I
nte
rna
t
io
na
l J
o
urna
l o
f
Adv
a
nces in Applie
d Science
s
(
I
J
AAS)
Vo
l.
1
3
,
No
.
4
,
Dec
em
b
er
2
0
2
4
,
p
p
.
8
3
4
~
8
43
I
SS
N:
2252
-
8
8
1
4
,
DOI
:
1
0
.
1
1
5
9
1
/ijaas
.
v
1
3
.
i
4
.
pp
834
-
8
43
834
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
a
s
.
ia
esco
r
e.
co
m
Ba
tak To
ba
lang
ua
g
e
-
Indo
nesia
n ma
chine t
ra
nsla
tio
n wit
h
trans
fer
l
ea
rning
using
no la
ng
ua
g
e lef
t
behin
d
Cev
in Sa
m
uel,
I
rsa
n T
a
ufik
Ali
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
i
c
a
l
En
g
i
n
e
e
r
i
n
g
,
F
a
c
u
l
t
y
o
f
En
g
i
n
e
e
r
i
n
g
,
U
n
i
v
e
r
s
i
t
a
s
R
i
a
u
,
P
e
k
a
n
b
a
r
u
,
I
n
d
o
n
e
si
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
No
v
1
6
,
2
0
2
3
R
ev
is
ed
J
u
n
3
,
2
0
2
4
Acc
ep
ted
J
u
n
1
9
,
2
0
2
4
Th
is
stu
d
y
fo
c
u
se
s
o
n
n
e
u
ra
l
m
a
c
h
in
e
tran
sla
ti
o
n
(NMT
)
f
o
r
l
o
w
-
re
so
u
rc
e
lan
g
u
a
g
e
s
(LRL
s)
p
a
ir,
Ba
tak
T
o
b
a
-
In
d
o
n
e
sia
n
(
b
b
c
↔in
d
).
T
h
e
B
a
tak
To
b
a
lan
g
u
a
g
e
is
a
c
rit
ica
ll
y
e
n
d
a
n
g
e
r
e
d
d
iale
c
t
o
f
a
n
In
d
o
n
e
sia
n
e
th
n
ic
g
ro
u
p
,
Ba
tak
.
Re
c
e
n
t
a
d
v
a
n
c
e
s
in
m
a
c
h
i
n
e
tran
sla
ti
o
n
o
ffe
r
p
o
ten
t
ial
so
lu
t
io
n
s,
wit
h
tran
sfe
r
lea
rn
in
g
e
m
e
rg
in
g
a
s
a
p
ro
m
isin
g
a
p
p
r
o
a
c
h
fo
r
t
h
is
lan
g
u
a
g
e
p
a
ir.
We
u
se
d
a
p
u
b
li
c
l
y
a
v
a
il
a
b
le
b
b
c
↔in
d
p
a
ra
ll
e
l
c
o
rp
o
ra
fro
m
th
e
Hu
g
g
i
n
g
F
a
c
e
d
a
tas
e
ts
h
u
b
a
n
d
e
m
p
lo
y
e
d
th
e
NLLB
-
2
0
0
'
s
d
isti
ll
e
d
6
0
0
M
v
a
rian
t
m
o
d
e
l
a
s
t
h
e
b
a
se
li
n
e
m
o
d
e
l.
Ou
r
m
o
d
e
ls
a
c
h
iev
e
d
sa
c
re
BLE
U
sc
o
re
s
a
s
fo
ll
o
ws
:
i)
f
o
r
b
b
c
→in
d
,
it
a
c
h
iev
e
d
a
sc
o
re
o
f
3
7
.
1
0
(+
2
5
.
6
7
,
u
p
fro
m
1
1
.
4
3
)
a
n
d
ii
)
f
o
r
i
n
d
→
b
b
c
,
it
a
c
h
iev
e
d
a
sc
o
re
o
f
3
0
.
8
4
(+
2
5
.
8
2
,
u
p
fro
m
5
.
0
2
).
Th
e
se
re
su
lt
s
o
u
t
p
e
rfo
rm
a
ll
p
re
v
i
o
u
s
wo
r
k
s
i
n
th
e
tas
k
b
b
c
↔in
d
m
a
c
h
in
e
tran
sla
ti
o
n
a
n
d
p
r
o
v
e
t
h
e
v
a
li
d
i
ty
o
f
o
u
r
a
p
p
r
o
a
c
h
.
K
ey
w
o
r
d
s
:
B
atak
T
o
b
a
lan
g
u
ag
e
L
o
w
-
r
eso
u
r
ce
la
n
g
u
a
g
es
Neu
r
al
m
ac
h
in
e
tr
a
n
s
latio
n
NL
L
B
-
2
0
0
m
o
d
el
s
ac
r
eBLE
U
s
co
r
e
T
r
an
s
f
er
lear
n
i
n
g
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
C
ev
in
Sam
u
el
Dep
ar
tm
en
t o
f
E
lectr
ical
E
n
g
i
n
ee
r
in
g
,
Facu
lty
o
f
E
n
g
in
ee
r
in
g
,
Un
iv
er
s
itas
R
iau
B
in
a
W
id
y
a
C
am
p
u
s
KM
.
1
2
.
5
,
Pek
an
b
a
r
u
2
8
2
9
3
,
I
n
d
o
n
esia
E
m
ail: c
ev
in
.
s
am
u
el3
7
7
8
@
s
tu
d
en
t.u
n
r
i.a
c.
id
1.
I
NT
RO
D
UCT
I
O
N
Am
o
n
g
th
e
d
iv
er
s
e
lin
g
u
is
tic
lan
d
s
ca
p
e
o
f
I
n
d
o
n
esia,
h
o
m
e
to
an
in
c
r
ed
ib
le
a
r
r
ay
o
f
lan
g
u
ag
es
an
d
d
ialec
ts
,
th
e
B
atak
T
o
b
a
la
n
g
u
ag
e,
s
p
o
k
e
n
b
y
m
illi
o
n
s
i
n
No
r
th
Su
m
atr
a,
is
in
tr
o
u
b
le.
C
lass
if
ied
as
en
d
an
g
e
r
ed
o
n
th
e
UNE
SC
O
lis
t
o
f
en
d
an
g
e
r
ed
lan
g
u
a
g
es
[
1
]
,
th
e
p
r
eser
v
atio
n
o
f
th
e
lan
g
u
a
g
e
is
o
f
p
ar
ticu
lar
im
p
o
r
tan
ce
in
th
e
f
ac
e
o
f
d
ec
lin
in
g
n
u
m
b
e
r
s
o
f
s
p
ea
k
er
s
an
d
r
eso
u
r
ce
s
.
An
ess
en
tial
s
tep
to
war
d
s
m
ain
tain
in
g
s
u
ch
a
la
n
g
u
a
g
e
is
o
p
en
in
g
d
o
o
r
s
f
o
r
co
m
m
u
n
icatio
n
th
r
o
u
g
h
th
e
m
o
d
er
n
ag
e,
allo
win
g
t
h
e
lan
g
u
ag
e
t
o
r
em
ain
r
elev
an
t a
n
d
ac
ce
s
s
ib
le
to
y
o
u
n
g
er
g
en
e
r
atio
n
s
in
th
e
f
ac
e
o
f
th
e
d
o
m
i
n
an
t la
n
g
u
ag
es
[
2
]
.
C
u
ltu
r
al
an
d
h
is
to
r
ical
s
ig
n
if
ican
ce
,
h
er
e
ar
e
s
o
m
e
s
ig
n
if
ican
t
f
ea
tu
r
es
o
f
its
im
p
o
r
tan
ce
:
i)
B
atak
T
o
b
a
is
an
ess
en
tial
co
m
p
o
n
e
n
t
o
f
th
e
B
atak
p
eo
p
le'
s
cu
ltu
r
al
id
en
tity
.
I
t
is
u
tili
ze
d
in
tr
ad
itio
n
al
ce
r
em
o
n
ies,
m
u
s
ic,
s
to
r
ies,
an
d
d
aily
life
,
r
ep
r
esen
tin
g
th
eir
d
is
tin
ct
c
u
ltu
r
al
leg
ac
y
an
d
h
is
to
r
y
;
ii)
B
atak
T
o
b
a
is
an
Au
s
tr
o
n
esian
lan
g
u
ag
e
,
co
n
tr
i
b
u
tin
g
to
I
n
d
o
n
esia'
s
an
d
th
e
wo
r
ld
'
s
lin
g
u
is
tic
v
ar
ie
ty
.
L
o
s
in
g
it
wo
u
ld
d
ev
alu
e
th
is
u
n
iq
u
e
lin
g
u
is
tic
h
er
itag
e;
an
d
iii)
B
atak
T
o
b
a
p
r
o
v
id
es
in
s
ig
h
ts
in
to
th
e
B
atak
p
eo
p
le'
s
h
is
to
r
ical
ev
o
lu
tio
n
a
n
d
c
o
n
tacts with
o
th
er
cu
ltu
r
es.
Po
ten
tial
im
p
ac
t
o
f
th
is
wo
r
k
,
th
er
e
ar
e
s
ev
er
al
way
s
in
wh
i
ch
th
is
r
esear
ch
m
ig
h
t
co
n
tr
ib
u
te
to
th
e
p
r
eser
v
atio
n
a
n
d
r
esto
r
atio
n
o
f
th
e
lan
g
u
ag
e.
T
h
e
m
o
d
el
ca
n
s
er
v
e
s
ev
er
al
p
u
r
p
o
s
es:
i)
p
r
o
m
o
tin
g
ac
ce
s
s
ib
ilit
y
is
b
y
f
ac
ilit
atin
g
co
m
m
u
n
icatio
n
b
etwe
en
B
atak
T
o
b
a
s
p
ea
k
er
s
a
n
d
I
n
d
o
n
esian
s
p
ea
k
er
s
,
it
c
an
ex
p
an
d
th
e
r
eso
u
r
ce
s
an
d
k
n
o
w
led
g
e
av
ailab
ilit
y
in
b
o
t
h
la
n
g
u
ag
es;
ii)
d
ev
elo
p
in
g
e
d
u
ca
tio
n
al
m
ater
ials
an
d
r
eso
u
r
ce
s
in
b
o
th
B
atak
T
o
b
a
an
d
I
n
d
o
n
esian
,
en
h
an
cin
g
ed
u
ca
tio
n
f
o
r
B
atak
T
o
b
a
co
m
m
u
n
ities
;
iii)
s
u
s
tai
n
th
e
cu
ltu
r
e
is
th
e
m
o
d
el
ca
n
b
e
in
teg
r
ated
in
to
o
th
er
d
i
g
ital
p
latf
o
r
m
s
an
d
ap
p
licatio
n
s
,
f
u
r
th
er
p
r
o
m
o
tin
g
B
atak
T
o
b
a;
an
d
,
iv
)
co
n
d
u
cti
n
g
ad
d
itio
n
al
r
esear
ch
,
t
h
is
s
tu
d
y
p
av
es
th
e
way
f
o
r
f
u
t
u
r
e
r
e
s
ea
r
ch
in
th
e
B
atak
T
o
b
a
lan
g
u
ag
e,
e
n
co
u
r
ag
in
g
s
cien
tis
ts
an
d
d
ev
elo
p
er
s
to
c
o
n
tr
ib
u
te
to
its
ad
v
an
ce
m
en
t a
n
d
p
r
eser
v
atio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ad
v
Ap
p
l Sci
I
SS
N:
2252
-
8
8
1
4
B
a
ta
k
To
b
a
la
n
g
u
a
g
e
-
I
n
d
o
n
es
ia
n
ma
ch
in
e
tr
a
n
s
la
tio
n
w
ith
tr
a
n
s
fer lea
r
n
in
g
u
s
in
g
n
o
…
(
C
ev
in
S
a
mu
el)
835
T
r
an
s
f
er
lear
n
i
n
g
,
a
s
tr
aig
h
tf
o
r
war
d
y
et
e
f
f
ec
tiv
e
a
p
p
r
o
ac
h
t
o
im
p
r
o
v
in
g
m
o
d
el
p
er
f
o
r
m
a
n
ce
o
f
lo
w
-
r
eso
u
r
ce
n
eu
r
al
m
ac
h
in
e
tr
a
n
s
latio
n
(
NM
T
)
,
o
f
f
er
s
a
p
r
o
m
is
in
g
ap
p
r
o
ac
h
f
o
r
B
atak
T
o
b
a
lan
g
u
ag
e
r
esto
r
atio
n
.
B
y
u
s
in
g
ex
is
tin
g
p
r
e
-
tr
ain
ed
m
ac
h
in
e
tr
an
s
latio
n
m
o
d
e
ls
lik
e
NL
L
B
-
2
0
0
,
we
ca
n
ac
ce
ler
ate
th
e
d
ev
elo
p
m
e
n
t
o
f
B
atak
T
o
b
a
t
r
an
s
latio
n
m
o
d
els,
ev
en
with
lim
ited
d
ata.
T
h
is
ap
p
r
o
ac
h
c
an
em
p
o
wer
B
atak
T
o
b
a
s
p
ea
k
er
s
to
ac
ce
s
s
in
f
o
r
m
atio
n
an
d
co
n
n
ec
t
with
b
r
o
ad
er
co
m
m
u
n
ities
,
u
ltima
tely
co
n
tr
ib
u
tin
g
to
th
e
lan
g
u
ag
e'
s
r
esto
r
atio
n
ef
f
o
r
ts
[
3
]
.
Fo
r
i
n
s
tan
ce
,
a
p
er
s
o
n
lear
n
in
g
B
atak
T
o
b
a
ca
n
u
tili
ze
th
is
m
o
d
el
to
f
in
d
tr
an
s
latio
n
s
th
ey
n
ee
d
,
m
ak
in
g
it a
p
r
ac
tical
to
o
l f
o
r
lan
g
u
ag
e
lear
n
in
g
.
T
h
is
p
ap
er
f
ills
th
is
g
ap
b
y
u
ti
lizin
g
tr
an
s
f
er
lear
n
in
g
,
a
tech
n
iq
u
e
th
at
is
p
er
f
ec
t
f
o
r
t
r
ain
i
n
g
m
o
d
els
with
lim
ited
d
ata.
W
e
in
ten
d
to
:
i)
cr
ea
te
a
B
atak
T
o
b
a
-
I
n
d
o
n
esian
(
b
b
c↔in
d
)
m
ac
h
i
n
e
tr
an
s
latio
n
m
o
d
el
u
s
in
g
tr
an
s
f
er
lear
n
in
g
,
d
em
o
n
s
tr
atin
g
its
f
ea
s
ib
ilit
y
f
o
r
lo
w
-
r
eso
u
r
ce
lan
g
u
ag
es
(
L
R
L
s
)
;
ii)
ev
alu
ate
th
e
m
o
d
el'
s
p
er
f
o
r
m
an
ce
to
ass
ess
it
s
p
o
ten
tial
f
o
r
r
ea
l
-
wo
r
l
d
ap
p
licatio
n
s
an
d
f
u
t
u
r
e
d
e
v
elo
p
m
en
t;
an
d
iii)
co
n
tr
ib
u
te
to
B
atak
T
o
b
a
p
r
eser
v
atio
n
b
y
p
r
o
v
id
in
g
a
t
o
o
l
f
o
r
cu
lt
u
r
al,
c
o
m
m
u
n
icatio
n
,
a
n
d
ed
u
ca
tio
n
p
r
o
m
o
ti
o
n
.
B
y
f
u
lf
illi
n
g
th
ese
g
o
als,
we
in
ten
d
to
p
a
v
e
th
e
p
ath
f
o
r
f
u
tu
r
e
s
tu
d
y
an
d
d
ev
elo
p
m
en
t
in
B
atak
T
o
b
a
m
ac
h
i
n
e
tr
an
s
latio
n
,
e
v
e
n
tu
ally
co
n
tr
i
b
u
tin
g
t
o
th
e
lan
g
u
ag
e'
s
s
u
r
v
iv
al
an
d
r
esil
ien
ce
in
th
e
d
ig
ital e
r
a
.
2.
RE
L
AT
E
D
WO
RK
2
.
1
.
L
o
w
-
re
s
o
urce
la
ng
ua
g
e
s
R
esear
ch
er
s
h
av
e
b
ee
n
ac
tiv
ely
wo
r
k
in
g
to
im
p
r
o
v
e
th
e
tr
an
s
latio
n
q
u
ality
f
o
r
L
R
L
s
,
aim
in
g
to
b
r
id
g
e
th
e
g
ap
with
h
ig
h
an
d
m
ed
i
u
m
-
r
eso
u
r
ce
la
n
g
u
a
g
e
s
.
On
e
p
o
s
s
ib
le
ap
p
r
o
ac
h
in
v
o
lv
es
tr
an
s
f
e
r
r
in
g
k
n
o
wled
g
e
f
r
o
m
a
h
ig
h
-
r
eso
u
r
ce
lan
g
u
a
g
e
to
th
e
L
R
L
s
v
ia
a
tech
n
iq
u
e
k
n
o
w
n
as
tr
an
s
f
er
lear
n
in
g
[
4
]
.
T
h
is
m
eth
o
d
u
s
es
cu
r
r
e
n
t
k
n
o
wled
g
e
to
in
cr
ea
s
e
tr
an
s
latio
n
ac
c
u
r
ac
y
in
L
R
L
s
.
An
o
th
er
im
p
o
r
tan
t
m
eth
o
d
is
d
ata
au
g
m
en
tatio
n
.
Un
lik
e
p
ar
allel
d
ata
(
wh
er
e
s
en
ten
ce
s
ex
is
t
in
b
o
th
lan
g
u
a
g
es),
L
R
L
s
ty
p
ically
h
av
e
m
u
ch
m
o
r
e
r
ea
d
ily
av
aila
b
le
m
o
n
o
l
in
g
u
al
d
ata
(
tex
t
in
o
n
ly
th
e
tar
g
et
lan
g
u
ag
e)
.
T
h
is
ad
v
an
t
ag
e
allo
ws
f
o
r
th
e
cr
ea
tio
n
o
f
s
y
n
th
etic
p
ar
allel
d
ata
p
air
s
,
wh
ich
h
a
v
e
b
ee
n
s
h
o
wn
to
en
h
a
n
ce
tr
an
s
latio
n
ac
c
u
r
ac
y
[
5
]
.
Fo
r
in
s
tan
ce
,
a
s
tu
d
y
f
o
cu
s
in
g
o
n
Fin
n
o
-
Ug
r
ic
lan
g
u
a
g
es
d
em
o
n
s
tr
ated
th
at
tr
ain
i
n
g
a
m
u
ltil
in
g
u
al
m
o
d
el
in
co
r
p
o
r
atin
g
o
th
er
la
n
g
u
ag
es
with
in
th
e
s
am
e
f
am
ily
led
to
s
ig
n
if
ican
t
im
p
r
o
v
em
en
ts
in
E
s
to
n
ian
tr
an
s
latio
n
q
u
ality
[
6
]
.
I
t
is
n
o
tab
ly
s
h
o
wn
in
[
7
]
th
at
th
e
clo
s
er
th
e
r
elatio
n
s
h
ip
b
etwe
en
t
h
e
h
ig
h
o
r
m
e
d
iu
m
-
r
eso
u
r
ce
l
an
g
u
ag
es
u
s
ed
in
tr
ain
in
g
an
d
th
e
L
R
L
s
,
th
e
g
r
ea
ter
th
e
tr
an
s
latio
n
ac
cu
r
ac
y
.
W
h
ile
tr
an
s
f
er
lear
n
in
g
h
as
p
r
o
v
en
v
alu
ab
le
f
o
r
L
R
L
s
with
lim
ited
p
ar
allel
r
eso
u
r
ce
s
[
8
]
,
th
e
s
tu
d
y
also
r
ev
ea
led
in
s
tan
ce
s
wh
er
e
u
s
in
g
an
u
n
r
elate
d
h
ig
h
-
r
eso
u
r
ce
lan
g
u
ag
e
r
esu
l
ted
in
ev
en
b
etter
tr
a
n
s
latio
n
s
.
2
.
2
.
NL
L
B
-
200
NL
L
B
-
2
0
0
is
a
lar
g
e
-
s
ca
le
m
u
ltil
in
g
u
al
m
ac
h
i
n
e
tr
a
n
s
latio
n
m
o
d
el
ca
p
a
b
le
o
f
tr
a
n
s
latin
g
b
etwe
en
2
0
2
lan
g
u
ag
es,
in
clu
d
in
g
m
a
n
y
L
R
L
s
as
id
en
tifie
d
in
[
9
]
.
Ho
wev
er
,
B
atak
T
o
b
a
is
o
n
e
o
f
th
e
lan
g
u
ag
es
cu
r
r
en
tly
n
o
t
s
u
p
p
o
r
te
d
.
T
h
e
m
o
d
el
was
tr
ain
e
d
o
n
1
8
b
illi
o
n
p
ar
allel
s
en
ten
ce
s
co
v
er
in
g
o
v
er
4
0
,
0
0
0
tr
an
s
latio
n
d
ir
ec
tio
n
s
.
T
h
e
tr
a
in
in
g
d
ata
was
cr
ea
ted
u
s
in
g
a
co
m
b
i
n
atio
n
o
f
h
u
m
a
n
-
tr
an
s
lated
b
en
ch
m
a
r
k
s
(
FLORES
-
2
0
0
)
an
d
a
n
ew
to
x
icity
b
en
ch
m
ar
k
en
co
m
p
ass
in
g
all
FLO
R
E
S
-
2
0
0
lan
g
u
ag
es,
en
s
u
r
in
g
tr
an
s
latio
n
q
u
ality
an
d
s
af
ety
.
As
d
escr
ib
ed
in
[
1
0
]
,
th
e
NL
L
B
team
em
p
lo
y
ed
b
itex
t
m
in
in
g
an
d
b
ac
k
-
tr
an
s
lati
o
n
tech
n
iq
u
es
to
co
n
s
tr
u
ct
th
e
n
o
n
-
E
n
g
lis
h
-
ce
n
tr
ic
tr
ain
in
g
d
ataset.
C
o
m
p
ar
ed
to
p
r
ev
io
u
s
s
tate
-
of
-
th
e
-
ar
t m
o
d
els lik
e
Dee
p
n
et
[
1
1
]
an
d
M2
M
-
1
0
0
[
1
0
]
,
NL
L
B
-
2
0
0
ac
h
iev
es a
4
4
% B
L
E
U
s
co
r
e
im
p
r
o
v
em
e
n
t.
As
r
ep
o
r
ted
in
[
9
]
,
b
y
h
av
i
n
g
ex
ten
s
iv
e
tr
ain
in
g
d
ata,
b
r
o
ad
lan
g
u
ag
e
s
u
p
p
o
r
t,
an
d
p
r
o
m
is
in
g
ev
alu
atio
n
r
esu
lts
,
th
e
NL
L
B
-
2
0
0
m
o
d
el
co
u
l
d
s
er
v
e
a
s
a
s
tr
o
n
g
f
o
u
n
d
atio
n
f
o
r
tr
ain
in
g
a
b
b
c↔in
d
tr
an
s
latio
n
s
y
s
tem
.
Fo
r
th
e
ex
p
er
im
en
ts
in
th
is
s
tu
d
y
,
we
u
s
ed
th
e
d
is
till
ed
6
0
0
M
v
a
r
ian
t
o
f
th
e
NL
L
B
-
2
0
0
m
o
d
el
as
th
e
b
aselin
e
f
o
r
b
o
t
h
o
u
r
b
ilin
g
u
al
u
n
id
ir
ec
tio
n
al
m
o
d
els,
i)
b
b
c→in
d
a
n
d
ii)
in
d
→b
b
c.
Fo
r
th
e
co
m
p
lete
d
etails o
n
h
o
w
we
b
u
ilt an
d
tr
ain
ed
th
ese
m
o
d
els,
p
lease
r
e
f
er
to
th
e
p
r
o
v
id
ed
c
o
d
e
h
er
e
[
1
2
]
.
3.
E
XP
E
R
I
M
E
N
T
A
L
SE
T
T
I
N
G
S
3
.
1
.
Da
t
a
s
et
a
nd
prepro
ce
s
s
i
ng
W
e
u
s
ed
th
e
Nu
s
aX
d
ataset,
a
p
ar
allel
co
r
p
u
s
th
at
co
n
tain
s
b
b
c↔in
d
[
1
3
]
lan
g
u
a
g
e
p
air
c
r
ea
ted
b
y
I
n
d
o
n
esian
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
(
I
n
d
o
NL
P).
T
h
e
I
n
d
o
NL
P tea
m
cr
ea
ted
th
e
d
ataset
[
1
4
]
b
y
tr
a
n
s
latin
g
Sm
SA
[
1
5
]
,
a
r
ec
o
g
n
ized
I
n
d
o
n
esian
s
en
tim
en
t
an
aly
s
is
d
ataset
th
at
co
n
s
is
t
s
o
f
r
ev
iews
an
d
co
m
m
e
n
ts
f
r
o
m
th
e
I
n
d
o
NL
U
b
en
c
h
m
ar
k
[
1
6
]
.
C
o
m
p
eten
t
b
ilin
g
u
al
s
p
ea
k
er
s
wer
e
r
esp
o
n
s
ib
le
f
o
r
th
e
tr
an
s
latio
n
.
I
t
wen
t
th
r
o
u
g
h
r
e
f
in
em
en
t
with
h
u
m
an
-
ass
is
ted
q
u
ality
ass
u
r
an
ce
.
W
e
r
ef
er
th
e
r
ea
d
er
to
[
1
4
]
f
o
r
m
o
r
e
o
n
th
e
co
n
s
tr
u
ctio
n
p
r
o
ce
s
s
es o
f
th
e
d
ataset.
At
p
r
ep
r
o
ce
s
s
in
g
tim
e,
we
u
s
ed
th
e
tr
ain
an
d
test
s
u
b
s
ets
o
f
th
e
Nu
s
aX
d
ataset
a
s
th
e
tr
a
in
in
g
an
d
ev
alu
atio
n
d
atasets
,
r
esp
ec
tiv
ely
.
T
ab
le
1
p
r
e
s
en
ts
th
e
in
f
o
r
m
atio
n
d
etailin
g
th
e
n
u
m
b
e
r
o
f
p
ar
allel
s
en
ten
ce
s
with
in
th
e
s
u
b
s
ets
we
u
s
ed
.
A
f
ter
lo
ad
in
g
u
p
th
e
d
ataset,
as
id
e
f
r
o
m
u
s
in
g
th
e
p
r
e
-
tr
ain
ed
b
u
ilt
-
in
to
k
en
ize
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
8
1
4
I
n
t J Ad
v
Ap
p
l Sci
,
Vo
l.
1
3
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
834
-
8
4
3
836
an
d
d
ata
c
o
llato
r
f
r
o
m
th
e
NL
L
B
-
2
0
0
m
o
d
el,
we
ch
o
s
e
n
o
t
to
p
e
r
f
o
r
m
an
y
ad
d
iti
o
n
al
n
o
r
m
aliza
tio
n
p
r
ep
r
o
ce
s
s
in
g
(
e.
g
.
,
co
llap
s
in
g
p
u
n
ctu
atio
n
,
r
em
o
v
in
g
s
p
ec
ia
l c
h
ar
ac
ter
s
)
o
n
t
o
th
e
d
ataset.
T
ab
le
1
.
Nu
m
b
er
o
f
p
a
r
allel
s
en
ten
ce
s
in
th
e
(
b
b
c↔in
d
)
d
ata
s
et
S
u
b
s
e
t
#
P
a
r
a
l
l
e
l
se
n
t
e
n
c
e
s
t
r
a
i
n
6
,
6
0
6
t
e
st
2
,
0
0
0
W
h
ile
r
em
o
v
in
g
p
u
n
ctu
atio
n
an
d
s
p
ec
ial
ch
ar
ac
ter
s
is
a
co
m
m
o
n
p
r
ep
r
o
ce
s
s
in
g
s
tep
in
s
o
m
e
NL
P
task
s
,
we
o
p
ted
t
o
r
etain
th
e
m
in
th
is
in
s
tan
ce
f
o
r
th
e
f
o
l
lo
win
g
r
ea
s
o
n
s
:
i)
p
r
eser
v
i
n
g
p
o
ten
tially
u
s
ef
u
l
in
f
o
r
m
atio
n
:
Pu
n
ctu
atio
n
m
ar
k
s
ca
n
co
n
v
ey
im
p
o
r
tan
t
in
f
o
r
m
atio
n
s
u
ch
a
s
s
en
ten
ce
en
d
i
n
g
s
,
q
u
esti
o
n
ty
p
es,
an
d
s
p
ea
k
e
r
em
o
tio
n
s
.
Similar
ly
,
s
p
ec
ial
ch
a
r
ac
ter
s
lik
e
cu
r
r
en
cy
s
y
m
b
o
ls
o
r
m
ath
e
m
atica
l
n
o
tatio
n
s
ca
n
h
o
l
d
s
p
ec
if
ic
m
ea
n
in
g
s
with
in
th
e
co
n
tex
t.
R
em
o
v
in
g
th
ese
ele
m
en
ts
co
u
ld
p
o
ten
tially
lead
t
o
m
is
in
ter
p
r
etatio
n
s
d
u
r
i
n
g
t
h
e
tr
an
s
latio
n
p
r
o
ce
s
s
an
d
ii)
Do
m
ain
-
s
p
ec
if
ic
co
n
s
id
er
atio
n
s
:
B
atak
T
o
b
a
m
ig
h
t
u
tili
ze
p
u
n
ctu
atio
n
an
d
s
p
ec
ial
ch
ar
ac
ter
s
in
u
n
i
q
u
e
way
s
ess
en
tial
f
o
r
ac
cu
r
ate
tr
an
s
latio
n
.
I
t
is
a
r
elativ
ely
u
n
d
er
-
r
eso
u
r
ce
d
lan
g
u
ag
e,
s
o
k
ee
p
i
n
g
th
ese
elem
en
ts
m
ig
h
t
b
e
cr
u
cial
to
ca
p
tu
r
in
g
th
e
lan
g
u
ag
e
n
u
a
n
ce
s
.
I
t
is
im
p
o
r
tan
t
to
n
o
te
th
at
th
e
d
ec
is
io
n
to
m
ai
n
tain
o
r
r
e
m
o
v
e
p
u
n
ctu
atio
n
a
n
d
s
p
ec
ial
ch
ar
ac
ter
s
v
ar
ies
b
as
ed
o
n
t
h
e
lan
g
u
ag
e
p
air
,
th
e
tr
an
s
latio
n
m
o
d
el'
s
in
ten
d
ed
ap
p
licatio
n
,
an
d
th
e
d
a
taset
'
s
f
ea
tu
r
es.
Fu
r
th
er
r
esear
ch
m
ig
h
t
b
e
n
ee
d
e
d
to
d
is
co
v
er
th
e
b
est p
r
ep
r
o
ce
s
s
in
g
ap
p
r
o
ac
h
f
o
r
s
u
b
s
eq
u
en
t it
er
atio
n
s
o
f
th
is
m
o
d
el.
T
h
e
m
ax
i
m
u
m
s
eq
u
en
ce
len
g
t
h
is
s
et
to
6
4
to
k
e
n
s
f
o
r
b
o
th
th
e
en
co
d
er
an
d
d
ec
o
d
er
.
T
h
i
s
v
alu
e
is
ch
o
s
en
b
ased
o
n
th
e
o
b
s
er
v
ed
d
is
tr
ib
u
tio
n
o
f
s
en
ten
ce
len
g
t
h
s
in
th
e
d
ataset.
Settin
g
th
e
m
ax
im
u
m
s
eq
u
e
n
ce
len
g
th
to
o
lo
w
co
u
ld
tr
u
n
ca
te
in
f
o
r
m
ativ
e
p
a
r
ts
o
f
th
e
s
en
ten
ce
s
wh
ile
s
ettin
g
it
to
o
h
ig
h
co
u
l
d
lead
to
ex
ce
s
s
iv
e
co
m
p
u
tatio
n
al
c
o
s
t
an
d
p
o
ten
tial
m
em
o
r
y
is
s
u
es.
An
aly
zin
g
th
e
d
ataset
r
e
v
ea
led
th
at
th
e
v
ast
m
ajo
r
ity
o
f
s
en
ten
ce
s
f
all
we
ll
with
in
th
e
6
4
-
to
k
e
n
lim
it,
s
tr
ik
in
g
a
b
ala
n
ce
b
etwe
en
c
ap
tu
r
in
g
n
ec
ess
ar
y
in
f
o
r
m
atio
n
an
d
co
m
p
u
tatio
n
a
l e
f
f
icien
cy
.
T
ab
les
2
an
d
3
s
h
o
w
th
at
t
h
e
s
ec
o
n
d
co
lu
m
n
,
te
x
t,
in
o
u
r
d
ataset
co
n
tain
s
s
en
t
en
ce
s
wr
itten
in
in
d
,
wh
ile
th
e
th
ir
d
c
o
lu
m
n
,
th
e
la
b
el
co
n
tain
s
s
en
ten
ce
s
wr
itten
in
b
b
c.
An
d
s
in
ce
o
u
r
ap
p
r
o
a
ch
in
th
is
s
tu
d
y
is
to
em
p
lo
y
two
c
o
m
p
letely
s
ep
ar
ate
an
d
d
is
tin
ct
b
ilin
g
u
al
u
n
id
ir
ec
tio
n
al
m
o
d
els,
n
a
m
ely
b
b
c→in
d
a
n
d
in
d
→b
b
c;
we
s
im
p
ly
s
witch
e
d
th
e
s
o
u
r
ce
a
n
d
tar
g
et
lan
g
u
ag
es.
Sp
ec
if
ically
,
f
o
r
b
b
c
→in
d
,
we
s
et
th
e
s
en
ten
ce
s
f
r
o
m
th
e
tex
t
c
o
lu
m
n
as
th
e
tar
g
et
lan
g
u
ag
e,
a
n
d
co
n
v
er
s
ely
,
f
o
r
in
d
→b
b
c,
we
s
et
it
as
th
e
s
o
u
r
ce
lan
g
u
ag
e.
3
.
2
.
I
m
ple
m
ent
a
t
io
n a
nd
ev
a
lua
t
io
n
W
e
u
s
ed
th
e
Hu
g
g
in
g
Face
T
ra
n
s
f
o
r
m
er
s
li
b
r
ar
y
an
d
th
e
G
o
o
g
le
C
o
lab
o
r
ato
r
y
s
er
v
ice
p
l
atf
o
r
m
f
o
r
all
o
f
o
u
r
tr
ain
in
g
a
n
d
im
p
l
em
en
tatio
n
.
Go
o
g
le
C
o
lab
o
r
ato
r
y
is
u
s
ed
to
cr
ea
te,
s
h
ar
e,
an
d
r
u
n
ju
p
y
ter
n
o
teb
o
o
k
s
o
n
lin
e
,
wh
ich
ar
e
d
ig
ital
d
o
cu
m
e
n
ts
th
at
ca
n
in
te
g
r
ate
ex
ec
u
tab
le
co
d
e,
im
a
g
e
s
,
an
d
te
x
t
in
o
n
e
f
ile
[
1
7
]
.
G
o
o
g
le
C
o
lab
o
r
ato
r
y
s
e
r
v
es
as
a
p
latf
o
r
m
f
o
r
r
u
n
n
in
g
an
d
s
h
ar
in
g
ea
s
ily
r
ep
licab
le
an
d
m
o
d
i
f
iab
le
co
d
e,
it
is
also
w
id
ely
u
tili
ze
d
in
m
ac
h
in
e
lear
n
in
g
r
esear
c
h
.
Hu
g
g
in
g
F
ac
e
tr
an
s
f
o
r
m
er
s
lib
r
ar
y
is
an
o
p
en
-
s
o
u
r
ce
lib
r
ar
y
t
h
at
p
r
o
v
id
es
a
co
m
p
r
eh
e
n
s
iv
e,
latest
tr
an
s
f
o
r
m
er
ar
ch
itectu
r
e
t
h
r
o
u
g
h
a
u
n
if
ied
API
[
1
8
]
to
s
u
p
p
o
r
t
a
r
an
g
e
o
f
m
ac
h
i
n
e
lear
n
in
g
task
s
.
W
e
th
en
ev
alu
ate
b
o
th
o
u
r
b
ilin
g
u
al
u
n
id
ir
ec
ti
o
n
al
m
o
d
els
b
ased
on
th
e
s
ac
r
eBLE
U
s
co
r
es
m
etr
ic;
an
en
h
an
ce
d
v
er
s
io
n
o
f
th
e
B
L
E
U
m
etr
ic
d
esig
n
ed
to
ad
d
r
ess
th
e
in
co
n
s
is
ten
cy
in
r
ep
o
r
tin
g
s
c
o
r
es
f
r
o
m
th
e
B
L
E
U
m
etr
ic.
T
h
e
m
ain
d
if
f
e
r
en
ce
b
etwe
e
n
s
ac
r
eBLE
U
an
d
B
L
E
U
is
th
at
s
ac
r
eBLE
U
p
r
o
v
id
es
a
s
tan
d
a
r
d
ized
way
o
f
co
m
p
u
tin
g
B
L
E
U
s
co
r
es,
m
ak
in
g
it
ea
s
ier
to
co
m
p
ar
e
s
co
r
es a
cr
o
s
s
d
if
f
er
e
n
t p
ap
er
s
[
1
9
]
.
T
ab
le
2
.
Par
allel
s
en
ten
ce
ex
a
m
p
les f
r
o
m
th
e
tr
ain
s
u
b
s
et
Id
Te
x
t
La
b
e
l
Eq
u
i
v
a
l
e
n
t
t
r
a
n
sl
a
t
i
o
n
i
n
E
n
g
l
i
s
h
0
H
o
b
i
b
a
n
g
e
t
n
g
a
p
a
i
n
?
se
l
f
i
e
,
s
a
m
a
m
a
i
n
a
n
h
p
w
k
w
k
b
e
l
a
j
a
r
j
u
g
a
si
h
t
a
p
i
g
a
k
k
e
ser
i
n
g
a
n
b
g
t
w
k
w
h
t
t
p
:
/
/
a
sk.
f
m
/
a
/
c
0
d
m
q
f
l
l
H
o
b
i
d
o
m
a
m
a
n
g
u
a
?
s
e
l
f
i
e
,
r
a
p
m
a
r
h
p
wkw
k
m
a
rsi
a
j
a
r
j
u
o
s
i
h
t
a
i
i
n
d
a
n
a
j
o
t
j
o
t
t
u
w
k
w
h
t
t
p
:
/
/
a
sk
.
f
m
/
a
/
c
0
d
m
q
f
l
l
Y
o
u
re
a
l
l
y
l
i
k
e
d
o
i
n
g
i
t
w
h
a
t
a
re
y
o
u
d
o
i
n
g
?
s
e
l
f
i
e
,
a
n
d
p
l
a
y
i
n
g
w
i
t
h
m
y
p
h
o
n
e
w
k
w
k
a
n
d
s
t
u
d
y
i
n
g
t
o
o
b
u
t
n
o
t
t
h
a
t
f
re
q
u
e
n
t
l
y
t
h
o
u
g
h
w
k
w
h
t
t
p
:
/
/
a
sk.
f
m
/
a
/
c
0
d
m
q
f
l
l
1
T
h
a
n
k
s
,
b
a
b
e
s
,
f
o
r
c
o
m
i
n
g
@
FN
H
a
z
i
q
a
h
_
@
Ft
i
n
At
i
g
h
a
h
@
_
_
_
szr.
b
a
j
u
d
h
s
a
n
t
e
k
m
a
i
p
o
si
n
g
h
a
n
c
o
q
?
#
H
a
s
n
o
l
A
n
i
s
S
o
l
e
m
n
i
z
a
t
i
o
n
Ma
u
l
i
a
t
e
b
a
b
e
s
f
o
r
c
o
m
i
n
g
@
FN
H
a
z
i
q
a
h
_
@
Ft
i
n
At
i
g
h
a
h
@
_
_
_
szr.
p
a
h
e
a
n
n
u
n
g
g
a
b
a
g
a
k
d
o
n
g
a
n
p
o
s
i
n
g
h
a
n
c
o
q
?
#
H
a
s
n
o
l
A
n
i
s
S
o
l
e
m
n
i
z
a
t
i
o
n
T
h
a
n
k
s
,
b
a
b
e
s,
f
o
r
c
o
m
i
n
g
@
F
N
H
a
z
i
q
a
h
_
@
Ft
i
n
At
i
g
h
a
h
@
_
_
_
sz
r.
m
y
c
l
o
t
h
e
s
w
e
r
e
so
g
o
o
d
w
h
e
n
I
w
a
n
t
e
d
t
o
p
o
se
,
t
h
e
y
g
o
t
ru
i
n
e
d
?
#
H
a
s
n
o
l
A
n
i
s
S
o
l
e
m
n
i
za
t
i
o
n
2
L
u
m
a
y
a
n
n
y
a
m
a
n
,
c
u
m
a
a
c
k
u
r
a
n
g
d
i
n
g
i
n
d
a
n
k
e
b
e
rs
i
h
a
n
p
e
r
l
u
d
i
t
i
n
g
k
a
t
k
a
n
Me
n
a
k
d
o
n
i
a
n
,
a
l
a
i
a
c
d
a
n
g
n
g
a
l
i
j
a
l
a
h
a
i
s
o
n
p
o
rl
u
d
i
t
i
n
g
k
a
t
h
o
n
I
t
'
s
q
u
i
t
e
c
o
m
f
o
rt
a
b
l
e
,
b
u
t
t
h
e
a
c
i
s
n
'
t
c
o
l
d
e
n
o
u
g
h
a
n
d
t
h
e
c
l
e
a
n
l
i
n
e
ss
n
e
e
d
s
t
o
b
e
i
m
p
r
o
v
e
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ad
v
Ap
p
l Sci
I
SS
N:
2252
-
8
8
1
4
B
a
ta
k
To
b
a
la
n
g
u
a
g
e
-
I
n
d
o
n
es
ia
n
ma
ch
in
e
tr
a
n
s
la
tio
n
w
ith
tr
a
n
s
fer lea
r
n
in
g
u
s
in
g
n
o
…
(
C
ev
in
S
a
mu
el)
837
T
ab
le
3
.
Par
allel
s
en
ten
ce
ex
a
m
p
les f
r
o
m
th
e
test
s
u
b
s
et
Id
Te
x
t
La
b
e
l
Eq
u
i
v
a
l
e
n
t
t
r
a
n
sl
a
t
i
o
n
i
n
E
n
g
l
i
s
h
0
Re
c
e
p
t
i
o
n
i
st
k
u
ra
n
g
r
a
m
a
h
,
l
i
f
t
k
o
t
o
r
d
a
n
b
a
u
,
a
i
r
k
a
m
a
r
m
a
n
d
i
k
u
r
a
n
g
l
a
n
c
a
r
t
a
p
i
a
i
r
p
a
n
a
s
d
a
ri
sh
o
w
e
r
m
a
n
t
a
b
,
k
e
b
e
rsi
h
a
n
b
a
n
t
a
l
d
a
n
s
p
r
e
i
p
e
r
l
u
d
i
e
v
a
l
u
a
si
l
a
g
i
.
Am
a
t
d
i
s
a
y
a
n
g
k
a
n
p
o
t
e
n
s
i
y
g
s
a
n
g
a
t
b
a
i
k
t
i
d
a
k
d
i
d
u
k
u
n
g
p
r
o
f
e
si
o
n
a
l
i
sm
e
S
D
M
n
y
a
,
f
u
n
g
s
i
k
o
n
t
r
o
l
n
y
a
p
e
r
l
u
d
i
t
i
n
g
k
a
t
k
a
n
Re
v
e
p
s
i
o
n
i
s
h
u
r
a
n
g
r
a
m
a
,
l
i
f
t
d
u
rs
u
n
ra
p
b
a
u
,
a
e
k
k
a
m
a
r
m
a
r
i
d
i
h
u
r
a
n
g
l
a
n
c
a
r
t
a
i
a
e
k
m
i
l
a
s
n
g
e
n
p
a
c
c
u
r
m
a
n
t
a
b
,
k
a
i
a
sa
n
b
a
t
t
a
l
ra
p
s
i
p
e
re
p
o
r
l
u
d
i
e
v
a
l
u
a
s
i
b
u
se.
a
m
a
n
a
d
i
s
a
y
a
n
g
k
o
n
p
o
t
e
n
s
i
n
a
sa
t
t
a
k
b
u
r
j
u
i
n
d
a
d
i
d
u
k
u
n
g
p
r
o
f
e
si
o
n
a
l
i
sm
e
S
D
M
n
a
,
f
u
n
g
si
k
o
n
t
r
o
l
n
a
p
o
r
l
u
d
i
t
i
n
g
k
a
t
k
o
n
T
h
e
r
e
c
e
p
t
i
o
n
i
st
i
s
n
o
t
f
r
i
e
n
d
l
y
,
t
h
e
e
l
e
v
a
t
o
r
i
s
d
i
r
t
y
a
n
d
s
mel
l
y
,
t
h
e
b
a
t
h
r
o
o
m
w
a
t
e
r
i
s
n
o
t
smo
o
t
h
b
u
t
t
h
e
h
o
t
w
a
t
e
r
f
r
o
m
t
h
e
s
h
o
w
e
r
i
s
g
o
o
d
,
t
h
e
c
l
e
a
n
l
i
n
e
ss
o
f
p
i
l
l
o
w
s
a
n
d
b
e
d
l
i
n
e
n
n
e
e
d
s
t
o
b
e
e
v
a
l
u
a
t
e
d
a
g
a
i
n
.
I
t
i
s
a
p
i
t
y
t
h
a
t
t
h
e
e
x
c
e
l
l
e
n
t
p
o
t
e
n
t
i
a
l
i
s
n
o
t
su
p
p
o
r
t
e
d
b
y
t
h
e
p
r
o
f
e
ssi
o
n
a
l
i
sm
o
f
i
t
s
h
u
m
a
n
r
e
so
u
r
c
e
s
,
t
h
e
c
o
n
t
r
o
l
f
u
n
c
t
i
o
n
n
e
e
d
s
t
o
b
e
i
mp
r
o
v
e
d
1
Bi
a
r
p
u
l
p
e
n
i
t
u
m
e
n
j
a
d
i
k
e
n
a
n
g
a
n
d
a
n
i
n
g
a
t
a
n
b
a
h
w
a
s
a
y
a
p
e
rn
a
h
d
a
t
a
n
g
u
n
t
u
k
b
e
r
t
a
m
u
d
a
n
b
e
r
c
e
r
i
t
a
h
i
n
g
g
a
m
e
n
u
l
i
s.
s
e
n
y
u
m
y
a
n
g
t
i
d
a
k
a
k
a
n
p
e
r
n
a
h
h
i
l
a
n
g
d
a
n
t
a
w
a
y
a
n
g
s
e
l
a
l
u
a
k
a
n
sa
y
a
i
n
g
a
t
,
t
e
r
i
m
a
k
a
si
h
.
m
a
r
i
k
i
t
a
m
e
n
u
t
u
p
h
a
l
i
t
u
d
e
n
g
a
n
l
a
g
u
i
n
i
:
[U
RL]
An
s
o
p
u
l
p
e
n
i
m
a
n
j
a
d
i
k
e
n
a
n
g
a
n
r
a
p
i
n
g
o
t
a
n
b
a
h
w
a
a
u
u
j
u
n
g
r
o
g
i
o
t
m
a
rt
a
m
u
r
a
p
m
a
rc
a
r
i
t
o
sa
m
p
e
m
a
n
u
l
i
s,
m
i
k
i
m
n
a
i
n
d
a
n
a
k
k
a
n
u
n
j
u
n
g
m
a
g
o
r
a
p
t
a
t
a
n
a
sa
l
a
l
u
a
u
i
n
g
o
t
,
t
a
r
i
m
o
k
a
s
i
h
.
m
a
r
i
h
i
t
a
m
a
n
u
t
u
p
h
a
l
i
ra
p
l
a
g
u
o
n
:
[U
RL]
Le
t
t
h
a
t
p
e
n
b
e
a
mem
o
r
y
a
n
d
a
r
e
mi
n
d
e
r
t
h
a
t
I
o
n
c
e
c
a
me
t
o
v
i
si
t
a
n
d
t
e
l
l
st
o
r
i
e
s
t
o
w
r
i
t
e
.
a
s
mi
l
e
t
h
a
t
w
i
l
l
n
e
v
e
r
d
i
sa
p
p
e
a
r
a
n
d
a
l
a
u
g
h
t
h
a
t
I
w
i
l
l
a
l
w
a
y
s
r
e
mem
b
e
r
,
t
h
a
n
k
y
o
u
.
l
e
t
's
c
l
o
se
i
t
w
i
t
h
t
h
i
s
s
o
n
g
:
[
U
R
L]
2
K
a
r
t
o
s
o
e
w
i
ry
o
i
t
u
b
e
r
a
g
a
m
a
d
a
n
m
a
u
b
u
n
u
h
S
o
e
k
a
r
n
o
.
Y
a
n
g
b
u
n
u
h
S
a
y
y
i
d
i
n
a
Al
i
k
e
t
i
k
a
s
o
l
a
t
s
u
b
u
h
i
t
u
seo
r
a
n
g
H
a
f
i
d
z
Q
u
r
a
n
,
I
b
n
u
Mu
l
j
a
m
.
S
a
m
p
a
i
d
i
s
i
n
i
s
u
d
a
h
j
e
l
a
s
b
e
l
u
m
,
b
r
o
?
[U
RL]
K
a
r
t
o
s
o
e
w
i
ry
o
m
a
r
a
g
a
m
a
ra
p
g
i
o
t
m
a
m
b
u
n
u
h
so
e
k
a
rn
o
.
n
a
m
a
m
b
u
n
u
h
sa
y
y
i
d
i
n
a
A
l
i
h
a
t
i
a
su
m
b
a
y
a
n
g
s
u
b
u
h
i
seo
r
a
n
g
H
a
f
i
d
z
Q
u
ra
n
,
i
b
n
u
m
u
l
j
a
m
.
sa
m
p
e
d
i
s
o
n
m
a
d
u
n
g
j
e
l
a
s
sa
n
g
a
n
a
p
e
d
o
,
b
r
o
?
[U
RL]
K
a
r
t
o
s
o
e
w
i
r
y
o
w
a
s
r
e
l
i
g
i
o
u
s
a
n
d
w
a
n
t
e
d
t
o
k
i
l
l
S
o
e
k
a
r
n
o
.
Th
e
o
n
e
w
h
o
k
i
l
l
e
d
S
a
y
y
i
d
i
n
a
A
l
i
d
u
r
i
n
g
t
h
e
d
a
w
n
p
r
a
y
e
r
w
a
s
a
Q
u
r
a
n
H
a
f
i
d
z
,
I
b
n
M
u
l
j
a
m
.
I
s
i
t
c
l
e
a
r
y
e
t
,
b
r
o
?
[
U
R
L]
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
4
.
1
.
Z
er
o
-
s
ho
t
c
a
pa
bil
it
ies o
f
NL
L
B
-
200
W
e
f
ir
s
t
as
s
es
s
ed
th
e
b
aselin
e
p
er
f
o
r
m
an
ce
o
f
th
e
NL
L
B
-
2
0
0
m
o
d
el
with
o
u
t
an
y
tr
ain
in
g
.
T
h
is
in
v
o
lv
ed
m
ea
s
u
r
in
g
its
e
v
alu
a
tio
n
lo
s
s
an
d
s
ac
r
eBLE
U
s
co
r
es
o
n
t
h
e
test
s
u
b
s
et
i
n
T
a
b
le
4
,
f
o
llo
wed
b
y
a
n
ev
alu
atio
n
o
f
its
ze
r
o
-
s
h
o
t
tr
a
n
s
latio
n
ca
p
ab
ilit
ies
f
o
r
b
b
c→
in
d
an
d
i
n
d
→b
b
c
la
n
g
u
a
g
e
p
a
ir
s
(
T
ab
les
5
an
d
6
)
Ou
r
f
in
d
in
g
s
u
p
o
n
lo
o
k
i
n
g
at
th
e
lo
w
s
ac
r
eBLE
U
s
co
r
es
(
0
.
0
to
1
2
.
4
4
)
an
d
c
o
d
e
-
m
ix
ed
tr
an
s
latio
n
s
(
E
n
g
lis
h
/en
g
in
s
tead
o
f
b
b
c
)
in
T
ab
les
5
an
d
6
in
d
icate
th
at
em
p
lo
y
in
g
a
p
r
e
-
tr
ain
e
d
lan
g
u
a
g
e
m
o
d
el,
p
ar
ticu
lar
ly
th
e
NL
L
B
-
2
0
0
m
o
d
el,
is
n
o
t
a
u
s
ab
le
s
tr
ateg
y
f
o
r
ze
r
o
-
s
h
o
t
tr
an
s
latio
n
in
th
e
ca
s
e
o
f
n
ew,
u
n
s
ee
n
lan
g
u
ag
e.
T
h
e
y
also
h
ig
h
lig
h
t
th
at
th
e
ab
s
en
ce
o
f
b
b
c
lab
els
in
th
e
NL
L
B
-
2
0
0
m
o
d
el
p
o
s
es
a
ch
allen
g
e,
ca
u
s
in
g
th
e
m
o
d
el
to
d
ef
au
lt
to
e
n
g
wh
en
b
b
c
is
n
o
t
r
e
co
g
n
ized
as
a
lab
el
o
p
tio
n
i
n
th
e
s
o
u
r
ce
/tar
g
et
lan
g
u
ag
e.
E
v
en
w
h
en
b
b
c
is
s
et
as a
n
o
p
tio
n
,
th
e
m
o
d
el
f
ails
to
r
ec
o
g
n
ize
it a
n
d
d
ef
au
lts
to
en
g
.
C
o
n
s
eq
u
en
tly
,
in
th
e
b
b
c→in
d
d
ir
ec
tio
n
,
wh
e
r
e
th
e
s
o
u
r
ce
s
en
ten
ce
s
ar
e
f
illed
with
u
n
r
ec
o
g
n
ized
en
g
tex
t
(
wh
ic
h
is
ac
tu
ally
i
n
b
b
c)
,
th
ey
ca
n
n
o
t
alig
n
wi
th
th
e
in
d
co
r
p
u
s
,
ca
u
s
in
g
th
e
m
ajo
r
ity
o
f
t
h
e
p
r
ed
ictio
n
s
to
r
em
ai
n
in
b
b
c
(
T
ab
le
5
)
.
O
n
th
e
o
th
e
r
h
a
n
d
,
in
th
e
i
n
d
→b
b
c
d
ir
ec
tio
n
,
wh
er
e
b
b
c
is
n
o
t
r
ec
o
g
n
ized
as
a
tar
g
et
lab
el,
t
h
e
task
b
ec
o
m
es
in
d
→en
g
d
i
r
ec
tio
n
,
r
esu
ltin
g
in
all
p
r
ed
ic
tio
n
s
b
ein
g
in
en
g
(
T
ab
le
6
)
.
T
h
ese
f
i
n
d
in
g
s
a
r
e
i
n
lin
e
with
[
2
0
]
,
wh
er
e
it
is
n
o
ted
th
at
wh
ile
lar
g
e
p
r
e
-
tr
ain
e
d
lan
g
u
ag
e
m
o
d
els
ca
n
g
en
er
ate
co
r
r
ec
t
tr
an
s
latio
n
s
in
ze
r
o
-
s
h
o
t
s
ce
n
ar
io
s
,
th
e
y
ar
e
also
p
r
o
n
e
to
p
r
o
d
u
ci
n
g
tr
an
s
latio
n
s
in
t
h
e
wr
o
n
g
lan
g
u
ag
e
.
4
.
2
.
T
ra
ini
ng
W
e
lev
er
ag
e
th
e
NL
L
B
-
2
0
0
m
o
d
el
f
o
r
b
b
c↔in
d
tr
an
s
latio
n
t
h
r
o
u
g
h
tr
an
s
f
e
r
lear
n
in
g
.
Fro
m
T
ab
le
4
,
we
ca
n
d
ep
ict
f
r
o
m
th
e
h
i
g
h
e
v
alu
atio
n
lo
s
s
as
well
as
th
e
l
o
w
s
ac
r
eBLE
U
s
co
r
es
th
at
th
e
NL
L
B
-
2
0
0
m
o
d
el
n
ee
d
s
f
u
r
th
e
r
tu
n
in
g
o
r
a
d
j
u
s
tm
en
ts
to
ac
h
iev
e
o
p
tim
a
l
p
er
f
o
r
m
an
ce
.
Her
e,
we
d
etail
th
e
s
elec
ted
h
y
p
er
p
ar
am
eter
s
an
d
th
eir
ju
s
t
if
icatio
n
s
d
u
r
in
g
tr
ain
in
g
:
i)
B
atch
s
ize
W
e
em
p
lo
y
an
in
itial
b
atch
s
ize
o
f
4
f
o
r
th
e
tr
ain
in
g
s
u
b
s
et
an
d
1
6
f
o
r
th
e
test
s
u
b
s
et.
T
h
ese
v
alu
es
ar
e
ch
o
s
e
n
b
ased
o
n
th
e
av
ai
lab
le
GPU
m
em
o
r
y
o
n
th
e
em
p
lo
y
ed
p
latf
o
r
m
(
NVI
DI
A
T
esla
T
4
3
2
GB
)
.
Sm
aller
b
atch
s
izes
allo
w
f
o
r
m
o
r
e
f
r
e
q
u
en
t
p
a
r
am
eter
u
p
d
ates
d
u
r
in
g
tr
ain
in
g
,
p
o
te
n
tially
lead
in
g
to
f
aster
co
n
v
er
g
en
ce
.
Ho
we
v
er
,
e
x
ce
s
s
iv
ely
s
m
all
b
atch
es
ca
n
also
in
cr
ea
s
e
tr
ain
in
g
tim
e
d
u
e
to
th
e
o
v
er
h
ea
d
ass
o
ciate
d
with
p
r
o
ce
s
s
in
g
in
d
iv
id
u
al
b
atch
es.
T
h
e
ch
o
s
en
b
atch
s
izes
r
ep
r
esen
t
a
co
m
p
r
o
m
is
e
b
etwe
en
th
ese
co
n
s
id
er
atio
n
s
,
allo
win
g
f
o
r
ef
f
icien
t
u
tili
za
tio
n
o
f
th
e
a
v
ailab
le
m
em
o
r
y
wh
ile
m
ai
n
tain
in
g
r
ea
s
o
n
ab
le
tr
ain
i
n
g
s
p
ee
d
.
ii)
Op
tim
izer
W
e
ap
p
ly
th
e
Ad
a
m
W
(
Ad
a
m
with
weig
h
t
d
ec
a
y
[
2
1
]
)
o
p
tim
izer
with
s
p
ec
if
ic
h
y
p
er
p
ar
am
eter
s
:
1
=
0
.
9
,
2
=
0
.
999
,
=
10
−
8
,
ℎ
_
=
0
.
3
.
Ad
am
W
is
a
wid
ely
u
s
ed
o
p
tim
izer
in
d
ee
p
lear
n
in
g
k
n
o
wn
f
o
r
its
ef
f
icien
cy
an
d
s
tab
ilit
y
in
h
an
d
lin
g
lar
g
e
m
o
d
els.
T
h
e
ch
o
s
en
h
y
p
er
p
a
r
am
et
er
v
alu
es
ar
e
b
ased
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
8
1
4
I
n
t J Ad
v
Ap
p
l Sci
,
Vo
l.
1
3
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
834
-
8
4
3
838
o
n
b
est
p
r
ac
tices
an
d
e
m
p
ir
ica
l
f
in
d
in
g
s
f
r
o
m
p
r
ev
io
u
s
s
tu
d
i
es
o
n
f
in
e
-
tu
n
in
g
p
r
e
-
tr
ain
e
d
m
o
d
els
f
o
r
m
ac
h
in
e
tr
an
s
latio
n
task
s
.
iii)
W
eig
h
t d
ec
ay
W
e
ap
p
ly
weig
h
t
d
ec
ay
to
all
p
ar
am
ete
r
s
ex
ce
p
t
th
e
b
ias
a
n
d
L
a
y
er
No
r
m
weig
h
ts
with
a
v
alu
e
o
f
0
.
3
.
T
h
is
tech
n
iq
u
e
h
elp
s
in
p
r
ev
en
tin
g
o
v
e
r
f
itti
n
g
b
y
p
e
n
al
izin
g
lar
g
e
p
ar
am
eter
v
alu
es
d
u
r
in
g
tr
ai
n
in
g
.
T
h
e
ch
o
s
en
v
alu
e
b
alan
ce
s
th
e
n
e
ed
f
o
r
r
eg
u
lar
izatio
n
t
o
av
o
id
o
v
er
f
itti
n
g
an
d
th
e
n
ee
d
to
a
llo
w
th
e
m
o
d
el
to
lear
n
ef
f
ec
tiv
ely
.
iv
)
L
ea
r
n
in
g
r
ate
s
ch
ed
u
ler
W
e
ad
o
p
t
a
lin
ea
r
lear
n
in
g
r
ate
s
ch
ed
u
ler
with
a
war
m
u
p
p
h
ase
(
Fig
u
r
e
1
)
,
as
im
p
lem
en
ted
in
[
2
2
]
.
T
h
e
in
itial
lear
n
in
g
r
ate
is
s
et
to
5
×
10
−
5
.
T
h
e
war
m
u
p
p
h
ase
last
s
f
o
r
8
2
5
s
tep
s
,
wh
er
e
th
e
lear
n
in
g
r
at
e
in
cr
ea
s
es
lin
ea
r
ly
f
r
o
m
0
to
5
×
10
−
5
.
Fo
r
th
e
r
em
ain
in
g
s
tep
s
(
7
,
4
2
5
)
,
th
e
lear
n
in
g
r
ate
lin
ea
r
ly
d
ec
r
ea
s
es
f
r
o
m
5
×
10
−
5
to
0
.
T
h
is
ap
p
r
o
ac
h
all
o
ws
th
e
m
o
d
el
to
in
itially
ex
p
lo
r
e
th
e
p
ar
am
eter
s
p
ac
e
m
o
r
e
f
r
ee
ly
,
f
o
llo
wed
b
y
a
g
r
ad
u
al
d
ec
r
ea
s
e
in
th
e
lear
n
in
g
r
ate
to
f
in
e
-
tu
n
e
th
e
p
ar
a
m
eter
s
f
o
r
o
p
tim
al
p
er
f
o
r
m
an
ce
.
v)
Mix
ed
p
r
ec
is
io
n
tr
ai
n
in
g
T
o
o
p
tim
ize
tr
ain
in
g
ef
f
icien
c
y
an
d
r
e
d
u
ce
m
em
o
r
y
u
s
ag
e
,
we
em
p
lo
y
m
ix
ed
p
r
ec
is
io
n
tr
ain
in
g
with
FP
1
6
as
im
p
lem
en
ted
in
th
e
F
air
s
eq
lib
r
ar
y
[
2
3
]
.
T
h
is
tech
n
iq
u
e
allo
w
s
th
e
m
o
d
el
to
p
e
r
f
o
r
m
co
m
p
u
tatio
n
s
u
s
in
g
a
lo
wer
p
r
ec
is
io
n
f
o
r
m
a
t
(
1
6
-
b
it)
wh
ile
m
ain
tain
in
g
th
e
g
r
ad
ien
ts
in
a
h
ig
h
er
p
r
ec
is
io
n
f
o
r
m
at
(
3
2
-
b
it)
.
T
h
is
ap
p
r
o
ac
h
r
ed
u
ce
s
m
em
o
r
y
u
s
ag
e
with
o
u
t
r
ed
u
ci
n
g
th
e
m
o
d
el'
s
ac
cu
r
ac
y
.
v
i)
T
o
tal
tr
ain
in
g
tim
e
T
h
e
tr
ain
in
g
p
r
o
ce
s
s
to
o
k
a
p
p
r
o
x
im
ately
1
h
o
u
r
,
5
1
m
in
u
tes
f
o
r
th
e
b
b
c→in
d
d
ir
ec
tio
n
,
an
d
2
h
o
u
r
s
,
2
1
m
i
n
u
tes
f
o
r
th
e
in
d
→b
b
c
d
ir
ec
tio
n
.
T
h
ese
d
u
r
atio
n
s
ar
e
i
n
f
lu
en
ce
d
b
y
f
ac
t
o
r
s
lik
e
t
h
e
d
ataset
s
ize,
ch
o
s
en
h
y
p
er
p
ar
am
eter
s
,
an
d
av
ailab
l
e
h
ar
d
war
e
r
eso
u
r
ce
s
.
T
h
e
tr
ai
n
in
g
lo
s
s
an
d
s
ac
r
eBLE
U
u
s
c
o
r
es
g
r
ap
h
s
f
o
r
o
u
r
two
tr
ain
in
g
s
ar
e
p
r
esen
ted
in
Fig
u
r
e
2
,
esp
ec
ially
in
Fig
u
r
e
s
2
(
a)
a
n
d
(
b
)
.
T
ab
le
4
.
E
v
alu
atio
n
lo
s
s
an
d
s
ac
r
eBLE
U
s
co
r
es o
f
th
e
m
o
d
e
l o
n
th
e
test
s
u
b
s
et
b
ef
o
r
e
tr
ain
in
g
D
i
r
e
c
t
i
o
n
Ev
a
l
u
a
t
i
o
n
l
o
ss (l
o
w
e
r
i
s
b
e
t
t
e
r
)
S
a
c
r
e
B
LEU
(
h
i
g
h
e
r
i
s
b
e
t
t
e
r
)
b
b
c
→
i
n
d
7
.
9
5
1
1
.
4
3
i
n
d
→
b
b
c
8
.
3
7
5
.
0
2
T
ab
le
5
.
E
v
alu
atio
n
o
f
ze
r
o
-
s
h
o
t c
ap
ab
ilit
y
in
b
b
c→in
d
d
ir
ec
tio
n
No
S
o
u
r
c
e
Ta
r
g
e
t
P
r
e
d
i
c
t
i
o
n
S
a
c
r
e
B
LEU
(
h
i
g
h
e
r
i
s
b
e
t
t
e
r
)
Eq
u
i
v
a
l
e
n
t
t
r
a
n
s
l
a
t
i
o
n
o
f
t
a
r
g
e
t
i
n
En
g
l
i
s
h
1
Ad
o
n
g
n
a
b
o
i
h
u
u
r
u
p
i
?
Ad
a
y
a
n
g
b
i
s
a
a
k
u
b
a
n
t
u
?
Ap
a
k
a
h
An
d
a
t
a
k
u
t
d
e
n
g
a
n
o
rg
a
n
i
s
a
si
?
8
.
1
2
I
s
t
h
e
r
e
a
n
y
t
h
i
n
g
I
c
a
n
h
e
l
p
y
o
u
w
i
t
h
?
2
U
b
a
t
n
i
s
u
d
e
s
a
h
i
t
r
o
h
a
n
a
so
n
a
n
g
.
O
b
a
t
s
e
g
a
l
a
p
e
n
y
a
k
i
t
a
d
a
l
a
h
h
a
t
i
y
a
n
g
g
e
m
b
i
r
a
.
U
b
a
t
a
d
a
l
a
h
b
a
g
i
a
n
sel
a
t
a
n
d
a
r
i
s
a
h
i
t
ro
h
a
d
a
n
so
n
a
n
g
.
4
.
9
9
Th
e
c
u
r
e
f
o
r
a
l
l
d
i
s
e
a
se
s
i
s
a
j
o
y
f
u
l
h
e
a
r
t
.
3
T
e
ra
t
u
r
m
a
n
g
a
n
d
o
h
o
t
c
u
k
u
p
p
a
rm
o
d
o
m
m
u
d
a
h
,
a
sa
u
n
a
n
g
m
a
rs
a
h
i
t
b
e
m
u
se.
Ma
k
a
n
t
e
r
a
t
u
r
d
a
n
i
st
i
ra
h
a
t
y
a
n
g
c
u
k
u
p
y
a
,
b
i
a
r
g
a
s
a
k
i
t
l
a
g
i
.
T
e
ra
t
u
r
m
a
n
g
a
n
d
o
h
o
t
c
u
k
u
p
p
a
rm
o
d
o
m
m
u
d
a
h
,
a
sa
u
n
a
n
g
m
a
rs
a
h
i
t
m
e
n
j
a
d
i
m
u
s
e
.
3
.
7
4
Ea
t
r
e
g
u
l
a
r
l
y
a
n
d
g
e
t
e
n
o
u
g
h
r
e
st
,
so
y
o
u
d
o
n
't
g
e
t
si
c
k
a
g
a
i
n
.
4
J
o
l
m
a
n
a
t
a
r
b
e
r
e
n
g
so
n
a
n
g
d
i
j
o
l
o
n
i
n
a
t
o
ro
p
,
d
a
n
g
t
e
n
t
u
j
o
l
m
a
n
a
so
a
d
o
n
g
m
a
s
a
l
a
h
n
a
,
a
l
a
i
i
m
a
j
o
l
m
a
n
a
t
a
n
g
i
s
d
i
s
i
h
a
b
u
n
i
a
n
.
O
ra
n
g
y
a
n
g
b
a
h
a
g
i
a
d
i
d
e
p
a
n
o
r
a
n
g
b
a
n
y
a
k
,
b
u
k
a
n
l
a
h
o
r
a
n
g
y
a
n
g
t
i
d
a
k
p
u
n
y
a
m
a
s
a
l
a
h
,
t
e
t
a
p
i
d
i
a
l
a
h
o
ra
n
g
y
a
n
g
m
e
n
a
n
g
i
s
d
i
k
a
l
a
o
r
a
n
g
t
i
d
a
k
a
d
a
.
J
o
l
m
a
d
a
n
t
a
r
b
e
r
e
n
g
so
n
a
n
g
d
i
j
o
l
o
n
i
n
a
t
o
ro
p
,
d
a
n
g
t
e
n
t
u
j
o
l
m
a
n
a
so
a
d
o
n
g
m
a
s
a
l
a
h
n
a
,
a
l
a
i
i
m
a
j
o
l
m
a
d
a
n
t
a
n
g
i
s
d
i
s
i
h
a
b
u
n
i
a
n
.
1
.
9
8
Th
e
p
e
r
s
o
n
w
h
o
i
s
h
a
p
p
y
i
n
f
r
o
n
t
o
f
p
e
o
p
l
e
i
s
n
o
t
t
h
e
o
n
e
w
h
o
h
a
s
n
o
p
r
o
b
l
e
ms,
b
u
t
h
e
i
s
t
h
e
o
n
e
w
h
o
c
r
i
e
s
w
h
e
n
p
e
o
p
l
e
a
r
e
n
o
t
a
r
o
u
n
d
.
5
Al
e
T
u
h
a
n
,
l
e
a
n
m
a
h
a
s
a
b
a
r
o
n
t
u
k
e
l
u
a
rg
a
n
a
d
i
b
a
g
a
s
p
a
rm
a
r
a
a
n
.
Al
a
n
a
t
a
k
k
a
s
d
o
h
u
b
o
t
o
n
a
m
i
,
s
u
d
e
n
a
m
a
r
h
o
sa
d
o
h
o
t
n
a
m
a
n
g
g
u
l
m
i
t
d
i
p
o
r
t
i
b
i
o
n
t
o
s
a
a
t
t
u
j
o
l
o
M
u
d
o
u
j
u
n
g
n
a
.
An
g
g
i
a
t
m
a
H
o
a
l
e
J
a
h
o
w
a
m
a
n
j
a
l
o
t
o
d
d
i
n
i
k
e
l
u
a
rg
a
n
a
m
i
o
n
,
sa
h
a
t
ro
d
i
t
a
n
g
a
n
Mu
m
a
i
b
a
n
a
,
j
a
l
a
p
a
h
u
d
d
u
l
m
a
i
d
i
i
n
g
a
n
a
n
n
a
so
n
a
n
g
.
Y
a
T
u
h
a
n
,
t
a
b
a
h
k
a
n
l
a
h
k
e
l
u
a
rg
a
y
a
n
g
s
e
d
a
n
g
d
i
l
a
n
d
a
m
u
si
b
a
h
.
K
a
r
e
n
a
,
se
m
u
a
m
a
k
h
l
u
k
d
i
d
u
n
i
a
p
a
s
t
i
a
k
a
n
a
d
a
sa
a
t
n
y
a
m
e
n
e
m
u
i
S
a
n
g
P
e
n
c
i
p
t
a
.
S
e
m
o
g
a
En
g
k
a
u
m
e
n
e
r
i
m
a
a
rw
a
h
s
a
u
d
a
ra
k
a
m
i
d
e
n
g
a
n
t
a
n
g
a
n
M
u
y
a
n
g
t
e
r
b
u
k
a
l
e
b
a
r
m
e
n
y
a
m
b
u
t
n
y
a
d
a
n
m
e
n
e
m
p
a
t
k
a
n
n
y
a
d
i
t
e
m
p
a
t
y
a
n
g
t
e
ri
n
d
a
h
.
Y
a
T
u
h
a
n
,
l
e
a
n
m
a
h
a
s
a
b
a
r
o
n
t
u
k
e
l
u
a
rg
a
d
a
n
d
i
b
a
g
a
s
p
a
r
m
a
r
a
a
n
.
Al
a
n
a
t
a
k
k
a
s
d
o
h
u
b
o
t
o
d
e
n
g
a
n
k
u
,
sel
a
t
a
n
d
a
n
m
a
rh
o
s
a
d
o
h
o
t
d
a
n
m
a
n
g
g
u
l
m
i
t
d
i
p
o
r
t
i
b
i
p
a
d
a
s
a
a
t
i
t
u
a
k
u
p
e
rg
i
k
e
m
u
a
k
h
i
rn
a
.
A
n
g
g
i
a
t
m
a
h
o
a
l
e
j
e
h
o
v
a
d
a
n
k
e
l
u
a
r
g
a
k
u
p
a
d
a
sa
a
t
i
n
i
,
s
a
h
a
t
k
a
m
i
b
e
r
a
d
a
d
i
t
a
n
g
a
n
M
u
d
a
n
i
b
a
n
a
,
j
a
l
a
p
a
h
u
d
d
u
l
d
a
n
d
i
s
i
n
i
s
a
m
a
d
e
n
g
a
n
so
n
a
n
g
.
3
.
1
0
O
G
o
d
,
b
e
s
t
r
o
n
g
f
o
r
t
h
e
f
a
mi
l
y
t
h
a
t
i
s
b
e
i
n
g
h
i
t
b
y
d
i
sas
t
e
r
.
B
e
c
a
u
s
e
,
a
l
l
c
r
e
a
t
u
r
e
s
i
n
t
h
e
w
o
r
l
d
w
i
l
l
d
e
f
i
n
i
t
e
l
y
h
a
v
e
t
h
e
i
r
t
i
m
e
t
o
mee
t
t
h
e
C
r
e
a
t
o
r
.
M
a
y
Y
o
u
a
c
c
e
p
t
o
u
r
b
r
o
t
h
e
r
's
sp
i
r
i
t
w
i
t
h
Y
o
u
r
a
r
ms
w
i
d
e
o
p
e
n
t
o
w
e
l
c
o
me
h
i
m
a
n
d
p
l
a
c
e
h
i
m
i
n
t
h
e
mo
st
b
e
a
u
t
i
f
u
l
p
l
a
c
e
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ad
v
Ap
p
l Sci
I
SS
N:
2252
-
8
8
1
4
B
a
ta
k
To
b
a
la
n
g
u
a
g
e
-
I
n
d
o
n
es
ia
n
ma
ch
in
e
tr
a
n
s
la
tio
n
w
ith
tr
a
n
s
fer lea
r
n
in
g
u
s
in
g
n
o
…
(
C
ev
in
S
a
mu
el)
839
T
ab
le
6
.
E
v
alu
atio
n
o
f
ze
r
o
-
s
h
o
t c
ap
ab
ilit
y
in
in
d
→b
b
c
d
ir
ec
tio
n
No
S
o
u
r
c
e
Ta
r
g
e
t
P
r
e
d
i
c
t
i
o
n
S
a
c
r
e
B
LEU
(
h
i
g
h
e
r
i
s
b
e
t
t
e
r
)
Eq
u
i
v
a
l
e
n
t
t
r
a
n
sl
a
t
i
o
n
o
f
t
a
r
g
e
t
i
n
E
n
g
l
i
sh
1
T
u
ru
n
d
i
s
i
n
i
l
a
h
k
i
t
a
.
T
u
ru
n
d
i
s
o
n
m
a
h
i
t
a
.
G
e
t
d
o
w
n
h
e
r
e
.
1
2
.
4
4
G
e
t
d
o
w
n
h
e
r
e
.
2
I
n
i
p
e
n
g
a
l
a
m
a
n
p
e
r
t
a
m
a
k
u
.
O
n
m
a
p
e
n
g
a
l
a
m
a
n
k
u
n
a
p
a
r
j
o
l
o
.
Th
i
s
i
s
m
y
f
i
r
st
e
x
p
e
r
i
e
n
c
e
.
8
.
1
2
Th
i
s
i
s
my
f
i
r
st
e
x
p
e
r
i
e
n
c
e
.
3
J
i
k
a
k
a
m
u
p
e
r
g
i
,
h
a
t
i
k
u
a
k
a
n
p
e
r
g
i
b
e
rsa
m
a
m
u
.
Mo
l
o
l
a
o
h
o
,
ro
h
a
k
k
u
d
o
h
o
t
m
a
i
l
a
o
.
I
f
y
o
u
g
o
,
my
h
e
a
r
t
w
i
l
l
g
o
w
i
t
h
y
o
u
.
4
.
4
6
I
f
y
o
u
g
o
,
my
h
e
a
r
t
w
i
l
l
g
o
w
i
t
h
y
o
u
.
4
K
e
t
i
k
a
m
e
n
g
h
a
d
a
p
i
m
a
s
a
l
a
h
a
p
a
p
u
n
d
a
l
a
m
k
e
h
i
d
u
p
a
n
i
n
i
m
a
k
a
t
e
r
i
m
a
l
a
h
d
e
n
g
a
n
h
a
t
i
y
a
n
g
s
a
b
a
r
d
a
n
i
m
a
n
y
a
n
g
k
u
a
t
.
Ba
h
e
n
m
a
d
i
ri
m
u
s
o
n
g
o
n
l
a
u
t
n
a
l
u
a
s
,
m
a
n
a
n
g
s
o
n
g
o
n
d
i
a
p
e
m
a
s
a
l
a
h
n
a
r
o
h
t
u
n
g
o
l
u
m
u
,
j
a
l
o
m
a
d
o
h
o
t
ro
h
a
n
a
s
e
re
p
d
o
h
o
t
i
m
a
n
n
a
g
o
g
o
.
N
/
A
0
.
0
W
h
e
n
f
a
c
i
n
g
a
n
y
p
r
o
b
l
e
m
i
n
l
i
f
e
,
a
c
c
e
p
t
i
t
w
i
t
h
a
p
a
t
i
e
n
t
h
e
a
r
t
a
n
d
st
r
o
n
g
f
a
i
t
h
.
5
Bi
a
r
l
a
h
o
r
a
n
g
l
a
i
n
m
e
m
u
j
i
e
n
g
k
a
u
d
a
n
b
u
k
a
n
m
u
l
u
t
m
u
,
o
ra
n
g
y
a
n
g
t
i
d
a
k
k
a
u
k
e
n
a
l
d
a
n
b
u
k
a
n
b
i
b
i
rm
u
s
e
n
d
i
ri
.
Ba
t
u
a
d
a
l
a
h
b
e
r
a
t
d
a
n
p
a
si
r
p
u
n
a
d
a
b
e
r
a
t
n
y
a
,
t
e
t
a
p
i
l
e
b
i
h
b
e
ra
t
d
a
r
i
k
e
d
u
a
-
d
u
a
n
y
a
a
d
a
l
a
h
s
a
k
i
t
h
a
t
i
t
e
r
h
a
d
a
p
o
r
a
n
g
b
o
d
o
h
.
H
a
l
a
k
n
a
a
si
n
g
t
a
g
o
n
a
n
m
a
m
u
j
i
h
o
,
u
n
a
n
g
t
u
n
g
p
a
m
a
n
g
a
n
m
u
sa
n
d
i
r
i
,
h
a
l
a
k
si
l
e
b
a
n
t
a
g
o
n
a
n
,
u
n
a
n
g
b
i
b
i
rm
u
s
a
n
d
i
ri
.
D
o
k
d
o
k
d
o
b
a
t
u
,
j
a
l
a
b
o
r
a
t
h
o
rsi
k
,
a
l
a
i
d
u
m
o
k
d
o
k
s
i
a
n
d
u
a
n
sa
d
o
a
n
g
g
o
h
a
m
u
r
h
i
n
g
o
n
n
i
h
a
l
a
k
n
a
o
t
o
.
Le
t
a
n
o
t
h
e
r
p
r
a
i
se
t
h
e
e
,
a
n
d
n
o
t
t
h
y
mo
u
t
h
;
a
s
t
r
a
n
g
e
r
,
a
n
d
n
o
t
t
h
y
o
w
n
l
i
p
s.
Th
e
st
o
n
e
i
s
h
e
a
v
y
,
a
n
d
t
h
e
san
d
h
e
a
v
y
;
b
u
t
mo
r
e
t
h
a
n
t
h
e
s
e
i
s
t
h
e
so
r
r
o
w
o
f
t
h
e
f
o
o
l.
1
.
3
7
Le
t
a
n
o
t
h
e
r
p
r
a
i
s
e
t
h
e
e
,
a
n
d
n
o
t
t
h
y
m
o
u
t
h
;
a
st
r
a
n
g
e
r
,
a
n
d
n
o
t
t
h
y
o
w
n
l
i
p
s.
T
h
e
st
o
n
e
i
s
h
e
a
v
y
,
a
n
d
t
h
e
sa
n
d
h
e
a
v
y
;
b
u
t
m
o
r
e
t
h
a
n
t
h
e
se
i
s
t
h
e
s
o
r
r
o
w
o
f
t
h
e
f
o
o
l
.
6
J
a
m
a
n
d
a
h
u
l
u
h
i
d
u
p
l
a
h
seo
r
a
n
g
ra
j
a
y
a
n
g
Be
rn
a
m
a
Ra
j
a
R
a
h
a
t
y
a
n
g
b
e
rk
u
a
s
a
d
i
S
a
m
o
si
r
.
R
a
j
a
Ra
h
a
t
a
d
a
l
a
h
ra
j
a
y
a
n
g
d
i
k
e
n
a
l
r
a
k
y
a
t
n
y
a
seb
a
g
a
i
ra
j
a
y
a
n
g
b
i
j
a
k
s
a
n
a
d
a
n
a
d
i
l
.
Pa
r
a
ra
k
y
a
t
n
y
a
p
u
n
sen
a
n
g
m
e
m
i
l
i
k
i
r
a
j
a
se
p
e
r
t
i
Ra
j
a
R
a
h
a
t
.
R
a
j
a
Ra
h
a
t
j
u
g
a
m
e
m
i
l
i
k
i
p
e
rm
a
i
s
u
ri
y
a
n
g
b
a
i
k
h
a
t
i
d
a
n
m
e
m
i
l
i
k
i
a
n
a
k
sa
t
u
y
a
n
g
b
e
r
i
n
a
m
a
Ma
n
g
g
a
l
e
.
N
a
j
o
l
o
a
d
o
n
g
m
a
r
a
j
a
n
a
m
a
rg
o
a
r
R
a
j
a
Ra
h
a
t
n
a
m
a
rh
u
t
a
i
S
a
m
o
si
r
.
R
a
j
a
Ra
h
a
t
t
a
k
k
a
s
d
o
i
t
a
n
d
a
a
k
k
a
ra
i
a
t
n
a
r
a
j
a
n
a
b
i
s
u
k
j
a
l
a
n
a
t
i
g
o
r.
S
o
n
a
n
g
d
o
a
k
k
a
ra
i
a
t
n
a
m
a
n
g
i
d
a
ra
j
a
R
a
h
a
t
o
n
.
R
a
j
a
Ra
h
a
t
o
n
m
a
n
g
a
ri
h
o
n
p
a
r
so
n
d
u
k
b
o
l
o
n
n
a
t
o
n
g
m
a
r
b
a
s
a
ro
h
a
j
a
l
a
m
a
n
g
a
ri
h
o
n
s
a
d
a
i
a
n
a
k
h
o
n
n
a
m
a
r
g
o
a
r
M
a
n
g
g
a
l
e
.
N
/
A
0
.
0
Lo
n
g
a
g
o
t
h
e
r
e
l
i
v
e
d
a
k
i
n
g
n
a
m
e
d
K
i
n
g
R
a
h
a
t
w
h
o
r
u
l
e
d
i
n
S
a
m
o
si
r
.
K
i
n
g
R
a
h
a
t
w
a
s
k
n
o
w
n
b
y
h
i
s
p
e
o
p
l
e
a
s
a
w
i
se
a
n
d
j
u
st
k
i
n
g
.
H
i
s
p
e
o
p
l
e
w
e
r
e
h
a
p
p
y
t
o
h
a
v
e
a
k
i
n
g
l
i
k
e
K
i
n
g
R
a
h
a
t
.
K
i
n
g
R
a
h
a
t
a
l
so
h
a
d
a
q
u
e
e
n
w
h
o
w
a
s
k
i
n
d
a
n
d
h
a
d
o
n
e
c
h
i
l
d
n
a
m
e
d
M
a
n
g
g
a
l
e
.
Fig
u
r
e
1
.
L
ea
r
n
in
g
r
ate
s
ch
ed
u
le
with
war
m
u
p
p
h
ase
f
o
r
t
r
ain
in
g
th
e
NL
L
B
-
2
0
0
m
o
d
el
o
n
th
e
b
b
c↔in
d
d
ataset
4
.
3
.
Resul
t
s
I
n
Fig
u
r
e
2
,
we
illu
s
tr
ate
th
e
r
esu
lts
ac
h
iev
ed
b
y
tr
ain
in
g
th
e
NL
L
B
-
2
0
0
m
o
d
el
o
n
L
R
L
s
p
ar
allel
co
r
p
o
r
a
f
o
r
b
o
t
h
in
d
→b
b
c
an
d
b
b
c→in
d
lan
g
u
a
g
e
d
ir
ec
tio
n
s
.
Ou
r
an
aly
s
is
o
f
th
ese
f
in
d
in
g
s
s
u
g
g
ests
th
at
lev
er
ag
in
g
L
R
L
s
p
ar
allel
co
r
p
o
r
a
f
o
r
p
r
e
-
tr
ain
ed
lan
g
u
ag
e
m
o
d
el
tr
ain
in
g
is
a
p
r
o
m
is
in
g
ap
p
r
o
ac
h
,
e
v
en
f
o
r
s
ce
n
ar
io
s
in
v
o
lv
in
g
n
ew,
u
n
s
ee
n
lan
g
u
ag
es,
Fig
u
r
e
2
(
a)
s
h
o
ws
T
r
ain
in
g
lo
s
s
an
d
Fig
u
r
e
2
(
b
)
s
h
o
ws
s
ac
r
eBLE
U
s
co
r
es.
T
h
is
ap
p
r
o
ac
h
ca
n
allev
iate
th
e
lim
itatio
n
s
ass
o
ciate
d
with
th
e
s
ca
r
ci
ty
o
f
d
ata
in
L
R
L
s
.
I
n
co
m
p
ar
is
o
n
to
th
e
s
ac
r
eBLE
U
s
co
r
es
o
f
th
e
m
o
d
el
b
ef
o
r
e
tr
ain
in
g
(
T
ab
le
4
)
,
th
e
m
o
d
el
d
em
o
n
s
tr
ated
a
s
ig
n
if
ican
t
im
p
r
o
v
em
en
t,
ac
h
i
ev
in
g
s
ac
r
eBLE
U
s
co
r
es
o
f
u
p
to
3
7
.
1
0
(
+
2
5
.
6
7
,
u
p
f
r
o
m
1
1
.
4
3
)
in
b
b
c→in
d
an
d
3
0
.
8
4
(
+2
5
.
8
2
,
u
p
f
r
o
m
5
.
0
2
)
in
i
n
d
→b
b
c.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
8
1
4
I
n
t J Ad
v
Ap
p
l Sci
,
Vo
l.
1
3
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
834
-
8
4
3
840
(
a)
(
b
)
Fig
u
r
e
2
.
Mo
d
el
af
ter
tr
ai
n
in
g
in
in
d
→b
b
c
a
n
d
b
b
c→in
d
d
ir
e
ctio
n
s
(
lo
wer
tr
ain
in
g
lo
s
s
is
b
etter
,
h
ig
h
er
s
ac
r
eBLE
U
s
co
r
e
is
b
etter
)
o
f
(
a)
T
r
ain
in
g
lo
s
s
an
d
(
b
)
s
ac
r
e
B
L
E
U
s
co
r
es
T
h
e
r
esu
lts
d
em
o
n
s
tr
ate
th
at
NL
L
B
-
2
0
0
is
s
u
ited
to
p
r
o
d
u
cin
g
h
ig
h
-
q
u
al
ity
tr
a
n
s
latio
n
s
,
ev
en
f
o
r
a
n
ew
an
d
u
n
s
ee
n
L
R
L
s
p
air
.
T
h
is
ca
p
ab
ilit
y
is
d
em
o
n
s
tr
ated
b
y
its
p
r
o
f
icien
cy
in
tr
ain
i
n
g
with
o
n
ly
6
,
6
0
0
p
ar
allel
s
en
ten
ce
s
f
r
o
m
th
e
t
r
ain
in
g
s
u
b
s
et,
wh
ich
wer
e
p
r
ev
io
u
s
ly
n
ew
an
d
u
n
s
ee
n
b
y
th
e
m
o
d
el
b
e
f
o
r
e
tr
ain
in
g
.
Fo
r
m
o
r
e
e
x
am
p
les,
we
ca
n
co
m
p
ar
e
th
e
s
ac
r
eBL
E
U
s
co
r
es
o
n
th
e
test
s
u
b
s
et
in
T
ab
les
5
an
d
6
.
Af
ter
th
e
t
r
ain
in
g
in
T
ab
les
7
an
d
8
,
th
e
im
p
r
o
v
em
e
n
t
in
s
ac
r
eBLE
U
s
co
r
es
r
an
g
es
f
r
o
m
+0
.
6
3
(
Par
allel
s
en
ten
ce
ex
am
p
le
n
o
.
2
o
f
T
a
b
les
6
a
n
d
8
)
u
p
to
+4
6
.
7
8
p
o
in
ts
(
Par
allel
s
en
ten
ce
e
x
am
p
l
e
n
o
.
3
o
f
T
a
b
les
6
an
d
8
)
.
T
ab
le
7
.
E
v
alu
atio
n
o
f
th
e
m
o
d
el
af
ter
tr
ain
in
g
in
b
b
c→in
d
d
ir
ec
tio
n
No
S
o
u
r
c
e
Ta
r
g
e
t
P
r
e
d
i
c
t
i
o
n
S
a
c
r
e
B
LEU
(
h
i
g
h
e
r
i
s
b
e
t
t
e
r
)
Eq
u
i
v
a
l
e
n
t
t
r
a
n
s
l
a
t
i
o
n
o
f
t
a
r
g
e
t
i
n
En
g
l
i
s
h
1
Ad
o
n
g
n
a
b
o
i
h
u
u
r
u
p
i
?
Ad
a
y
a
n
g
b
i
s
a
a
k
u
b
a
n
t
u
?
Ad
a
y
a
n
g
b
i
s
a
s
a
y
a
b
a
n
t
u
?
3
7
.
9
9
I
s
t
h
e
r
e
a
n
y
t
h
i
n
g
I
c
a
n
h
e
l
p
y
o
u
w
i
t
h
?
2
U
b
a
t
n
i
s
u
d
e
sa
h
i
t
r
o
h
a
n
a
so
n
a
n
g
.
O
b
a
t
s
e
g
a
l
a
p
e
n
y
a
k
i
t
a
d
a
l
a
h
h
a
t
i
y
a
n
g
g
e
m
b
i
ra
.
O
b
a
t
sem
u
a
s
a
k
i
t
h
a
t
i
y
g
b
a
h
a
g
i
a
.
6
.
7
7
Th
e
c
u
r
e
f
o
r
a
l
l
d
i
s
e
a
se
s
i
s
a
j
o
y
f
u
l
h
e
a
r
t
.
3
T
e
ra
t
u
r
m
a
n
g
a
n
d
o
h
o
t
c
u
k
u
p
p
a
rm
o
d
o
m
m
u
d
a
h
,
a
sa
u
n
a
n
g
m
a
rsa
h
i
t
b
e
m
u
se
.
Ma
k
a
n
t
e
r
a
t
u
r
d
a
n
i
st
i
ra
h
a
t
y
a
n
g
c
u
k
u
p
y
a
,
b
i
a
r
g
a
s
a
k
i
t
l
a
g
i
.
T
e
ra
t
u
r
m
a
k
a
n
d
a
n
c
u
k
u
p
t
i
d
u
r
y
a
,
b
i
a
r
g
a
s
a
k
i
t
l
a
g
i
.
5
0
.
5
2
Ea
t
r
e
g
u
l
a
r
l
y
a
n
d
g
e
t
e
n
o
u
g
h
r
e
s
t
,
so
y
o
u
d
o
n
'
t
g
e
t
si
c
k
a
g
a
i
n
.
4
J
o
l
m
a
n
a
t
a
r
b
e
r
e
n
g
so
n
a
n
g
d
i
j
o
l
o
n
i
n
a
t
o
r
o
p
,
d
a
n
g
t
e
n
t
u
j
o
l
m
a
n
a
s
o
a
d
o
n
g
m
a
s
a
l
a
h
n
a
,
a
l
a
i
i
m
a
j
o
l
m
a
n
a
t
a
n
g
i
s
d
i
s
i
h
a
b
u
n
i
a
n
.
O
ra
n
g
y
a
n
g
b
a
h
a
g
i
a
d
i
d
e
p
a
n
o
ra
n
g
b
a
n
y
a
k
,
b
u
k
a
n
l
a
h
o
r
a
n
g
y
a
n
g
t
i
d
a
k
p
u
n
y
a
m
a
s
a
l
a
h
,
t
e
t
a
p
i
d
i
a
l
a
h
o
r
a
n
g
y
a
n
g
m
e
n
a
n
g
i
s
d
i
k
a
l
a
o
r
a
n
g
t
i
d
a
k
a
d
a
.
O
ra
n
g
y
a
n
g
t
e
r
l
i
h
a
t
b
a
h
a
g
i
a
d
i
k
a
m
p
u
n
g
h
a
l
a
m
a
n
,
b
u
k
a
n
t
e
n
t
u
o
r
a
n
g
y
a
n
g
t
i
d
a
k
p
u
n
y
a
m
a
sa
l
a
h
,
t
a
p
i
a
d
a
l
a
h
o
r
a
n
g
y
a
n
g
m
e
n
a
n
g
i
s
d
i
se
b
u
a
h
ru
m
a
h
.
2
8
.
6
3
Th
e
p
e
r
so
n
w
h
o
i
s
h
a
p
p
y
i
n
f
r
o
n
t
o
f
p
e
o
p
l
e
i
s n
o
t
t
h
e
o
n
e
w
h
o
h
a
s
n
o
p
r
o
b
l
e
ms,
b
u
t
h
e
i
s
t
h
e
o
n
e
w
h
o
c
r
i
e
s
w
h
e
n
p
e
o
p
l
e
a
r
e
n
o
t
a
r
o
u
n
d
.
5
Al
e
T
u
h
a
n
,
l
e
a
n
m
a
h
a
s
a
b
a
r
o
n
t
u
k
e
l
u
a
rg
a
n
a
d
i
b
a
g
a
s
p
a
rm
a
ra
a
n
.
Al
a
n
a
t
a
k
k
a
s
d
o
h
u
b
o
t
o
n
a
m
i
,
s
u
d
e
n
a
m
a
r
h
o
s
a
d
o
h
o
t
n
a
m
a
n
g
g
u
l
m
i
t
d
i
p
o
r
t
i
b
i
o
n
t
o
s
a
a
t
t
u
j
o
l
o
M
u
d
o
u
j
u
n
g
n
a
.
An
g
g
i
a
t
m
a
H
o
a
l
e
J
a
h
o
w
a
m
a
n
j
a
l
o
t
o
d
d
i
n
i
k
e
l
u
a
rg
a
n
a
m
i
o
n
,
sa
h
a
t
ro
d
i
t
a
n
g
a
n
M
u
m
a
i
b
a
n
a
,
j
a
l
a
p
a
h
u
d
d
u
l
m
a
i
d
i
i
n
g
a
n
a
n
n
a
so
n
a
n
g
.
Y
a
T
u
h
a
n
,
t
a
b
a
h
k
a
n
l
a
h
k
e
l
u
a
rg
a
y
a
n
g
se
d
a
n
g
d
i
l
a
n
d
a
m
u
s
i
b
a
h
.
K
a
re
n
a
,
semu
a
m
a
k
h
l
u
k
d
i
d
u
n
i
a
p
a
s
t
i
a
k
a
n
a
d
a
s
a
a
t
n
y
a
m
e
n
e
m
u
i
S
a
n
g
Pe
n
c
i
p
t
a
.
S
e
m
o
g
a
E
n
g
k
a
u
m
e
n
e
r
i
m
a
a
rw
a
h
s
a
u
d
a
ra
k
a
m
i
d
e
n
g
a
n
t
a
n
g
a
n
M
u
y
a
n
g
t
e
r
b
u
k
a
l
e
b
a
r
m
e
n
y
a
m
b
u
t
n
y
a
d
a
n
m
e
n
e
m
p
a
t
k
a
n
n
y
a
d
i
t
e
m
p
a
t
y
a
n
g
t
e
ri
n
d
a
h
.
Y
a
T
u
h
a
n
,
b
e
ri
l
a
h
k
e
s
a
b
a
ra
n
k
e
p
a
d
a
k
e
l
u
a
rg
a
y
a
n
g
b
e
r
a
d
a
d
i
d
a
l
a
m
su
r
g
a
.
K
a
r
e
n
a
k
a
m
i
p
a
st
i
t
a
h
u
,
b
a
h
w
a
sem
u
a
y
a
n
g
b
e
rd
o
s
a
d
a
n
m
e
n
g
i
k
h
l
a
s
k
a
n
d
i
r
i
d
i
d
u
n
i
a
i
n
i
a
k
a
n
d
a
t
a
n
g
m
e
n
j
e
l
a
n
g
Mu
p
a
d
a
a
k
h
i
rn
y
a
.
S
e
m
o
g
a
En
g
k
a
u
t
e
t
a
p
i
j
u
g
a
Y
a
Al
l
a
h
m
e
n
e
ri
m
a
se
l
u
ru
h
k
e
l
u
a
rg
a
k
a
m
i
,
sa
m
p
a
i
m
e
rek
a
sa
m
p
a
i
d
i
t
a
n
g
a
n
M
u
,
d
a
n
m
e
n
d
u
d
u
k
k
a
n
m
e
re
k
a
d
i
t
e
m
p
a
t
y
a
n
g
m
e
n
y
e
n
a
n
g
k
a
n
.
7
.
4
6
O
G
o
d
,
b
e
s
t
r
o
n
g
f
o
r
t
h
e
f
a
mi
l
y
t
h
a
t
i
s
b
e
i
n
g
h
i
t
b
y
d
i
sas
t
e
r
.
B
e
c
a
u
s
e
,
a
l
l
c
r
e
a
t
u
r
e
s
i
n
t
h
e
w
o
r
l
d
w
i
l
l
d
e
f
i
n
i
t
e
l
y
h
a
v
e
t
h
e
i
r
t
i
me
t
o
mee
t
t
h
e
C
r
e
a
t
o
r
.
M
a
y
Y
o
u
a
c
c
e
p
t
o
u
r
b
r
o
t
h
e
r
's
s
p
i
r
i
t
w
i
t
h
Y
o
u
r
a
r
ms
w
i
d
e
o
p
e
n
t
o
w
e
l
c
o
me
h
i
m
a
n
d
p
l
a
c
e
h
i
m
i
n
t
h
e
m
o
st
b
e
a
u
t
i
f
u
l
p
l
a
c
e
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ad
v
Ap
p
l Sci
I
SS
N:
2252
-
8
8
1
4
B
a
ta
k
To
b
a
la
n
g
u
a
g
e
-
I
n
d
o
n
es
ia
n
ma
ch
in
e
tr
a
n
s
la
tio
n
w
ith
tr
a
n
s
fer lea
r
n
in
g
u
s
in
g
n
o
…
(
C
ev
in
S
a
mu
el)
841
T
ab
le
8
.
E
v
alu
atio
n
o
f
th
e
m
o
d
el
af
ter
tr
ain
in
g
in
in
d
→b
b
c
d
ir
ec
tio
n
No
S
o
u
r
c
e
Ta
r
g
e
t
P
r
e
d
i
c
t
i
o
n
S
a
c
r
e
B
LEU
(
h
i
g
h
e
r
i
s
b
e
t
t
e
r
)
Eq
u
i
v
a
l
e
n
t
t
r
a
n
sl
a
t
i
o
n
o
f
Ta
r
g
e
t
i
n
E
n
g
l
i
s
h
1
T
u
ru
n
d
i
s
i
n
i
l
a
h
k
i
t
a
.
T
u
ru
n
d
i
s
o
n
m
a
h
i
t
a
.
T
u
ru
n
t
u
s
o
n
m
a
h
i
t
a
.
4
2
.
7
3
G
e
t
d
o
w
n
h
e
r
e
.
2
I
n
i
p
e
n
g
a
l
a
m
a
n
p
e
r
t
a
m
a
k
u
.
O
n
m
a
p
e
n
g
a
l
a
m
a
n
k
u
n
a
p
a
r
j
o
l
o
.
o
n
p
e
n
g
a
l
a
m
a
n
p
e
rt
a
m
a
k
u
.
8
.
7
5
Th
i
s
i
s
m
y
f
i
r
st
e
x
p
e
r
i
e
n
c
e
.
3
J
i
k
a
k
a
m
u
p
e
r
g
i
,
h
a
t
i
k
u
a
k
a
n
p
e
r
g
i
b
e
r
sa
m
a
m
u
.
Mo
l
o
l
a
o
h
o
,
r
o
h
a
k
k
u
d
o
h
o
t
m
a
i
l
a
o
.
Mo
l
o
l
a
o
h
o
,
r
o
h
a
k
u
l
a
h
o
ra
p
d
o
h
o
t
h
o
.
2
9
.
0
7
I
f
y
o
u
g
o
,
m
y
h
e
a
r
t
w
i
l
l
g
o
w
i
t
h
y
o
u
.
4
K
e
t
i
k
a
m
e
n
g
h
a
d
a
p
i
m
a
sa
l
a
h
a
p
a
p
u
n
d
a
l
a
m
k
e
h
i
d
u
p
a
n
i
n
i
m
a
k
a
t
e
ri
m
a
l
a
h
d
e
n
g
a
n
h
a
t
i
y
a
n
g
s
a
b
a
r
d
a
n
i
m
a
n
y
a
n
g
k
u
a
t
.
Ba
h
e
n
m
a
d
i
r
i
m
u
so
n
g
o
n
l
a
u
t
n
a
l
u
a
s,
m
a
n
a
n
g
so
n
g
o
n
d
i
a
p
e
m
a
s
a
l
a
h
n
a
r
o
h
t
u
n
g
o
l
u
m
u
,
j
a
l
o
m
a
d
o
h
o
t
ro
h
a
n
a
ser
e
p
d
o
h
o
t
i
m
a
n
n
a
g
o
g
o
.
H
a
t
i
a
m
a
n
g
a
d
o
p
i
m
a
sa
l
a
h
a
h
a
p
e
d
a
l
a
m
h
i
d
u
p
o
n
m
a
k
a
t
e
ri
m
a
m
a
ra
p
h
a
t
i
n
a
s
a
b
a
r
r
a
p
i
m
a
n
n
a
g
o
g
o
.
1
1
.
2
9
W
h
e
n
f
a
c
i
n
g
a
n
y
p
r
o
b
l
e
m
i
n
l
i
f
e
,
a
c
c
e
p
t
i
t
w
i
t
h
a
p
a
t
i
e
n
t
h
e
a
r
t
a
n
d
s
t
r
o
n
g
f
a
i
t
h
.
5
Bi
a
r
l
a
h
o
r
a
n
g
l
a
i
n
m
e
m
u
j
i
e
n
g
k
a
u
d
a
n
b
u
k
a
n
m
u
l
u
t
m
u
,
o
ra
n
g
y
a
n
g
t
i
d
a
k
k
a
u
k
e
n
a
l
d
a
n
b
u
k
a
n
b
i
b
i
rm
u
se
n
d
i
r
i
.
B
a
t
u
a
d
a
l
a
h
b
e
ra
t
d
a
n
p
a
si
r
p
u
n
a
d
a
b
e
r
a
t
n
y
a
,
t
e
t
a
p
i
l
e
b
i
h
b
e
ra
t
d
a
r
i
k
e
d
u
a
-
d
u
a
n
y
a
a
d
a
l
a
h
s
a
k
i
t
h
a
t
i
t
e
r
h
a
d
a
p
o
r
a
n
g
b
o
d
o
h
.
H
a
l
a
k
n
a
a
s
i
n
g
t
a
g
o
n
a
n
m
a
m
u
j
i
h
o
,
u
n
a
n
g
t
u
n
g
p
a
m
a
n
g
a
n
m
u
s
a
n
d
i
ri
,
h
a
l
a
k
si
l
e
b
a
n
t
a
g
o
n
a
n
,
u
n
a
n
g
b
i
b
i
rm
u
s
a
n
d
i
ri
.
D
o
k
d
o
k
d
o
b
a
t
u
,
j
a
l
a
b
o
ra
t
h
o
rs
i
k
,
a
l
a
i
d
u
m
o
k
d
o
k
si
a
n
d
u
a
n
s
a
d
o
a
n
g
g
o
h
a
m
u
r
h
i
n
g
o
n
n
i
h
a
l
a
k
n
a
o
t
o
.
Asa
d
i
p
u
j
i
h
o
j
o
l
m
a
n
a
a
si
n
g
j
a
l
a
d
a
n
g
d
i
su
n
g
k
u
n
m
u
,
j
o
l
m
a
n
a
d
a
n
g
d
i
b
o
t
o
h
o
j
a
l
a
d
a
n
g
d
i
b
i
b
i
rm
u
s
a
n
d
i
r
i
.
B
a
t
u
n
a
b
o
r
a
t
j
a
l
a
p
a
s
i
r
p
e
a
d
o
n
g
b
o
r
a
t
n
a
,
a
l
a
i
l
o
b
i
b
o
r
a
t
si
a
n
n
a
d
u
a
-
d
u
a
i
i
m
a
h
a
t
c
i
t
r
o
h
a
t
u
j
o
l
m
a
o
t
o
.
5
.
4
4
Le
t
a
n
o
t
h
e
r
p
r
a
i
s
e
t
h
e
e
,
a
n
d
n
o
t
t
h
y
mo
u
t
h
;
a
s
t
r
a
n
g
e
r
,
a
n
d
n
o
t
t
h
y
o
w
n
l
i
p
s
.
Th
e
st
o
n
e
i
s
h
e
a
v
y
,
a
n
d
t
h
e
san
d
h
e
a
v
y
;
b
u
t
m
o
r
e
t
h
a
n
t
h
e
se
i
s
t
h
e
so
r
r
o
w
o
f
t
h
e
f
o
o
l
.
6
J
a
m
a
n
d
a
h
u
l
u
h
i
d
u
p
l
a
h
seo
r
a
n
g
ra
j
a
y
a
n
g
Be
rn
a
m
a
Ra
j
a
Ra
h
a
t
y
a
n
g
b
e
r
k
u
a
s
a
d
i
S
a
m
o
s
i
r.
R
a
j
a
Ra
h
a
t
a
d
a
l
a
h
r
a
j
a
y
a
n
g
d
i
k
e
n
a
l
ra
k
y
a
t
n
y
a
se
b
a
g
a
i
ra
j
a
y
a
n
g
b
i
j
a
k
s
a
n
a
d
a
n
a
d
i
l
.
Pa
r
a
r
a
k
y
a
t
n
y
a
p
u
n
sen
a
n
g
m
e
m
i
l
i
k
i
r
a
j
a
sep
e
rt
i
R
a
j
a
R
a
h
a
t
.
R
a
j
a
Ra
h
a
t
j
u
g
a
m
e
m
i
l
i
k
i
p
e
rm
a
i
s
u
ri
y
a
n
g
b
a
i
k
h
a
t
i
d
a
n
m
e
m
i
l
i
k
i
a
n
a
k
s
a
t
u
y
a
n
g
b
e
ri
n
a
m
a
M
a
n
g
g
a
l
e
.
N
a
j
o
l
o
a
d
o
n
g
m
a
r
a
j
a
n
a
m
a
rg
o
a
r
R
a
j
a
R
a
h
a
t
n
a
m
a
rh
u
t
a
i
S
a
m
o
s
i
r.
R
a
j
a
Ra
h
a
t
t
a
k
k
a
s
d
o
i
t
a
n
d
a
a
k
k
a
ra
i
a
t
n
a
r
a
j
a
n
a
b
i
s
u
k
j
a
l
a
n
a
t
i
g
o
r
.
S
o
n
a
n
g
d
o
a
k
k
a
r
a
i
a
t
n
a
m
a
n
g
i
d
a
r
a
j
a
Ra
h
a
t
o
n
.
R
a
j
a
R
a
h
a
t
o
n
m
a
n
g
a
ri
h
o
n
p
a
rs
o
n
d
u
k
b
o
l
o
n
n
a
t
o
n
g
m
a
r
b
a
s
a
ro
h
a
j
a
l
a
m
a
n
g
a
r
i
h
o
n
s
a
d
a
i
a
n
a
k
h
o
n
n
a
m
a
r
g
o
a
r
Ma
n
g
g
a
l
e
.
Z
a
m
a
n
n
a
j
o
l
o
m
a
n
g
o
l
u
m
a
sa
d
a
r
a
j
a
n
a
m
a
r
g
o
a
r
Ra
j
a
R
a
h
a
t
n
a
m
a
rk
u
a
sa
d
i
S
a
m
o
si
r.
R
a
j
a
Ra
h
a
t
i
m
a
ra
j
a
n
a
d
i
b
o
t
o
ra
k
y
a
t
n
a
s
e
b
a
g
a
i
r
a
j
a
n
a
b
i
j
a
k
j
a
l
a
a
d
i
l
.
Pa
ra
ra
k
y
a
t
n
a
p
e
s
o
n
a
n
g
p
u
n
a
ra
j
a
so
n
g
o
n
Ra
j
a
R
a
h
a
t
.
Ra
j
a
Ra
h
a
t
p
e
p
u
n
a
p
e
rm
a
i
s
u
ri
n
a
b
u
r
j
u
r
o
h
a
j
a
l
a
p
u
n
a
a
n
a
k
sa
d
a
n
a
m
a
n
g
a
l
e
a
n
g
o
a
r
Ma
n
g
g
a
l
e
.
1
5
.
1
2
Lo
n
g
a
g
o
t
h
e
r
e
l
i
v
e
d
a
k
i
n
g
n
a
m
e
d
K
i
n
g
R
a
h
a
t
w
h
o
r
u
l
e
d
i
n
S
a
mo
s
i
r
.
K
i
n
g
R
a
h
a
t
w
a
s
k
n
o
w
n
b
y
h
i
s
p
e
o
p
l
e
a
s
a
w
i
se
a
n
d
j
u
s
t
k
i
n
g
.
H
i
s
p
e
o
p
l
e
w
e
r
e
h
a
p
p
y
t
o
h
a
v
e
a
k
i
n
g
l
i
k
e
K
i
n
g
R
a
h
a
t
.
K
i
n
g
R
a
h
a
t
a
l
s
o
h
a
d
a
q
u
e
e
n
w
h
o
w
a
s
k
i
n
d
a
n
d
h
a
d
o
n
e
c
h
i
l
d
n
a
m
e
d
M
a
n
g
g
a
l
e
.
4
.
4
.
B
enchm
a
r
k
ing
o
ur
re
s
ults a
g
a
ins
t
prio
r
s
t
a
t
e
-
of
-
t
he
-
ar
t
B
a
t
a
k
T
o
ba
-
I
nd
o
nes
ia
n t
ra
ns
la
t
io
n
On
e
o
f
o
u
r
m
o
s
t
im
p
o
r
tan
t
a
ch
iev
em
en
ts
is
th
at,
in
co
m
p
ar
is
o
n
to
p
r
ev
io
u
s
s
tu
d
ies
o
n
b
b
c↔in
d
m
ac
h
in
e
tr
a
n
s
latio
n
,
s
p
ec
if
ica
lly
r
ef
er
e
n
cin
g
[
1
4
]
,
[
2
4
]
,
o
u
r
ap
p
r
o
ac
h
ac
h
iev
es
th
e
b
est
r
esu
lt
in
s
ac
r
eBLE
U
s
co
r
es.
As
s
ee
n
in
T
ab
le
9
,
o
u
r
m
o
d
els
h
av
e
a
s
ac
r
eBLE
U
o
f
i)
3
7
.
1
0
f
o
r
b
b
c→in
d
an
d
ii)
3
0
.
8
4
f
o
r
in
d
→b
b
c;
b
o
t
h
ar
e
h
ig
h
er
co
m
p
ar
ed
to
[
1
4
]
’
s
s
ac
r
eBLE
U
o
f
i)
2
0
.
9
4
f
o
r
b
b
c→in
d
an
d
ii
)
1
8
.
4
1
f
o
r
in
d
→b
b
c
;
[
2
4
]
’
s
s
ac
r
eBLE
U
o
f
i)
2
0
.
4
8
f
o
r
b
b
c→in
d
an
d
ii)
1
8
.
4
1
f
o
r
in
d
→b
b
c.
I
t
’
s
wo
r
th
n
o
t
in
g
th
at
wh
ile
[
2
4
]
em
p
lo
y
ed
th
e
s
am
e
b
aselin
e
m
o
d
el
as
o
u
r
s
t
h
e
NL
L
B
-
2
0
0
’
s
d
is
till
ed
6
0
0
M
v
ar
ia
n
t
m
o
d
el
in
o
n
e
o
f
th
eir
ex
p
er
im
en
ts
,
th
e
y
d
i
d
n
’
t
i
n
tr
o
d
u
ce
it
to
th
e
b
b
c↔in
d
lan
g
u
ag
e
p
air
.
T
h
is
lim
itatio
n
ar
is
e
s
f
r
o
m
th
e
f
ac
t
th
at
th
e
NL
L
B
-
2
0
0
m
o
d
el
is
n
o
t
tr
ain
ed
o
n
t
h
e
b
b
c↔in
d
lan
g
u
ag
e
p
air
[
9
]
,
wh
ich
is
p
r
ec
is
e
ly
th
e
f
o
cu
s
o
f
o
u
r
wo
r
k
.
T
o
v
iew
th
e
r
esu
lts
as
liv
e
m
o
d
els,
ac
ce
s
s
th
ese:
in
d
→b
b
c
[
2
5
]
,
b
b
c→in
d
[
2
6
]
.
T
h
e
co
d
e
in
clu
d
in
g
th
e
im
p
lem
en
tatio
n
d
etails ca
n
b
e
f
o
u
n
d
at
[
1
2
]
.
T
ab
le
9
.
C
o
m
p
a
r
is
o
n
o
f
o
u
r
m
o
d
el
in
b
b
c↔in
d
with
p
r
ev
io
u
s
s
tu
d
ies
S
t
u
d
y
b
b
c
→
i
n
d
S
a
c
r
e
B
LEU
(
h
i
g
h
e
r
i
s
b
e
t
t
e
r
)
i
n
d
→
b
b
c
S
a
c
r
e
B
LEU
(
h
i
g
h
e
r
i
s
b
e
t
t
e
r
)
Th
e
b
a
se
l
i
n
e
mo
d
e
l
u
se
d
w
h
i
c
h
a
c
h
i
e
v
e
s
t
h
e
h
i
g
h
e
st
S
a
c
r
e
B
LEU
(
b
b
c
→
i
n
d
)
Th
e
b
a
se
l
i
n
e
mo
d
e
l
u
se
d
w
h
i
c
h
a
c
h
i
e
v
e
s
t
h
e
h
i
g
h
e
st
S
a
c
r
e
B
LEU
(
i
n
d
→
b
b
c
)
[
1
4
]
2
0
.
9
4
1
8
.
4
1
P
B
S
M
T
I
n
d
o
B
A
R
T
v
2
[
2
4
]
2
0
.
4
8
1
8
.
4
1
mT5
I
n
d
o
B
A
R
T
v
2
O
u
r
s
3
7
.
1
0
3
0
.
8
4
N
LLB
-
200
N
LLB
-
200
5.
CO
NCLU
SI
O
N
B
ased
o
n
o
u
r
f
in
d
in
g
s
,
u
s
in
g
a
lar
g
e
p
r
e
-
tr
ai
n
ed
lan
g
u
ag
e
m
o
d
el
lik
e
th
e
NL
L
B
-
2
0
0
as
a
b
aselin
e
p
r
o
v
es
to
b
e
ef
f
ec
tiv
e
f
o
r
L
R
L
s
tr
an
s
latio
n
,
s
p
ec
if
icall
y
f
o
r
b
b
c↔in
d
p
air
s
,
ev
e
n
f
o
r
en
tire
ly
u
n
s
ee
n
lan
g
u
ag
es.
Fo
r
ex
am
p
le,
s
o
m
e
o
f
o
u
r
ev
alu
atio
n
s
s
h
o
w
im
p
r
e
s
s
iv
e
g
ain
s
in
s
ac
r
eBLE
U
s
co
r
es,
r
ea
ch
in
g
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
8
1
4
I
n
t J Ad
v
Ap
p
l Sci
,
Vo
l.
1
3
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
834
-
8
4
3
842
+4
6
.
7
8
(
b
b
c→in
d
)
an
d
+
3
0
.
2
9
(
in
d
→b
b
c
)
.
Ho
wev
er
,
tr
a
n
s
latin
g
to
/f
r
o
m
en
tire
ly
u
n
s
ee
n
lan
g
u
ag
es
in
ze
r
o
-
s
h
o
t
s
ettin
g
s
r
ev
ea
led
a
cr
u
cial
lim
itatio
n
o
f
th
e
NL
L
B
-
2
0
0
m
o
d
el:
g
e
n
er
atin
g
c
o
r
r
ec
t
tr
an
s
latio
n
s
in
th
e
wr
o
n
g
lan
g
u
ag
e.
W
h
ile
p
r
e
-
tr
ain
ed
m
o
d
els
h
o
ld
p
r
o
m
is
e
f
o
r
ze
r
o
-
s
h
o
t
tr
an
s
latio
n
,
ch
allen
g
es
r
em
ain
,
esp
ec
ially
f
o
r
u
n
s
ee
n
lan
g
u
ag
es
an
d
d
o
m
ain
s
.
Fu
r
th
er
r
esear
ch
is
n
ec
ess
ar
y
to
o
v
er
co
m
e
th
ese
h
u
r
d
les
an
d
en
h
an
ce
p
r
e
-
tr
ain
e
d
m
o
d
el
ef
f
ec
tiv
en
ess
in
ze
r
o
-
s
h
o
t tr
a
n
s
latio
n
.
RE
F
E
R
E
NC
E
S
[
1
]
U
N
E
S
C
O
W
A
L
,
“
B
a
t
a
k
T
o
b
a
,
”
U
N
E
S
C
O
.
A
c
c
e
s
s
e
d
:
F
e
b
.
2
5
,
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
e
n
.
w
a
l
.
u
n
e
s
c
o
.
o
r
g
/
l
a
n
g
u
a
g
e
s
/
b
a
t
a
k
-
t
o
b
a
[
2
]
H
.
V
.
S
,
S
u
mars
i
h
,
a
n
d
R
.
H
u
se
i
n
,
“
T
o
b
a
B
a
t
a
k
l
a
n
g
u
a
g
e
s
h
i
f
t
i
n
R
a
n
t
a
u
S
e
l
a
t
a
n
,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
T
h
e
4
t
h
An
n
u
a
l
I
n
t
e
r
n
a
t
i
o
n
a
l
S
e
m
i
n
a
r
o
n
T
ra
n
s
f
o
rm
a
t
i
v
e
E
d
u
c
a
t
i
o
n
a
n
d
Ed
u
c
a
t
i
o
n
a
l
L
e
a
d
e
rs
h
i
p
(
AI
S
T
E
EL)
,
M
e
d
a
n
:
EPr
i
n
t
s,
2
0
1
9
.
[
3
]
E.
S
i
n
a
m
b
e
l
a
a
n
d
L
.
L.
H
u
t
a
g
a
o
l
,
“
I
d
i
o
m
a
t
i
c
t
r
a
n
s
l
a
t
i
o
n
o
f
u
m
p
a
s
a
i
n
d
e
l
i
v
e
r
i
n
g
u
l
o
s
i
n
T
o
b
a
B
a
t
a
k
w
e
d
d
i
n
g
c
e
r
e
m
o
n
y
,
”
Ex
p
l
o
ra
,
v
o
l
.
9
,
n
o
.
1
,
p
p
.
6
7
–
8
7
,
A
p
r
.
2
0
2
3
,
d
o
i
:
1
0
.
5
1
6
2
2
/
e
x
p
l
o
r
a
.
v
9
i
1
.
1
2
4
3
.
[
4
]
J.
G
u
,
H
.
H
a
ssa
n
,
J
.
D
e
v
l
i
n
,
a
n
d
V
.
O
.
K
.
L
i
,
“
U
n
i
v
e
r
sa
l
n
e
u
r
a
l
ma
c
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
f
o
r
e
x
t
r
e
me
l
y
l
o
w
r
e
s
o
u
r
c
e
l
a
n
g
u
a
g
e
s,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
1
8
C
o
n
f
e
re
n
c
e
o
f
t
h
e
N
o
r
t
h
Am
e
r
i
c
a
n
C
h
a
p
t
e
r
o
f
t
h
e
Ass
o
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
st
i
c
s:
H
u
m
a
n
L
a
n
g
u
a
g
e
T
e
c
h
n
o
l
o
g
i
e
s,
V
o
l
u
m
e
1
(
L
o
n
g
Pa
p
e
rs)
,
S
t
r
o
u
d
s
b
u
r
g
,
P
A
,
U
S
A
:
A
sso
c
i
a
t
i
o
n
f
o
r
C
o
mp
u
t
a
t
i
o
n
a
l
Li
n
g
u
i
s
t
i
c
s,
2
0
1
8
,
p
p
.
3
4
4
–
3
5
4
.
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
N
1
8
-
1
0
3
2
.
[
5
]
R
.
S
e
n
n
r
i
c
h
a
n
d
B
.
Zh
a
n
g
,
“
R
e
v
i
s
i
t
i
n
g
l
o
w
-
r
e
s
o
u
r
c
e
n
e
u
r
a
l
mac
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
:
a
c
a
s
e
s
t
u
d
y
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
5
7
t
h
A
n
n
u
a
l
Me
e
t
i
n
g
o
f
t
h
e
Ass
o
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
st
i
c
s
,
S
t
r
o
u
d
s
b
u
r
g
,
P
A
,
U
S
A
:
A
sso
c
i
a
t
i
o
n
f
o
r
C
o
mp
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s
,
2
0
1
9
,
p
p
.
2
1
1
–
2
2
1
.
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
P
1
9
-
1
0
2
1
.
[
6
]
M
.
Ta
r
s
,
T.
P
u
r
a
so
n
,
a
n
d
A
.
Tä
t
t
a
r
,
“
Te
a
c
h
i
n
g
u
n
s
e
e
n
l
o
w
-
r
e
s
o
u
r
c
e
l
a
n
g
u
a
g
e
s
t
o
l
a
r
g
e
t
r
a
n
sl
a
t
i
o
n
mo
d
e
l
s,
”
i
n
P
ro
c
e
e
d
i
n
g
s
o
f
t
h
e
S
e
v
e
n
t
h
C
o
n
f
e
r
e
n
c
e
o
n
Ma
c
h
i
n
e
T
ra
n
s
l
a
t
i
o
n
(
WM
T
)
,
A
b
u
D
h
a
b
i
,
U
n
i
t
e
d
A
r
a
b
Emi
r
a
t
e
s
:
A
sso
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
Li
n
g
u
i
s
t
i
c
s,
2
0
2
2
,
p
p
.
3
7
5
–
3
8
0
.
[
7
]
M
.
Ta
r
s,
A
.
Tä
t
t
a
r
,
a
n
d
M
.
F
i
še
l
,
“
Ex
t
r
e
me
l
y
l
o
w
-
r
e
so
u
r
c
e
ma
c
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
f
o
r
c
l
o
sel
y
r
e
l
a
t
e
d
l
a
n
g
u
a
g
e
s
,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
2
3
rd
N
o
r
d
i
c
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s
(
N
o
D
a
L
i
D
a
)
,
R
e
y
k
j
a
v
i
k
,
I
c
e
l
a
n
d
:
Li
n
k
ö
p
i
n
g
U
n
i
v
e
r
si
t
y
E
l
e
c
t
r
o
n
i
c
P
r
e
ss,
S
w
e
d
e
n
,
2
0
2
1
,
p
p
.
4
1
–
52
.
[
8
]
T.
K
o
c
mi
a
n
d
O
.
B
o
j
a
r
,
“
Tr
i
v
i
a
l
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
f
o
r
l
o
w
-
r
e
s
o
u
r
c
e
n
e
u
r
a
l
m
a
c
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
T
h
i
rd
C
o
n
f
e
re
n
c
e
o
n
M
a
c
h
i
n
e
T
r
a
n
s
l
a
t
i
o
n
:
Re
se
a
rc
h
Pa
p
e
rs
,
S
t
r
o
u
d
s
b
u
r
g
,
P
A
,
U
S
A
:
A
sso
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
st
i
c
s
,
2
0
1
8
,
p
p
.
2
4
4
–
2
5
2
.
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
W
1
8
-
6
3
2
5
.
[
9
]
N
LLB
T
e
a
m
e
t
a
l
.
,
“
N
o
l
a
n
g
u
a
g
e
l
e
f
t
b
e
h
i
n
d
:
sc
a
l
i
n
g
h
u
ma
n
-
c
e
n
t
e
r
e
d
mac
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
,
”
Ju
l
.
2
0
2
2
,
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
a
r
x
i
v
.
o
r
g
/
a
b
s/
2
2
0
7
.
0
4
6
7
2
[
1
0
]
A
.
F
a
n
e
t
a
l
.
,
“
B
e
y
o
n
d
E
n
g
l
i
sh
-
c
e
n
t
r
i
c
mu
l
t
i
l
i
n
g
u
a
l
m
a
c
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
,
”
J
o
u
r
n
a
l
o
f
M
a
c
h
i
n
e
L
e
a
rn
i
n
g
R
e
se
a
rc
h
,
v
o
l
.
2
2
,
p
p
.
1
–
4
8
,
2
0
2
1
.
[
1
1
]
H
.
W
a
n
g
,
S
.
M
a
,
L.
D
o
n
g
,
S
.
H
u
a
n
g
,
D
.
Z
h
a
n
g
,
a
n
d
F
.
W
e
i
,
“
D
e
e
p
N
e
t
:
sca
l
i
n
g
t
r
a
n
sf
o
r
mers
t
o
1
,
0
0
0
La
y
e
r
s,
”
i
n
I
E
EE
T
ra
n
s
a
c
t
i
o
n
s
o
n
P
a
t
t
e
r
n
A
n
a
l
y
si
s
a
n
d
Ma
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
2
0
2
4
,
p
p
.
1
–
14
,
d
o
i
:
1
0
.
1
1
0
9
/
TPA
M
I
.
2
0
2
4
.
3
3
8
6
9
2
7
.
[
1
2
]
G
i
t
H
u
b
,
“
B
a
t
a
k
T
o
b
a
l
a
n
g
u
a
g
e
-
I
n
d
o
n
e
si
a
n
mac
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
w
i
t
h
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
u
s
i
n
g
N
o
La
n
g
u
a
g
e
Le
f
t
B
e
h
i
n
d
,
”
G
i
t
H
u
b
,
I
n
c
.
A
c
c
e
sse
d
:
F
e
b
.
2
5
,
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
g
i
t
h
u
b
.
c
o
m/
c
a
f
f
e
i
n
e
e
e
e
/
b
a
t
a
k
_
t
o
b
a
_
i
n
d
o
n
e
si
a
n
_
n
mt
[
1
3
]
H
u
g
g
i
n
g
F
a
c
e
,
“
D
a
t
a
s
e
t
s
:
i
n
d
o
n
l
p
/
n
u
s
a
t
r
a
n
s
l
a
t
i
o
n
_
m
t
,
”
H
u
g
g
i
n
g
F
a
c
e
.
A
c
c
e
ss
e
d
:
F
e
b
.
2
4
,
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
h
u
g
g
i
n
g
f
a
c
e
.
c
o
/
d
a
t
a
s
e
t
s
/
i
n
d
o
n
l
p
/
n
u
sat
r
a
n
s
l
a
t
i
o
n
_
m
t
[
1
4
]
G
.
I
.
W
i
n
a
t
a
e
t
a
l
.
,
“
N
u
s
a
X
:
mu
l
t
i
l
i
n
g
u
a
l
p
a
r
a
l
l
e
l
s
e
n
t
i
me
n
t
d
a
t
a
set
f
o
r
1
0
I
n
d
o
n
e
s
i
a
n
l
o
c
a
l
l
a
n
g
u
a
g
e
s
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s o
f
t
h
e
1
7
t
h
C
o
n
f
e
re
n
c
e
o
f
t
h
e
E
u
r
o
p
e
a
n
C
h
a
p
t
e
r
o
f
t
h
e
Asso
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
st
i
c
s
,
S
t
r
o
u
d
sb
u
r
g
,
P
A
,
U
S
A
:
A
sso
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s,
2
0
2
3
,
p
p
.
8
1
5
–
8
3
4
.
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
3
.
e
a
c
l
-
mai
n
.
5
7
.
[
1
5
]
A
.
P
u
r
w
a
r
i
a
n
t
i
a
n
d
I
.
A
.
P
.
A
.
C
r
i
s
d
a
y
a
n
t
i
,
“
I
mp
r
o
v
i
n
g
B
i
-
LSTM
p
e
r
f
o
r
m
a
n
c
e
f
o
r
I
n
d
o
n
e
si
a
n
s
e
n
t
i
me
n
t
a
n
a
l
y
s
i
s
u
s
i
n
g
p
a
r
a
g
r
a
p
h
v
e
c
t
o
r
,
”
i
n
2
0
1
9
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
f
A
d
v
a
n
c
e
d
I
n
f
o
rm
a
t
i
c
s:
C
o
n
c
e
p
t
s
,
T
h
e
o
ry
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
(
I
C
AI
C
T
A)
,
I
EEE,
S
e
p
.
2
0
1
9
,
p
p
.
1
–
5
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
A
I
C
TA
.
2
0
1
9
.
8
9
0
4
1
9
9
.
[
1
6
]
B
.
W
i
l
i
e
e
t
a
l
.
,
“
I
n
d
o
N
L
U
:
b
e
n
c
h
m
a
r
k
a
n
d
r
e
s
o
u
r
c
e
s
f
o
r
e
v
a
l
u
a
t
i
n
g
I
n
d
o
n
e
s
i
a
n
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
u
n
d
e
r
s
t
a
n
d
i
n
g
,
”
i
n
P
r
o
c
e
e
d
i
n
g
s
o
f
t
h
e
1
s
t
C
o
n
f
e
r
e
n
c
e
o
f
t
h
e
A
s
i
a
-
P
a
c
i
f
i
c
C
h
a
p
t
e
r
o
f
t
h
e
A
s
s
o
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s
a
n
d
t
h
e
1
0
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
i
n
t
C
o
n
f
e
r
e
n
c
e
o
n
N
a
t
u
r
a
l
L
a
n
g
u
a
g
e
P
r
o
c
e
s
s
i
n
g
,
S
u
z
h
o
u
,
C
h
i
n
a
:
A
s
s
o
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s
,
2
0
2
0
,
p
p
.
8
4
3
–
8
5
7
.
[
1
7
]
L.
C
.
T
i
mp
e
,
“
A
n
o
n
l
i
n
e
G
o
o
g
l
e
C
o
l
a
b
p
r
o
j
e
c
t
f
o
r
e
x
p
l
o
r
i
n
g
t
h
e
S
A
R
S
C
o
V
-
2
g
e
n
o
m
e
a
n
d
mRNA
v
a
c
c
i
n
e
s,
”
Bi
o
c
h
e
m
i
s
t
ry
a
n
d
Mo
l
e
c
u
l
a
r
Bi
o
l
o
g
y
E
d
u
c
a
t
i
o
n
,
v
o
l
.
5
1
,
n
o
.
2
,
p
p
.
2
0
9
–
2
1
1
,
M
a
r
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
2
/
b
mb
.
2
1
7
1
1
.
[
1
8
]
T.
W
o
l
f
e
t
a
l
.
,
“
Tr
a
n
sf
o
r
m
e
r
s:
s
t
a
t
e
-
of
-
t
h
e
-
a
r
t
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
ssi
n
g
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
2
0
C
o
n
f
e
re
n
c
e
o
n
Em
p
i
r
i
c
a
l
Me
t
h
o
d
s
i
n
N
a
t
u
ra
l
L
a
n
g
u
a
g
e
Pr
o
c
e
ssi
n
g
:
S
y
s
t
e
m
D
e
m
o
n
s
t
ra
t
i
o
n
s
,
S
t
r
o
u
d
s
b
u
r
g
,
P
A
,
U
S
A
:
A
ss
o
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
Li
n
g
u
i
s
t
i
c
s,
2
0
2
0
,
p
p
.
3
8
–
4
5
.
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
2
0
2
0
.
e
m
n
l
p
-
d
e
m
o
s.
6
.
[
1
9
]
M
.
P
o
s
t
,
“
A
c
a
l
l
f
o
r
c
l
a
r
i
t
y
i
n
r
e
p
o
r
t
i
n
g
B
L
E
U
s
c
o
r
e
s
,
”
i
n
P
r
o
c
e
e
d
i
n
g
s
o
f
t
h
e
T
h
i
r
d
C
o
n
f
e
r
e
n
c
e
o
n
M
a
c
h
i
n
e
T
r
a
n
s
l
a
t
i
o
n
:
R
e
s
e
a
r
c
h
P
a
p
e
r
s
,
S
t
r
o
u
d
s
b
u
r
g
,
P
A
,
U
S
A
:
A
s
s
o
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
L
i
n
g
u
i
s
t
i
c
s
,
2
0
1
8
,
p
p
.
1
8
6
–
1
9
1
.
d
o
i
:
1
0
.
1
8
6
5
3
/
v
1
/
W
1
8
-
6
3
1
9
.
[
2
0
]
K
.
G
u
p
t
a
,
“
M
A
L
M
:
mi
x
i
n
g
a
u
g
m
e
n
t
e
d
l
a
n
g
u
a
g
e
m
o
d
e
l
i
n
g
f
o
r
z
e
r
o
-
sh
o
t
m
a
c
h
i
n
e
t
r
a
n
s
l
a
t
i
o
n
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
2
n
d
I
n
t
e
r
n
a
t
i
o
n
a
l
W
o
rks
h
o
p
o
n
N
a
t
u
r
a
l
L
a
n
g
u
a
g
e
Pr
o
c
e
ssi
n
g
f
o
r
D
i
g
i
t
a
l
H
u
m
a
n
i
t
i
e
s
,
T
a
i
p
e
i
,
Ta
i
w
a
n
:
A
sso
c
i
a
t
i
o
n
f
o
r
C
o
m
p
u
t
a
t
i
o
n
a
l
Li
n
g
u
i
s
t
i
c
s,
2
0
2
2
,
p
p
.
5
3
–
58
.
[
2
1
]
I
.
Lo
sh
c
h
i
l
o
v
a
n
d
F
.
H
u
t
t
e
r
,
“
D
e
c
o
u
p
l
e
d
w
e
i
g
h
t
d
e
c
a
y
r
e
g
u
l
a
r
i
z
a
t
i
o
n
,
”
N
o
v
.
2
0
1
7
,
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
a
r
x
i
v
.
o
r
g
/
a
b
s/
1
7
1
1
.
0
5
1
0
1
.
[
2
2
]
E.
A
.
P
l
a
t
a
n
i
o
s,
O
.
S
t
r
e
t
c
u
,
G
.
N
e
u
b
i
g
,
B
.
P
o
c
z
o
s,
a
n
d
T.
M
.
M
i
t
c
h
e
l
l
,
“
C
o
m
p
e
t
e
n
c
e
-
b
a
se
d
c
u
r
r
i
c
u
l
u
m
l
e
a
r
n
i
n
g
f
o
r
n
e
u
r
a
l
mac
h
i
n
e
t
r
a
n
sl
a
t
i
o
n
,
”
M
a
r
.
2
0
1
9
,
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
a
r
x
i
v
.
o
r
g
/
a
b
s
/
1
9
0
3
.
0
9
8
4
8
[
2
3
]
S
.
S
a
msi
,
M
.
J
o
n
e
s,
a
n
d
M
.
M
.
V
e
i
l
l
e
t
t
e
,
“
C
o
m
p
u
t
e
,
t
i
me
a
n
d
e
n
e
r
g
y
c
h
a
r
a
c
t
e
r
i
z
a
t
i
o
n
o
f
e
n
c
o
d
e
r
-
d
e
c
o
d
e
r
n
e
t
w
o
r
k
s
w
i
t
h
a
u
t
o
m
a
t
i
c
m
i
x
e
d
p
r
e
c
i
si
o
n
t
r
a
i
n
i
n
g
,
”
i
n
2
0
2
0
I
EE
E
H
i
g
h
P
e
rf
o
rm
a
n
c
e
Ex
t
re
m
e
C
o
m
p
u
t
i
n
g
C
o
n
f
e
r
e
n
c
e
(
H
PE
C
)
,
I
EEE,
S
e
p
.
2
0
2
0
,
p
p
.
1
–
6
.
d
o
i
:
1
0
.
1
1
0
9
/
H
P
E
C
4
3
6
7
4
.
2
0
2
0
.
9
2
8
6
2
4
1
.
[
2
4
]
W
.
W
o
n
g
s
o
,
A
.
J
o
y
o
a
d
i
k
u
su
mo
,
B
.
S
.
B
u
a
n
a
,
a
n
d
D
.
S
u
h
a
r
t
o
n
o
,
“
M
a
n
y
-
to
-
man
y
mu
l
t
i
l
i
n
g
u
a
l
t
r
a
n
sl
a
t
i
o
n
mo
d
e
l
f
o
r
l
a
n
g
u
a
g
e
s
o
f
I
n
d
o
n
e
si
a
,
”
I
EEE
A
c
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
9
1
3
8
5
–
9
1
3
9
7
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
3
0
8
8
1
8
.
[
2
5
]
C
.
S
a
m
u
e
l
,
“
i
n
d
o
n
e
s
i
a
n
_
t
o
_
b
a
t
a
k
n
e
se_
t
r
a
n
sl
a
t
i
o
n
_
m
o
d
e
l
,
”
H
u
g
g
i
n
g
F
a
c
e
.
A
c
c
e
ss
e
d
:
F
e
b
.
2
4
,
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
h
u
g
g
i
n
g
f
a
c
e
.
c
o
/
k
e
p
i
n
sa
m/
i
nd
-
to
-
b
b
c
-
n
mt
-
v1
[
2
6
]
C
.
S
a
m
u
e
l
,
“
b
a
t
a
k
n
e
s
e
_
t
o
_
i
n
d
o
n
e
si
a
n
_
t
r
a
n
s
l
a
t
i
o
n
_
m
o
d
e
l
,
”
H
u
g
g
i
n
g
F
a
c
e
.
A
c
c
e
ss
e
d
:
F
e
b
.
2
4
,
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
h
u
g
g
i
n
g
f
a
c
e
.
c
o
/
k
e
p
i
n
sa
m/
b
b
c
-
to
-
i
n
d
-
n
m
t
-
v1
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ad
v
Ap
p
l Sci
I
SS
N:
2252
-
8
8
1
4
B
a
ta
k
To
b
a
la
n
g
u
a
g
e
-
I
n
d
o
n
es
ia
n
ma
ch
in
e
tr
a
n
s
la
tio
n
w
ith
tr
a
n
s
fer lea
r
n
in
g
u
s
in
g
n
o
…
(
C
ev
in
S
a
mu
el)
843
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Ce
v
in
S
a
m
u
e
l
is
a
n
In
f
o
r
m
a
ti
c
s
u
n
d
e
rg
ra
d
u
a
te
stu
d
e
n
t
fr
o
m
De
p
a
rtme
n
t
o
f
El
e
c
tri
c
a
l
En
g
in
e
e
rin
g
,
F
a
c
u
l
ty
o
f
En
g
i
n
e
e
rin
g
,
Un
iv
e
rsitas
Riau
,
P
e
k
a
n
b
a
ru
,
In
d
o
n
e
sia
.
He
is
sk
il
led
in
f
u
ll
-
sta
c
k
we
b
d
e
v
e
l
o
p
m
e
n
t
a
n
d
h
a
s
a
re
se
a
rc
h
in
tere
st
in
d
e
e
p
lea
rn
in
g
,
p
a
rti
c
u
larly
lan
g
u
a
g
e
m
o
d
e
ls.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
c
e
v
in
.
sa
m
u
e
l3
7
7
8
@s
t
u
d
e
n
t.
u
n
ri.
a
c
.
i
d
a
n
d
c
e
v
in
.
sa
m
u
e
l@y
a
h
o
o
.
c
o
m
.
Ir
sa
n
Ta
u
fi
k
Ali
is
a
lec
tu
re
r
a
t
th
e
De
p
a
rtme
n
t
o
f
E
lec
tri
c
a
l
En
g
i
n
e
e
rin
g
,
F
a
c
u
lt
y
o
f
E
n
g
i
n
e
e
rin
g
,
Un
i
v
e
rsitas
Riau
,
P
e
k
a
n
b
a
ru
,
I
n
d
o
n
e
sia
.
He
o
b
tain
e
d
h
is
Do
c
to
ra
te
i
n
In
fo
rm
a
ti
c
s
fro
m
Un
i
v
e
rsitas
I
n
d
o
n
e
sia
(UI),
In
d
o
n
e
sia
,
M
a
st
e
r’s
d
e
g
re
e
i
n
I
n
fo
rm
a
ti
o
n
Tec
h
n
o
l
o
g
y
fr
o
m
U
n
iv
e
rsitas
G
a
d
jah
M
a
d
a
(UG
M
),
In
d
o
n
e
sia
,
a
n
d
Ba
c
h
e
lo
r’s
d
e
g
re
e
in
In
fo
rm
a
ti
c
s
fr
o
m
Un
i
v
e
rsitas
Isla
m
In
d
o
n
e
sia
(
UII)
,
I
n
d
o
n
e
sia
.
Hi
s
re
se
a
rc
h
in
tere
st
i
n
c
lu
d
e
s
m
a
c
h
in
e
lea
rn
in
g
,
d
e
e
p
lea
rn
in
g
,
a
n
d
i
n
fo
rm
a
ti
o
n
tec
h
n
o
l
o
g
y
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
irsa
n
.
a
li
@le
c
tu
re
r.
u
n
ri.
a
c
.
id
.
Evaluation Warning : The document was created with Spire.PDF for Python.