
cheap B-Price
|
||||
belgian B-Cuisine
|
||||
restaurant O
|
||||
on O
|
||||
the O
|
||||
north B-Location
|
||||
end I-Location
|
||||
|
||||
does O
|
||||
the O
|
||||
new O
|
||||
mexican B-Cuisine
|
||||
restaurant O
|
||||
have O
|
||||
good B-Rating
|
||||
reviews I-Rating
|
||||
|
||||
what O
|
||||
are O
|
||||
the O
|
||||
hours O
|
||||
of O
|
||||
the O
|
||||
nearest B-Location
|
||||
shoneys B-Restaurant_Name
|
||||
breakfast I-Restaurant_Name
|
||||
bar I-Restaurant_Name
|
||||
today O
|
||||
|
||||
is O
|
||||
there O
|
||||
a O
|
||||
restaurant O
|
||||
within B-Location
|
||||
25 I-Location
|
||||
miles I-Location
|
||||
that O
|
||||
serves O
|
||||
pizza B-Dish
|
||||
with I-Dish
|
||||
gluten B-Amenity
|
||||
free I-Amenity
|
||||
crust B-Dish
|
||||
|
||||
im O
|
||||
looking O
|
||||
for O
|
||||
a O
|
||||
cheap B-Price
|
||||
restaurant O
|
||||
|
||||
are O
|
||||
than O
|
||||
any O
|
||||
sushi B-Cuisine
|
||||
restaurants O
|
||||
on O
|
||||
the O
|
||||
north B-Location
|
||||
side I-Location
|
||||
of I-Location
|
||||
town I-Location
|
||||
and O
|
||||
i O
|
||||
need O
|
||||
to O
|
||||
know O
|
||||
their O
|
||||
phone O
|
||||
number O
|
||||
and O
|
||||
directions O
|
||||
|
||||
make O
|
||||
me O
|
||||
a O
|
||||
reservation O
|
||||
for O
|
||||
4 O
|
||||
people O
|
||||
tonight B-Hours
|
||||
at O
|
||||
the O
|
||||
best B-Rating
|
||||
sushi B-Cuisine
|
||||
restaurant O
|
||||
in B-Location
|
||||
this I-Location
|
||||
city I-Location
|
||||
|
||||
please O
|
||||
make O
|
||||
me O
|
||||
a O
|
||||
reservation O
|
||||
at O
|
||||
the O
|
||||
best B-Rating
|
||||
rated I-Rating
|
||||
restaurant O
|
||||
with O
|
||||
modern B-Cuisine
|
||||
american I-Cuisine
|
||||
cuisine O
|
||||
thank O
|
||||
you O
|
||||
|
||||
show O
|
||||
me O
|
||||
restaurants O
|
||||
that B-Hours
|
||||
are I-Hours
|
||||
open I-Hours
|
||||
|
||||
does O
|
||||
michaels B-Restaurant_Name
|
||||
have O
|
||||
a O
|
||||
smoking B-Amenity
|
||||
area I-Amenity
|
||||
|
||||
where O
|
||||
can O
|
||||
i O
|
||||
find O
|
||||
the O
|
||||
best B-Rating
|
||||
desserts B-Cuisine
|
||||
made O
|
||||
with O
|
||||
real B-Amenity
|
||||
cane I-Amenity
|
||||
sugar I-Amenity
|
||||
|
||||
do O
|
||||
they O
|
||||
have O
|
||||
good B-Price
|
||||
waterfront B-Amenity
|
||||
dining I-Amenity
|
||||
|
||||
what O
|
||||
is O
|
||||
the O
|
||||
best B-Rating
|
||||
rated I-Rating
|
||||
restaurant O
|
||||
near B-Location
|
||||
the I-Location
|
||||
next I-Location
|
||||
exit I-Location
|
||||
for O
|
||||
bismark B-Location
|
||||
north I-Location
|
||||
dakota I-Location
|
||||
|
||||
make O
|
||||
reservations O
|
||||
at O
|
||||
olive B-Restaurant_Name
|
||||
garden I-Restaurant_Name
|
||||
on O
|
||||
main B-Location
|
||||
street I-Location
|
||||
|
||||
wheres O
|
||||
the O
|
||||
closest B-Location
|
||||
deli B-Cuisine
|
||||
|
||||
i O
|
||||
am O
|
||||
looking O
|
||||
for O
|
||||
a O
|
||||
restaurant O
|
||||
that O
|
||||
features O
|
||||
wild B-Dish
|
||||
game I-Dish
|
||||
in B-Location
|
||||
the I-Location
|
||||
area I-Location
|
||||
|
||||
where O
|
||||
is O
|
||||
the O
|
||||
closet B-Location
|
||||
fast B-Cuisine
|
||||
food I-Cuisine
|
||||
restaurant O
|
||||
|
||||
can O
|
||||
you O
|
||||
look O
|
||||
for O
|
||||
an O
|
||||
italian B-Cuisine
|
||||
restaurant O
|
||||
in O
|
||||
this B-Location
|
||||
area I-Location
|
||||
|
||||
is O
|
||||
there O
|
||||
a O
|
||||
spuds B-Restaurant_Name
|
||||
with O
|
||||
lots B-Amenity
|
||||
of I-Amenity
|
||||
beer I-Amenity
|
||||
selections I-Amenity
|
||||
|
||||
does O
|
||||
nobles B-Restaurant_Name
|
||||
on O
|
||||
main B-Location
|
||||
have O
|
||||
valet B-Amenity
|
||||
parking I-Amenity
|
||||
|
||||
where O
|
||||
can O
|
||||
i O
|
||||
get O
|
||||
sushi B-Dish
|
||||
that O
|
||||
isnt B-Price
|
||||
real I-Price
|
||||
pricey O
|
||||
|
||||
is O
|
||||
there O
|
||||
an O
|
||||
upscale B-Restaurant_Name
|
||||
french B-Cuisine
|
||||
restaurant B-Restaurant_Name
|
||||
in O
|
||||
this O
|
||||
part O
|
||||
of O
|
||||
town B-Location
|
||||
that O
|
||||
doesnt O
|
||||
require O
|
||||
a O
|
||||
reservation O
|
||||
|
||||
do O
|
||||
you O
|
||||
show O
|
||||
any O
|
||||
mcdonalds B-Restaurant_Name
|
||||
nearby B-Location
|
||||
|
||||
are O
|
||||
there O
|
||||
any O
|
||||
italian B-Cuisine
|
||||
restaurants O
|
||||
open B-Hours
|
||||
before I-Hours
|
||||
7 I-Hours
|
||||
am I-Hours
|
||||
|
||||
i O
|
||||
want O
|
||||
the O
|
||||
phone O
|
||||
numbers O
|
||||
to O
|
||||
the O
|
||||
best B-Rating
|
||||
5 I-Rating
|
||||
star I-Rating
|
||||
restaurants O
|
||||
in O
|
||||
nyc B-Location
|
||||
|
||||
where O
|
||||
can O
|
||||
i O
|
||||
get O
|
||||
a O
|
||||
cheap B-Price
|
||||
meal O
|
||||
before B-Hours
|
||||
8 I-Hours
|
||||
am I-Hours
|
||||
|
||||
can O
|
||||
you O
|
||||
make O
|
||||
reservations O
|
||||
for O
|
||||
a O
|
||||
party O
|
||||
of O
|
||||
eight O
|
||||
at O
|
||||
the O
|
||||
north B-Restaurant_Name
|
||||
brook I-Restaurant_Name
|
||||
diner I-Restaurant_Name
|
||||
|
||||
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
|
|
@ -0,0 +1,99 @@
|
|||
import os
|
||||
os.environ["CUDA_DEVICE_ORDER"]="PCI_BUS_ID"
|
||||
os.environ["CUDA_VISIBLE_DEVICES"]='0'
|
||||
import logging
|
||||
import sys
|
||||
sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), "../")))
|
||||
|
||||
import hydra
|
||||
from hydra import utils
|
||||
from torch.utils.data import DataLoader
|
||||
from deepke.name_entity_re.few_shot.models.model import PromptBartModel, PromptGeneratorModel
|
||||
from deepke.name_entity_re.few_shot.module.datasets import ConllNERProcessor, ConllNERDataset
|
||||
from deepke.name_entity_re.few_shot.module.train import Trainer
|
||||
from deepke.name_entity_re.few_shot.utils.utils import set_seed
|
||||
from deepke.name_entity_re.few_shot.module.mapping_type import mit_movie_mapping, mit_restaurant_mapping, atis_mapping
|
||||
|
||||
import warnings
|
||||
warnings.filterwarnings("ignore", category=UserWarning)
|
||||
from tensorboardX import SummaryWriter
|
||||
writer = SummaryWriter(log_dir='logs')
|
||||
|
||||
logging.basicConfig(format = '%(asctime)s - %(levelname)s - %(name)s - %(message)s',
|
||||
datefmt = '%m/%d/%Y %H:%M:%S',
|
||||
level = logging.INFO)
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
|
||||
DATASET_CLASS = {
|
||||
'conll2003': ConllNERDataset,
|
||||
'mit-movie': ConllNERDataset,
|
||||
'mit-restaurant': ConllNERDataset,
|
||||
'atis': ConllNERDataset
|
||||
}
|
||||
|
||||
DATA_PROCESS = {
|
||||
'conll2003': ConllNERProcessor,
|
||||
'mit-movie': ConllNERProcessor,
|
||||
'mit-restaurant': ConllNERProcessor,
|
||||
'atis': ConllNERProcessor
|
||||
}
|
||||
|
||||
DATA_PATH = {
|
||||
'conll2003': {'train': 'data/conll2003/train.txt',
|
||||
'dev': 'data/conll2003/dev.txt',
|
||||
'test': 'data/conll2003/test.txt'},
|
||||
'mit-movie': {'train': 'data/mit-movie/20-shot-train.txt',
|
||||
'dev': 'data/mit-movie/test.txt'},
|
||||
'mit-restaurant': {'train': 'data/mit-restaurant/10-shot-train.txt',
|
||||
'dev': 'data/mit-restaurant/test.txt'},
|
||||
'atis': {'train': 'data/atis/20-shot-train.txt',
|
||||
'dev': 'data/atis/test.txt'}
|
||||
}
|
||||
|
||||
MAPPING = {
|
||||
'conll2003': {'loc': '<<location>>',
|
||||
'per': '<<person>>',
|
||||
'org': '<<organization>>',
|
||||
'misc': '<<others>>'},
|
||||
'mit-movie': mit_movie_mapping,
|
||||
'mit-restaurant': mit_restaurant_mapping,
|
||||
'atis': atis_mapping
|
||||
}
|
||||
|
||||
|
||||
@hydra.main(config_path="conf/config.yaml")
|
||||
def main(cfg):
|
||||
cwd = utils.get_original_cwd()
|
||||
cfg.cwd = cwd
|
||||
print(cfg)
|
||||
|
||||
data_path = DATA_PATH[cfg.dataset_name]
|
||||
for mode, path in data_path.items():
|
||||
data_path[mode] = os.path.join(cfg.cwd, path)
|
||||
dataset_class, data_process = DATASET_CLASS[cfg.dataset_name], DATA_PROCESS[cfg.dataset_name]
|
||||
mapping = MAPPING[cfg.dataset_name]
|
||||
|
||||
set_seed(cfg.seed) # set seed, default is 1
|
||||
if cfg.save_path is not None: # make save_path dir
|
||||
cfg.save_path = os.path.join(cfg.save_path, cfg.dataset_name+"_"+str(cfg.batch_size)+"_"+str(cfg.learning_rate)+cfg.notes)
|
||||
if not os.path.exists(cfg.save_path):
|
||||
os.makedirs(cfg.save_path, exist_ok=True)
|
||||
|
||||
process = data_process(data_path=data_path, mapping=mapping, bart_name=cfg.bart_name, learn_weights=cfg.learn_weights)
|
||||
test_dataset = dataset_class(data_processor=process, mode='test')
|
||||
test_dataloader = DataLoader(test_dataset, collate_fn=test_dataset.collate_fn, batch_size=cfg.batch_size, num_workers=4)
|
||||
|
||||
label_ids = list(process.mapping2id.values())
|
||||
prompt_model = PromptBartModel(tokenizer=process.tokenizer, label_ids=label_ids, args=cfg)
|
||||
model = PromptGeneratorModel(prompt_model=prompt_model, bos_token_id=0,
|
||||
eos_token_id=1,
|
||||
max_length=cfg.tgt_max_len, max_len_a=cfg.src_seq_ratio,num_beams=cfg.num_beams, do_sample=False,
|
||||
repetition_penalty=1, length_penalty=cfg.length_penalty, pad_token_id=1,
|
||||
restricter=None)
|
||||
trainer = Trainer(train_data=None, dev_data=None, test_data=test_dataloader, model=model, process=process, args=cfg, logger=logger, loss=None, metrics=None, writer=writer)
|
||||
trainer.predict()
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
|
|
@ -0,0 +1,3 @@
|
|||
transformers==3.4.0
|
||||
pytorch==1.7.0
|
||||
tensorboardX==2.4
|
||||
|
|
@ -0,0 +1,109 @@
|
|||
import os
|
||||
|
||||
import hydra
|
||||
os.environ["CUDA_DEVICE_ORDER"]="PCI_BUS_ID"
|
||||
os.environ["CUDA_VISIBLE_DEVICES"]='1'
|
||||
import logging
|
||||
import sys
|
||||
sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), "../")))
|
||||
|
||||
from hydra import utils
|
||||
from torch.utils.data import DataLoader
|
||||
from deepke.name_entity_re.few_shot.models.model import PromptBartModel, PromptGeneratorModel
|
||||
from deepke.name_entity_re.few_shot.module.datasets import ConllNERProcessor, ConllNERDataset
|
||||
from deepke.name_entity_re.few_shot.module.train import Trainer
|
||||
from deepke.name_entity_re.few_shot.module.metrics import Seq2SeqSpanMetric
|
||||
from deepke.name_entity_re.few_shot.utils.utils import get_loss, set_seed
|
||||
from deepke.name_entity_re.few_shot.module.mapping_type import mit_movie_mapping, mit_restaurant_mapping, atis_mapping
|
||||
|
||||
import warnings
|
||||
warnings.filterwarnings("ignore", category=UserWarning)
|
||||
from tensorboardX import SummaryWriter
|
||||
writer = SummaryWriter(log_dir='logs')
|
||||
|
||||
logging.basicConfig(format = '%(asctime)s - %(levelname)s - %(name)s - %(message)s',
|
||||
datefmt = '%m/%d/%Y %H:%M:%S',
|
||||
level = logging.INFO)
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
|
||||
DATASET_CLASS = {
|
||||
'conll2003': ConllNERDataset,
|
||||
'mit-movie': ConllNERDataset,
|
||||
'mit-restaurant': ConllNERDataset,
|
||||
'atis': ConllNERDataset
|
||||
}
|
||||
|
||||
DATA_PROCESS = {
|
||||
'conll2003': ConllNERProcessor,
|
||||
'mit-movie': ConllNERProcessor,
|
||||
'mit-restaurant': ConllNERProcessor,
|
||||
'atis': ConllNERProcessor
|
||||
}
|
||||
|
||||
DATA_PATH = {
|
||||
'conll2003': {'train': 'data/conll2003/train.txt',
|
||||
'dev': 'data/conll2003/dev.txt',
|
||||
'test': 'data/conll2003/test.txt'},
|
||||
'mit-movie': {'train': 'data/mit-movie/20-shot-train.txt',
|
||||
'dev': 'data/mit-movie/test.txt'},
|
||||
'mit-restaurant': {'train': 'data/mit-restaurant/10-shot-train.txt',
|
||||
'dev': 'data/mit-restaurant/test.txt'},
|
||||
'atis': {'train': 'data/atis/20-shot-train.txt',
|
||||
'dev': 'data/atis/test.txt'}
|
||||
}
|
||||
|
||||
MAPPING = {
|
||||
'conll2003': {'loc': '<<location>>',
|
||||
'per': '<<person>>',
|
||||
'org': '<<organization>>',
|
||||
'misc': '<<others>>'},
|
||||
'mit-movie': mit_movie_mapping,
|
||||
'mit-restaurant': mit_restaurant_mapping,
|
||||
'atis': atis_mapping
|
||||
}
|
||||
|
||||
@hydra.main(config_path="conf/config.yaml")
|
||||
def main(cfg):
|
||||
cwd = utils.get_original_cwd()
|
||||
cfg.cwd = cwd
|
||||
print(cfg)
|
||||
|
||||
data_path = DATA_PATH[cfg.dataset_name]
|
||||
for mode, path in data_path.items():
|
||||
data_path[mode] = os.path.join(cfg.cwd, path)
|
||||
dataset_class, data_process = DATASET_CLASS[cfg.dataset_name], DATA_PROCESS[cfg.dataset_name]
|
||||
mapping = MAPPING[cfg.dataset_name]
|
||||
|
||||
set_seed(cfg.seed) # set seed, default is 1
|
||||
if cfg.save_path is not None: # make save_path dir
|
||||
cfg.save_path = os.path.join(cfg.save_path, cfg.dataset_name+"_"+str(cfg.batch_size)+"_"+str(cfg.learning_rate)+cfg.notes)
|
||||
if not os.path.exists(cfg.save_path):
|
||||
os.makedirs(cfg.save_path, exist_ok=True)
|
||||
|
||||
process = data_process(data_path=data_path, mapping=mapping, bart_name=cfg.bart_name, learn_weights=cfg.learn_weights)
|
||||
train_dataset = dataset_class(data_processor=process, mode='train')
|
||||
train_dataloader = DataLoader(train_dataset, collate_fn=train_dataset.collate_fn, batch_size=cfg.batch_size, num_workers=4)
|
||||
|
||||
dev_dataset = dataset_class(data_processor=process, mode='dev')
|
||||
dev_dataloader = DataLoader(dev_dataset, collate_fn=dev_dataset.collate_fn, batch_size=cfg.batch_size, num_workers=4)
|
||||
|
||||
label_ids = list(process.mapping2id.values())
|
||||
|
||||
prompt_model = PromptBartModel(tokenizer=process.tokenizer, label_ids=label_ids, args=cfg)
|
||||
model = PromptGeneratorModel(prompt_model=prompt_model, bos_token_id=0,
|
||||
eos_token_id=1,
|
||||
max_length=cfg.tgt_max_len, max_len_a=cfg.src_seq_ratio,num_beams=cfg.num_beams, do_sample=False,
|
||||
repetition_penalty=1, length_penalty=cfg.length_penalty, pad_token_id=1,
|
||||
restricter=None)
|
||||
metrics = Seq2SeqSpanMetric(eos_token_id=1, num_labels=len(label_ids), target_type='word')
|
||||
loss = get_loss
|
||||
|
||||
trainer = Trainer(train_data=train_dataloader, dev_data=dev_dataloader, test_data=None, model=model, args=cfg, logger=logger, loss=loss, metrics=metrics, writer=writer)
|
||||
trainer.train()
|
||||
|
||||
writer.close()
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
Loading…
Reference in New Issue