H7.1 top features

Top feature 0 in H7.1: (feature 21707)

TOP ACTIVATIONS
MAX = 2.536

at
Token at
Feature activation+0.000
the
Token the
Feature activation+0.000
time
Token time
Feature activation+0.000
that
Token that
Feature activation+0.000
II
Token II
Feature activation+0.000
IR
TokenIR
Feature activation+2.536
's
Token's
Feature activation+1.274
claims
Token claims
Feature activation+0.000
to
Token to
Feature activation+0.084
the
Token the
Feature activation+0.112
trademark
Token trademark
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
Department
Token Department
Feature activation+1.136
of
Token of
Feature activation+0.000
Justice
Token Justice
Feature activation+2.201
appreci
Token appreci
Feature activation+0.000
ates
Tokenates
Feature activation+0.000
the
Token the
Feature activation+0.000
offer
Token offer
Feature activation+0.000
of
Token of
Feature activation+0.000
power
Token power
Feature activation+0.000
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.000
King
Token King
Feature activation+0.295
and
Token and
Feature activation+0.199
Queen
Token Queen
Feature activation+2.053
concoct
Token concoct
Feature activation+0.000
ed
Tokened
Feature activation+0.000
a
Token a
Feature activation+0.000
plan
Token plan
Feature activation+0.000
to
Token to
Feature activation+0.000
future
Token future
Feature activation+0.000
lawsuit
Token lawsuit
Feature activation+0.000
if
Token if
Feature activation+0.000
the
Token the
Feature activation+0.000
nud
Token nud
Feature activation+0.000
ists
Tokenists
Feature activation+2.044
could
Token could
Feature activation+0.000
prove
Token prove
Feature activation+0.000
that
Token that
Feature activation+0.000
the
Token the
Feature activation+0.000
law
Token law
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
KEY
TokenKEY
Feature activation+0.000
E
TokenE
Feature activation+0.986
TV
Token TV
Feature activation+2.023
asked
Token asked
Feature activation+0.000
the
Token the
Feature activation+0.000
judge
Token judge
Feature activation+0.000
,
Token,
Feature activation+0.000
was
Token was
Feature activation+0.000
instead
Token instead
Feature activation+0.000
,
Token,
Feature activation+0.000
it
Token it
Feature activation+0.000
was
Token was
Feature activation+0.000
everything
Token everything
Feature activation+0.000
they
Token they
Feature activation+1.654
wanted
Token wanted
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Trump
TokenTrump
Feature activation+0.000
and
Token and
Feature activation+0.000
Kerala
Token Kerala
Feature activation+0.000
police
Token police
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.103
person
Token person
Feature activation+1.582
is
Token is
Feature activation+1.123
asking
Token asking
Feature activation+0.000
its
Token its
Feature activation+0.000
operatives
Token operatives
Feature activation+0.000
to
Token to
Feature activation+0.000
also
Token also
Feature activation+0.000
noted
Token noted
Feature activation+0.000
that
Token that
Feature activation+0.000
the
Token the
Feature activation+0.000
Center
Token Center
Feature activation+0.457
is
Token is
Feature activation+1.492
funded
Token funded
Feature activation+0.000
by
Token by
Feature activation+0.000
many
Token many
Feature activation+0.000
prominent
Token prominent
Feature activation+0.000
conservative
Token conservative
Feature activation+0.000
#
Token #
Feature activation+0.000
SD
TokenSD
Feature activation+0.000
Live
TokenLive
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
When
TokenWhen
Feature activation+1.481
?
Token?
Feature activation+0.000
#
Token #
Feature activation+0.000
Surv
TokenSurv
Feature activation+0.000
iv
Tokeniv
Feature activation+0.000
or
Tokenor
Feature activation+0.000
the
Token the
Feature activation+0.000
lines
Token lines
Feature activation+0.000
.
Token.
Feature activation+0.000
Chand
Token Chand
Feature activation+0.000
rab
Tokenrab
Feature activation+0.000
han
Tokenhan
Feature activation+1.463
is
Token is
Feature activation+0.000
insin
Token insin
Feature activation+0.000
uating
Tokenuating
Feature activation+0.000
that
Token that
Feature activation+0.000
he
Token he
Feature activation+0.000
the
Token the
Feature activation+0.337
film
Token film
Feature activation+0.000
shows
Token shows
Feature activation+0.000
us
Token us
Feature activation+0.000
that
Token that
Feature activation+0.000
he
Token he
Feature activation+1.460
knows
Token knows
Feature activation+0.000
who
Token who
Feature activation+0.000
the
Token the
Feature activation+0.000
important
Token important
Feature activation+0.000
people
Token people
Feature activation+0.000
has
Token has
Feature activation+0.000
also
Token also
Feature activation+0.000
revealed
Token revealed
Feature activation+0.000
that
Token that
Feature activation+0.048
federal
Token federal
Feature activation+0.492
agencies
Token agencies
Feature activation+1.422
,
Token,
Feature activation+0.000
including
Token including
Feature activation+0.000
security
Token security
Feature activation+0.000
services
Token services
Feature activation+0.000
,
Token,
Feature activation+0.000
,
Token,
Feature activation+0.000
it
Token it
Feature activation+0.000
was
Token was
Feature activation+0.000
like
Token like
Feature activation+0.000
âĢ¦
Token âĢ¦
Feature activation+0.000
he
Token he
Feature activation+1.349
just
Token just
Feature activation+1.046
,
Token,
Feature activation+0.000
I
Token I
Feature activation+0.000
don
Token don
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
yet
Token yet
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Par
TokenPar
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+1.327
,
Token,
Feature activation+0.141
who
Token who
Feature activation+0.000
has
Token has
Feature activation+0.000
been
Token been
Feature activation+0.000
batting
Token batting
Feature activation+0.000
T
TokenT
Feature activation+0.000
rying
Tokenrying
Feature activation+0.000
to
Token to
Feature activation+0.000
dismiss
Token dismiss
Feature activation+0.000
the
Token the
Feature activation+0.000
journalist
Token journalist
Feature activation+1.286
he
Token he
Feature activation+0.000
continued
Token continued
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
this
Tokenthis
Feature activation+0.000
the
Token the
Feature activation+0.000
time
Token time
Feature activation+0.000
that
Token that
Feature activation+0.000
II
Token II
Feature activation+0.000
IR
TokenIR
Feature activation+2.536
's
Token's
Feature activation+1.274
claims
Token claims
Feature activation+0.000
to
Token to
Feature activation+0.084
the
Token the
Feature activation+0.112
trademark
Token trademark
Feature activation+0.000
were
Token were
Feature activation+0.000
House
Token House
Feature activation+0.000
of
Token of
Feature activation+0.000
Representatives
Token Representatives
Feature activation+0.000
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.000
President
Token President
Feature activation+1.274
's
Token's
Feature activation+0.978
chief
Token chief
Feature activation+0.000
critic
Token critic
Feature activation+0.000
complained
Token complained
Feature activation+0.000
of
Token of
Feature activation+0.000
mach
Token mach
Feature activation+0.000
o
Tokeno
Feature activation+0.000
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.000
Super
Token Super
Feature activation+0.000
hero
Tokenhero
Feature activation+1.271
fails
Token fails
Feature activation+0.091
to
Token to
Feature activation+0.000
see
Token see
Feature activation+0.000
that
Token that
Feature activation+0.000
work
Token work
Feature activation+0.000
Mr
Token Mr
Feature activation+0.000
.
Token.
Feature activation+0.000
Dah
Token Dah
Feature activation+0.000
mer
Tokenmer
Feature activation+0.000
's
Token's
Feature activation+0.000
urges
Token urges
Feature activation+1.245
to
Token to
Feature activation+0.000
have
Token have
Feature activation+0.000
sex
Token sex
Feature activation+0.000
with
Token with
Feature activation+0.000
the
Token the
Feature activation+0.000
said
Token said
Feature activation+0.000
"
Token "
Feature activation+0.000
NO
TokenNO
Feature activation+0.000
"
Token"
Feature activation+0.000
and
Token and
Feature activation+0.000
he
Token he
Feature activation+1.212
dropped
Token dropped
Feature activation+0.000
out
Token out
Feature activation+0.000
(
Token (
Feature activation+0.000
said
Tokensaid
Feature activation+0.000
he
Token he
Feature activation+0.000

Top DFA by src position
MAX = 2.744

Google
Token Google
Feature activation+0.003
Top resid features:
said
Token said
Feature activation-0.011
Top resid features:
II
Token II
Feature activation-0.003
Top resid features:
IR
TokenIR
Feature activation+0.002
Top resid features:
had
Token had
Feature activation+0.016
Top resid features:
asked
Token asked
Feature activation+2.702
Top resid features:
for
Token for
Feature activation+0.355
Top resid features:
an
Token an
Feature activation+0.004
Top resid features:
"
Token "
Feature activation-0.002
Top resid features:
ex
Tokenex
Feature activation-0.001
Top resid features:
hor
Tokenhor
Feature activation-0.000
Top resid features:
The
TokenThe
Feature activation+0.001
Top resid features:
Department
Token Department
Feature activation+0.002
Top resid features:
of
Token of
Feature activation-0.001
Top resid features:
Justice
Token Justice
Feature activation-0.001
Top resid features:
recently
Token recently
Feature activation+0.026
Top resid features:
requested
Token requested
Feature activation+2.713
Top resid features:
documents
Token documents
Feature activation+0.017
Top resid features:
from
Token from
Feature activation-0.004
Top resid features:
the
Token the
Feature activation-0.010
Top resid features:
Senate
Token Senate
Feature activation+0.000
Top resid features:
Judiciary
Token Judiciary
Feature activation+0.001
Top resid features:
Ferdinand
Token Ferdinand
Feature activation-0.009
Top resid features:
and
Token and
Feature activation+0.036
Top resid features:
Queen
Token Queen
Feature activation+0.007
Top resid features:
Is
Token Is
Feature activation+0.013
Top resid features:
abella
Tokenabella
Feature activation+0.017
Top resid features:
wanted
Token wanted
Feature activation+2.333
Top resid features:
to
Token to
Feature activation+0.050
Top resid features:
do
Token do
Feature activation+0.051
Top resid features:
away
Token away
Feature activation-0.003
Top resid features:
with
Token with
Feature activation+0.000
Top resid features:
many
Token many
Feature activation+0.003
Top resid features:
exactly
Token exactly
Feature activation-0.002
Top resid features:
what
Token what
Feature activation-0.003
Top resid features:
the
Token the
Feature activation+0.002
Top resid features:
nud
Token nud
Feature activation-0.005
Top resid features:
ists
Tokenists
Feature activation+0.008
Top resid features:
wanted
Token wanted
Feature activation+2.704
Top resid features:
.
Token.
Feature activation+0.061
Top resid features:
U
Token U
Feature activation-0.001
Top resid features:
.
Token.
Feature activation-0.000
Top resid features:
S
TokenS
Feature activation-0.000
Top resid features:
.
Token.
Feature activation+0.000
Top resid features:
Ċ
TokenĊ
Feature activation-0.004
Top resid features:
But
TokenBut
Feature activation-0.013
Top resid features:
KEY
Token KEY
Feature activation-0.005
Top resid features:
E
TokenE
Feature activation+0.002
Top resid features:
TV
Token TV
Feature activation-0.071
Top resid features:
wanted
Token wanted
Feature activation+2.744
Top resid features:
to
Token to
Feature activation+0.028
Top resid features:
know
Token know
Feature activation+0.145
Top resid features:
if
Token if
Feature activation-0.015
Top resid features:
Judge
Token Judge
Feature activation-0.003
Top resid features:
Brady
Token Brady
Feature activation+0.000
Top resid features:
long
Token long
Feature activation-0.001
Top resid features:
been
Token been
Feature activation-0.002
Top resid features:
united
Token united
Feature activation+0.001
Top resid features:
in
Token in
Feature activation+0.016
Top resid features:
their
Token their
Feature activation-0.001
Top resid features:
demands
Token demands
Feature activation+1.270
Top resid features:
for
Token for
Feature activation+0.914
Top resid features:
a
Token a
Feature activation+0.043
Top resid features:
lower
Token lower
Feature activation+0.003
Top resid features:
tax
Token tax
Feature activation+0.000
Top resid features:
burden
Token burden
Feature activation+0.002
Top resid features:
police
Token police
Feature activation-0.000
Top resid features:
,
Token,
Feature activation-0.002
Top resid features:
a
Token a
Feature activation-0.003
Top resid features:
person
Token person
Feature activation+0.005
Top resid features:
is
Token is
Feature activation+0.096
Top resid features:
asking
Token asking
Feature activation+2.199
Top resid features:
its
Token its
Feature activation+0.054
Top resid features:
operatives
Token operatives
Feature activation-0.001
Top resid features:
to
Token to
Feature activation-0.009
Top resid features:
carry
Token carry
Feature activation-0.003
Top resid features:
out
Token out
Feature activation-0.002
Top resid features:
Mack
Token Mack
Feature activation+0.000
Top resid features:
in
Tokenin
Feature activation+0.000
Top resid features:
ack
Tokenack
Feature activation-0.001
Top resid features:
Center
Token Center
Feature activation-0.015
Top resid features:
is
Token is
Feature activation-0.005
Top resid features:
demanding
Token demanding
Feature activation+2.185
Top resid features:
the
Token the
Feature activation+0.016
Top resid features:
emails
Token emails
Feature activation+0.006
Top resid features:
of
Token of
Feature activation+0.002
Top resid features:
anybody
Token anybody
Feature activation-0.004
Top resid features:
who
Token who
Feature activation-0.003
Top resid features:
whenever
Token whenever
Feature activation-0.025
Top resid features:
I
Token I
Feature activation-0.000
Top resid features:
âĢ
TokenâĢ
Feature activation-0.016
Top resid features:
Ļ
TokenĻ
Feature activation+0.005
Top resid features:
m
Tokenm
Feature activation-0.028
Top resid features:
needed
Token needed
Feature activation+1.880
Top resid features:
.
Token.
Feature activation+0.060
Top resid features:
Ċ
TokenĊ
Feature activation+0.028
Top resid features:
Ċ
TokenĊ
Feature activation+0.014
Top resid features:
Where
TokenWhere
Feature activation+0.006
Top resid features:
?
Token?
Feature activation+0.058
Top resid features:
,
Token,
Feature activation-0.003
Top resid features:
Chand
Token Chand
Feature activation-0.001
Top resid features:
rab
Tokenrab
Feature activation-0.004
Top resid features:
han
Tokenhan
Feature activation-0.006
Top resid features:
is
Token is
Feature activation+0.020
Top resid features:
asking
Token asking
Feature activation+2.062
Top resid features:
people
Token people
Feature activation-0.013
Top resid features:
to
Token to
Feature activation-0.004
Top resid features:
read
Token read
Feature activation-0.011
Top resid features:
between
Token between
Feature activation-0.002
Top resid features:
the
Token the
Feature activation-0.002
Top resid features:
al
Tokenal
Feature activation-0.010
Top resid features:
,
Token,
Feature activation-0.003
Top resid features:
âĢ
TokenâĢ
Feature activation+0.001
Top resid features:
Ļ
TokenĻ
Feature activation+0.001
Top resid features:
W
Token W
Feature activation-0.001
Top resid features:
ants
Tokenants
Feature activation+1.861
Top resid features:
More
Token More
Feature activation+0.014
Top resid features:
Such
Token Such
Feature activation+0.001
Top resid features:
Movies
Token Movies
Feature activation-0.006
Top resid features:
in
Token in
Feature activation-0.000
Top resid features:
China
Token China
Feature activation+0.001
Top resid features:
that
Token that
Feature activation-0.002
Top resid features:
federal
Token federal
Feature activation-0.000
Top resid features:
authorities
Token authorities
Feature activation+0.000
Top resid features:
already
Token already
Feature activation+0.011
Top resid features:
routinely
Token routinely
Feature activation-0.010
Top resid features:
ask
Token ask
Feature activation+1.232
Top resid features:
for
Token for
Feature activation+0.230
Top resid features:
âĢĶ
Token âĢĶ
Feature activation-0.003
Top resid features:
and
Token and
Feature activation-0.002
Top resid features:
receive
Token receive
Feature activation+0.473
Top resid features:
âĢĶ
Token âĢĶ
Feature activation+0.001
Top resid features:
like
Token like
Feature activation-0.005
Top resid features:
he
Token he
Feature activation+0.003
Top resid features:
really
Token really
Feature activation-0.002
Top resid features:
,
Token,
Feature activation-0.004
Top resid features:
really
Token really
Feature activation-0.001
Top resid features:
needed
Token needed
Feature activation+1.792
Top resid features:
this
Token this
Feature activation+0.162
Top resid features:
,
Token,
Feature activation+0.014
Top resid features:
âĢ
TokenâĢ
Feature activation+0.000
Top resid features:
Ŀ
TokenĿ
Feature activation+0.000
Top resid features:
she
Token she
Feature activation-0.000
Top resid features:
ar
Tokenar
Feature activation+0.012
Top resid features:
Par
Token Par
Feature activation-0.001
Top resid features:
rik
Tokenrik
Feature activation-0.001
Top resid features:
ar
Tokenar
Feature activation-0.000
Top resid features:
had
Token had
Feature activation+0.081
Top resid features:
directed
Token directed
Feature activation+1.799
Top resid features:
the
Token the
Feature activation+0.007
Top resid features:
three
Token three
Feature activation-0.003
Top resid features:
Service
Token Service
Feature activation+0.002
Top resid features:
Chiefs
Token Chiefs
Feature activation+0.002
Top resid features:
to
Token to
Feature activation+0.001
Top resid features:
prevention
Token prevention
Feature activation-0.000
Top resid features:
,
Token,
Feature activation-0.001
Top resid features:
when
Token when
Feature activation-0.003
Top resid features:
a
Token a
Feature activation+0.008
Top resid features:
journalist
Token journalist
Feature activation-0.021
Top resid features:
asked
Token asked
Feature activation+1.898
Top resid features:
him
Token him
Feature activation-0.000
Top resid features:
for
Token for
Feature activation+0.029
Top resid features:
comment
Token comment
Feature activation-0.020
Top resid features:
on
Token on
Feature activation-0.022
Top resid features:
the
Token the
Feature activation-0.003
Top resid features:
Google
Token Google
Feature activation+0.001
Top resid features:
said
Token said
Feature activation-0.003
Top resid features:
II
Token II
Feature activation-0.002
Top resid features:
IR
TokenIR
Feature activation-0.001
Top resid features:
had
Token had
Feature activation+0.004
Top resid features:
asked
Token asked
Feature activation+1.121
Top resid features:
for
Token for
Feature activation+0.657
Top resid features:
an
Token an
Feature activation+0.023
Top resid features:
"
Token "
Feature activation-0.005
Top resid features:
ex
Tokenex
Feature activation-0.005
Top resid features:
hor
Tokenhor
Feature activation-0.000
Top resid features:
granted
Token granted
Feature activation+0.006
Top resid features:
the
Token the
Feature activation+0.002
Top resid features:
authority
Token authority
Feature activation+0.101
Top resid features:
the
Token the
Feature activation+0.007
Top resid features:
President
Token President
Feature activation+0.023
Top resid features:
requested
Token requested
Feature activation+1.605
Top resid features:
in
Token in
Feature activation+0.027
Top resid features:
October
Token October
Feature activation-0.009
Top resid features:
.
Token.
Feature activation+0.002
Top resid features:
In
Token In
Feature activation-0.016
Top resid features:
the
Token the
Feature activation+0.001
Top resid features:
The
Token The
Feature activation+0.004
Top resid features:
Super
Token Super
Feature activation-0.002
Top resid features:
hero
Tokenhero
Feature activation-0.013
Top resid features:
also
Token also
Feature activation+0.030
Top resid features:
often
Token often
Feature activation+0.025
Top resid features:
expects
Token expects
Feature activation+1.694
Top resid features:
co
Token co
Feature activation-0.002
Top resid features:
-
Token-
Feature activation-0.001
Top resid features:
workers
Tokenworkers
Feature activation+0.001
Top resid features:
to
Token to
Feature activation+0.002
Top resid features:
not
Token not
Feature activation+0.002
Top resid features:
Mr
Token Mr
Feature activation+0.006
Top resid features:
.
Token.
Feature activation+0.002
Top resid features:
Dah
Token Dah
Feature activation+0.010
Top resid features:
mer
Tokenmer
Feature activation+0.011
Top resid features:
's
Token's
Feature activation+0.080
Top resid features:
urges
Token urges
Feature activation+1.630
Top resid features:
to
Token to
Feature activation+0.000
Top resid features:
have
Token have
Feature activation+0.000
Top resid features:
sex
Token sex
Feature activation+0.000
Top resid features:
with
Token with
Feature activation+0.000
Top resid features:
the
Token the
Feature activation+0.000
Top resid features:
Senator
TokenSenator
Feature activation+0.001
Top resid features:
Bob
Token Bob
Feature activation-0.002
Top resid features:
Corker
Token Corker
Feature activation-0.012
Top resid features:
"
Token "
Feature activation-0.001
Top resid features:
be
Tokenbe
Feature activation-0.008
Top resid features:
gged
Tokengged
Feature activation+1.657
Top resid features:
"
Token"
Feature activation+0.291
Top resid features:
me
Token me
Feature activation-0.011
Top resid features:
to
Token to
Feature activation-0.002
Top resid features:
endorse
Token endorse
Feature activation-0.011
Top resid features:
him
Token him
Feature activation-0.008
Top resid features:

Decoder Weights Distribution

Head 0: 0.04

Head 1: 0.51

Head 2: 0.02

Head 3: 0.05

Head 4: 0.03

Head 5: 0.06

Head 6: 0.07

Head 7: 0.04

Head 8: 0.04

Head 9: 0.06

Head 10: 0.06

Head 11: 0.02

Positive logits

requests3.64

demands3.56

request3.38

requesting3.37

demand3.25

orders3.24

requ3.20

demanding3.19

desires3.17

demanded3.17

desired3.11

Request3.07

Demand3.04

requested2.94

Request2.86

orders2.83

ordered2.83

request2.82

desire2.80

Desire2.78

Negative logits

VD-2.23

GC-2.17

tarn-2.08

UV-1.88

CG-1.87

ecd-1.86

-1.85

-1.84

hindsight-1.81

lineback-1.81

xit-1.80

derivatives-1.78

recomb-1.78

-1.77

dt-1.76

appearances-1.75

bleach-1.75

DX-1.75

fared-1.74

Coll-1.74

INTERVAL 2.282 - 2.536
CONTAINS 0.000%

at
Token at
Feature activation+0.000
the
Token the
Feature activation+0.000
time
Token time
Feature activation+0.000
that
Token that
Feature activation+0.000
II
Token II
Feature activation+0.000
IR
TokenIR
Feature activation+2.536
's
Token's
Feature activation+1.274
claims
Token claims
Feature activation+0.000
to
Token to
Feature activation+0.084
the
Token the
Feature activation+0.112
trademark
Token trademark
Feature activation+0.000

INTERVAL 2.029 - 2.282
CONTAINS 0.000%

âĢ
TokenâĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
Department
Token Department
Feature activation+1.136
of
Token of
Feature activation+0.000
Justice
Token Justice
Feature activation+2.201
appreci
Token appreci
Feature activation+0.000
ates
Tokenates
Feature activation+0.000
the
Token the
Feature activation+0.000
offer
Token offer
Feature activation+0.000
of
Token of
Feature activation+0.000
power
Token power
Feature activation+0.000
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.000
King
Token King
Feature activation+0.295
and
Token and
Feature activation+0.199
Queen
Token Queen
Feature activation+2.053
concoct
Token concoct
Feature activation+0.000
ed
Tokened
Feature activation+0.000
a
Token a
Feature activation+0.000
plan
Token plan
Feature activation+0.000
to
Token to
Feature activation+0.000
future
Token future
Feature activation+0.000
lawsuit
Token lawsuit
Feature activation+0.000
if
Token if
Feature activation+0.000
the
Token the
Feature activation+0.000
nud
Token nud
Feature activation+0.000
ists
Tokenists
Feature activation+2.044
could
Token could
Feature activation+0.000
prove
Token prove
Feature activation+0.000
that
Token that
Feature activation+0.000
the
Token the
Feature activation+0.000
law
Token law
Feature activation+0.000

INTERVAL 1.775 - 2.029
CONTAINS 0.000%

.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
KEY
TokenKEY
Feature activation+0.000
E
TokenE
Feature activation+0.986
TV
Token TV
Feature activation+2.023
asked
Token asked
Feature activation+0.000
the
Token the
Feature activation+0.000
judge
Token judge
Feature activation+0.000
,
Token,
Feature activation+0.000
was
Token was
Feature activation+0.000

INTERVAL 1.521 - 1.775
CONTAINS 0.000%

and
Token and
Feature activation+0.000
Kerala
Token Kerala
Feature activation+0.000
police
Token police
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.103
person
Token person
Feature activation+1.582
is
Token is
Feature activation+1.123
asking
Token asking
Feature activation+0.000
its
Token its
Feature activation+0.000
operatives
Token operatives
Feature activation+0.000
to
Token to
Feature activation+0.000
instead
Token instead
Feature activation+0.000
,
Token,
Feature activation+0.000
it
Token it
Feature activation+0.000
was
Token was
Feature activation+0.000
everything
Token everything
Feature activation+0.000
they
Token they
Feature activation+1.654
wanted
Token wanted
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Trump
TokenTrump
Feature activation+0.000

INTERVAL 1.268 - 1.521
CONTAINS 0.001%

the
Token the
Feature activation+0.337
film
Token film
Feature activation+0.000
shows
Token shows
Feature activation+0.000
us
Token us
Feature activation+0.000
that
Token that
Feature activation+0.000
he
Token he
Feature activation+1.460
knows
Token knows
Feature activation+0.000
who
Token who
Feature activation+0.000
the
Token the
Feature activation+0.000
important
Token important
Feature activation+0.000
people
Token people
Feature activation+0.000
the
Token the
Feature activation+0.000
lines
Token lines
Feature activation+0.000
.
Token.
Feature activation+0.000
Chand
Token Chand
Feature activation+0.000
rab
Tokenrab
Feature activation+0.000
han
Tokenhan
Feature activation+1.463
is
Token is
Feature activation+0.000
insin
Token insin
Feature activation+0.000
uating
Tokenuating
Feature activation+0.000
that
Token that
Feature activation+0.000
he
Token he
Feature activation+0.000
#
Token #
Feature activation+0.000
SD
TokenSD
Feature activation+0.000
Live
TokenLive
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
When
TokenWhen
Feature activation+1.481
?
Token?
Feature activation+0.000
#
Token #
Feature activation+0.000
Surv
TokenSurv
Feature activation+0.000
iv
Tokeniv
Feature activation+0.000
or
Tokenor
Feature activation+0.000
the
Token the
Feature activation+0.000
time
Token time
Feature activation+0.000
that
Token that
Feature activation+0.000
II
Token II
Feature activation+0.000
IR
TokenIR
Feature activation+2.536
's
Token's
Feature activation+1.274
claims
Token claims
Feature activation+0.000
to
Token to
Feature activation+0.084
the
Token the
Feature activation+0.112
trademark
Token trademark
Feature activation+0.000
were
Token were
Feature activation+0.000
T
TokenT
Feature activation+0.000
rying
Tokenrying
Feature activation+0.000
to
Token to
Feature activation+0.000
dismiss
Token dismiss
Feature activation+0.000
the
Token the
Feature activation+0.000
journalist
Token journalist
Feature activation+1.286
he
Token he
Feature activation+0.000
continued
Token continued
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
this
Tokenthis
Feature activation+0.000

INTERVAL 1.014 - 1.268
CONTAINS 0.002%

it
Token it
Feature activation+0.000
was
Token was
Feature activation+0.000
like
Token like
Feature activation+0.000
âĢ¦
Token âĢ¦
Feature activation+0.000
he
Token he
Feature activation+1.349
just
Token just
Feature activation+1.046
,
Token,
Feature activation+0.000
I
Token I
Feature activation+0.000
don
Token don
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
said
Token said
Feature activation+0.000
"
Token "
Feature activation+0.000
NO
TokenNO
Feature activation+0.000
"
Token"
Feature activation+0.000
and
Token and
Feature activation+0.000
he
Token he
Feature activation+1.212
dropped
Token dropped
Feature activation+0.000
out
Token out
Feature activation+0.000
(
Token (
Feature activation+0.000
said
Tokensaid
Feature activation+0.000
he
Token he
Feature activation+0.000
to
Token to
Feature activation+0.000
go
Token go
Feature activation+0.000
home
Token home
Feature activation+0.000
,
Token,
Feature activation+0.000
does
Token does
Feature activation+0.489
he
Token he
Feature activation+1.049
say
Token say
Feature activation+0.045
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
I
TokenI
Feature activation+0.000
want
Token want
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
Department
Token Department
Feature activation+1.136
of
Token of
Feature activation+0.000
Justice
Token Justice
Feature activation+2.201
appreci
Token appreci
Feature activation+0.000
ates
Tokenates
Feature activation+0.000
the
Token the
Feature activation+0.000
than
Token than
Feature activation+0.000
we
Token we
Feature activation+0.000
wanted
Token wanted
Feature activation+0.008
,
Token,
Feature activation+0.000
and
Token and
Feature activation+0.000
we
Token we
Feature activation+1.051
still
Token still
Feature activation+0.753
sold
Token sold
Feature activation+0.000
a
Token a
Feature activation+0.000
lot
Token lot
Feature activation+0.000
of
Token of
Feature activation+0.000

INTERVAL 0.761 - 1.014
CONTAINS 0.003%

jobs
Token jobs
Feature activation+0.000
in
Token in
Feature activation+0.000
order
Token order
Feature activation+0.000
to
Token to
Feature activation+0.000
cover
Token cover
Feature activation+0.561
all
Token all
Feature activation+0.936
her
Token her
Feature activation+0.722
legal
Token legal
Feature activation+0.000
fees
Token fees
Feature activation+0.000
.
Token.
Feature activation+0.000
She
Token She
Feature activation+0.000
whip
Token whip
Feature activation+0.000
rape
Token rape
Feature activation+0.000
victims
Token victims
Feature activation+0.000
if
Token if
Feature activation+0.000
you
Token you
Feature activation+0.000
believe
Token believe
Feature activation+0.955
God
Token God
Feature activation+0.912
commands
Token commands
Feature activation+0.000
it
Token it
Feature activation+0.000
?
Token?
Feature activation+0.000
Yes
Token Yes
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ms
TokenMs
Feature activation+0.027
Mal
Token Mal
Feature activation+0.061
hot
Tokenhot
Feature activation+0.000
ra
Tokenra
Feature activation+0.911
,
Token,
Feature activation+0.000
who
Token who
Feature activation+0.000
resigned
Token resigned
Feature activation+0.000
as
Token as
Feature activation+0.000
shadow
Token shadow
Feature activation+0.000
arrest
Token arrest
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
KEY
TokenKEY
Feature activation+0.000
E
TokenE
Feature activation+0.986
TV
Token TV
Feature activation+2.023
asked
Token asked
Feature activation+0.000
the
Token the
Feature activation+0.000
judge
Token judge
Feature activation+0.000
,
Token,
Feature activation+0.000
Lok
Token Lok
Feature activation+0.000
pal
Tokenpal
Feature activation+0.000
bill
Token bill
Feature activation+0.000
as
Token as
Feature activation+0.000
a
Token a
Feature activation+0.000
product
Token product
Feature activation+0.838
facing
Token facing
Feature activation+0.000
similar
Token similar
Feature activation+0.000
challenges
Token challenges
Feature activation+0.000
,
Token,
Feature activation+0.000
and
Token and
Feature activation+0.000

INTERVAL 0.507 - 0.761
CONTAINS 0.008%

this
Token this
Feature activation+0.000
month
Token month
Feature activation+0.000
that
Token that
Feature activation+0.000
Trump
Token Trump
Feature activation+0.000
never
Token never
Feature activation+0.000
asked
Token asked
Feature activation+0.537
him
Token him
Feature activation+0.000
about
Token about
Feature activation+0.000
Russia
Token Russia
Feature activation+0.000
's
Token's
Feature activation+0.000
interference
Token interference
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
He
TokenHe
Feature activation+0.000
said
Token said
Feature activation+0.382
he
Token he
Feature activation+0.551
dreamed
Token dreamed
Feature activation+0.000
of
Token of
Feature activation+0.000
living
Token living
Feature activation+0.000
under
Token under
Feature activation+0.000
Shar
Token Shar
Feature activation+0.000
Getty
Token Getty
Feature activation+0.000
Images
Token Images
Feature activation+0.000
)
Token)
Feature activation+0.000
Story
Token Story
Feature activation+0.000
Highlights
Token Highlights
Feature activation+0.000
Groups
Token Groups
Feature activation+0.676
:
Token:
Feature activation+0.000
Int
Token Int
Feature activation+0.000
im
Tokenim
Feature activation+0.000
acy
Tokenacy
Feature activation+0.000
is
Token is
Feature activation+0.000
in
Token in
Feature activation+0.000
a
Token a
Feature activation+0.000
position
Token position
Feature activation+0.000
where
Token where
Feature activation+0.000
even
Token even
Feature activation+0.000
if
Token if
Feature activation+0.537
we
Token we
Feature activation+0.682
don
Token don
Feature activation+0.526
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
t
Tokent
Feature activation+0.332
that
Token that
Feature activation+0.000
debate
Token debate
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
They
TokenThey
Feature activation+0.697
compromised
Token compromised
Feature activation+0.000
on
Token on
Feature activation+0.000
a
Token a
Feature activation+0.000
sidewalk
Token sidewalk
Feature activation+0.000
on
Token on
Feature activation+0.000

INTERVAL 0.254 - 0.507
CONTAINS 0.029%

.
Token.
Feature activation+0.000
My
Token My
Feature activation+0.000
dogs
Token dogs
Feature activation+0.000
also
Token also
Feature activation+0.000
respond
Token respond
Feature activation+0.000
to
Token to
Feature activation+0.294
"
Token "
Feature activation+0.000
Where
TokenWhere
Feature activation+0.000
's
Token's
Feature activation+0.000
Jo
Token Jo
Feature activation+0.000
annie
Tokenannie
Feature activation+0.000
a
Token a
Feature activation+0.000
Minnesota
Token Minnesota
Feature activation+0.000
federal
Token federal
Feature activation+0.000
judge
Token judge
Feature activation+0.000
has
Token has
Feature activation+0.000
issued
Token issued
Feature activation+0.436
an
Token an
Feature activation+0.022
order
Token order
Feature activation+0.000
(
Token (
Feature activation+0.000
PDF
TokenPDF
Feature activation+0.000
)
Token)
Feature activation+0.000
pursue
Token pursue
Feature activation+0.000
him
Token him
Feature activation+0.000
now
Token now
Feature activation+0.000
.
Token.
Feature activation+0.000
Other
Token Other
Feature activation+0.000
teams
Token teams
Feature activation+0.341
that
Token that
Feature activation+0.036
could
Token could
Feature activation+0.103
be
Token be
Feature activation+0.057
good
Token good
Feature activation+0.000
fits
Token fits
Feature activation+0.000
domain
Token domain
Feature activation+0.000
objects
Token objects
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Organ
TokenOrgan
Feature activation+0.000
izer
Tokenizer
Feature activation+0.430
Att
Token Att
Feature activation+0.000
endants
Tokenendants
Feature activation+0.000
Events
Token Events
Feature activation+0.000
Registration
Token Registration
Feature activation+0.000
Attend
Token Attend
Feature activation+0.000
them
Token them
Feature activation+0.000
.
Token.
Feature activation+0.000
Occasionally
Token Occasionally
Feature activation+0.000
,
Token,
Feature activation+0.000
we
Token we
Feature activation+0.217
came
Token came
Feature activation+0.270
across
Token across
Feature activation+0.000
a
Token a
Feature activation+0.000
product
Token product
Feature activation+0.000
that
Token that
Feature activation+0.000
is
Token is
Feature activation+0.000

INTERVAL 0.000 - 0.254
CONTAINS 99.955%

other
Token other
Feature activation+0.000
researchers
Token researchers
Feature activation+0.000
.
Token.
Feature activation+0.000
<|endoftext|>
Token<|endoftext|>
Feature activation+0.000
About
TokenAbout
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Bay
TokenBay
Feature activation+0.000
kok
Tokenkok
Feature activation+0.000
is
Token is
Feature activation+0.000
an
Token an
Feature activation+0.000
,
Token,
Feature activation+0.000
is
Token is
Feature activation+0.000
resisting
Token resisting
Feature activation+0.000
moves
Token moves
Feature activation+0.000
to
Token to
Feature activation+0.000
craft
Token craft
Feature activation+0.000
rules
Token rules
Feature activation+0.000
cracking
Token cracking
Feature activation+0.000
down
Token down
Feature activation+0.000
on
Token on
Feature activation+0.000
banks
Token banks
Feature activation+0.000
she
Token she
Feature activation+0.000
says
Token says
Feature activation+0.000
she
Token she
Feature activation+0.000
has
Token has
Feature activation+0.000
been
Token been
Feature activation+0.000
stalk
Token stalk
Feature activation+0.000
ed
Tokened
Feature activation+0.000
and
Token and
Feature activation+0.000
sp
Token sp
Feature activation+0.000
ied
Tokenied
Feature activation+0.000
upon
Token upon
Feature activation+0.000
President
Token President
Feature activation+0.000
Barack
Token Barack
Feature activation+0.000
Obama
Token Obama
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.000
health
Token health
Feature activation+0.000
care
Token care
Feature activation+0.000
law
Token law
Feature activation+0.000
and
Token and
Feature activation+0.000
rules
Token rules
Feature activation+0.000
before
Token before
Feature activation+0.000
heading
Token heading
Feature activation+0.000
back
Token back
Feature activation+0.000
to
Token to
Feature activation+0.000
Bruce
Token Bruce
Feature activation+0.000
's
Token's
Feature activation+0.000
home
Token home
Feature activation+0.000
base
Token base
Feature activation+0.000
in
Token in
Feature activation+0.000
the
Token the
Feature activation+0.000
Northeast
Token Northeast
Feature activation+0.000

BOTTOM ACTIVATIONS
MIN = 0.000

atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
Development
Token Development
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
<|endoftext|>
Token<|endoftext|>
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000

Top feature 1 in H7.1: (feature 40375)

TOP ACTIVATIONS
MAX = 1.966

the
Token the
Feature activation+0.000
snow
Token snow
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.003
I
TokenI
Feature activation+1.966
have
Token have
Feature activation+1.361
helped
Token helped
Feature activation+0.000
push
Token push
Feature activation+0.000
a
Token a
Feature activation+0.000
stranger
Token stranger
Feature activation+0.000
someone
Token someone
Feature activation+0.000
detailing
Token detailing
Feature activation+0.000
how
Token how
Feature activation+0.302
POS
Token POS
Feature activation+0.525
RE
Token RE
Feature activation+0.107
P
TokenP
Feature activation+1.891
was
Token was
Feature activation+1.561
instrumental
Token instrumental
Feature activation+0.930
in
Token in
Feature activation+0.000
a
Token a
Feature activation+0.000
suicide
Token suicide
Feature activation+0.000
job
Token job
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Professor
TokenProfessor
Feature activation+0.000
Rosen
Token Rosen
Feature activation+1.677
wrote
Token wrote
Feature activation+0.167
a
Token a
Feature activation+0.000
compelling
Token compelling
Feature activation+0.000
ARTICLE
Token ARTICLE
Feature activation+0.000
in
Token in
Feature activation+0.000
way
Token way
Feature activation+0.000
.
Token.
Feature activation+0.000
And
Token And
Feature activation+0.000
if
Token if
Feature activation+0.000
their
Token their
Feature activation+0.000
parents
Token parents
Feature activation+1.630
were
Token were
Feature activation+1.160
raised
Token raised
Feature activation+0.000
right
Token right
Feature activation+0.000
,
Token,
Feature activation+0.000
they
Token they
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
Cook
TokenCook
Feature activation+0.000
ing
Tokening
Feature activation+0.000
with
Token with
Feature activation+0.000
Ell
Token Ell
Feature activation+0.000
a
Tokena
Feature activation+1.619
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
or
Token or
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
,
Token,
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
our
Tokenour
Feature activation+0.668
map
Token map
Feature activation+0.000
editors
Token editors
Feature activation+1.574
will
Token will
Feature activation+1.210
make
Token make
Feature activation+0.641
that
Token that
Feature activation+0.000
change
Token change
Feature activation+0.000
.
Token.
Feature activation+0.050
detailing
Token detailing
Feature activation+0.000
how
Token how
Feature activation+0.302
POS
Token POS
Feature activation+0.525
RE
Token RE
Feature activation+0.107
P
TokenP
Feature activation+1.891
was
Token was
Feature activation+1.561
instrumental
Token instrumental
Feature activation+0.930
in
Token in
Feature activation+0.000
a
Token a
Feature activation+0.000
suicide
Token suicide
Feature activation+0.000
intervention
Token intervention
Feature activation+0.000
,
Token,
Feature activation+0.000
suggested
Token suggested
Feature activation+0.000
that
Token that
Feature activation+0.042
the
Token the
Feature activation+0.089
Indian
Token Indian
Feature activation+0.000
government
Token government
Feature activation+1.467
should
Token should
Feature activation+1.207
do
Token do
Feature activation+1.067
"
Token "
Feature activation+0.000
what
Tokenwhat
Feature activation+0.102
China
Token China
Feature activation+0.000
which
Token which
Feature activation+0.000
we
Token we
Feature activation+0.072
could
Token could
Feature activation+0.000
completely
Token completely
Feature activation+0.000
depend
Token depend
Feature activation+0.901
on
Token on
Feature activation+1.374
others
Token others
Feature activation+0.860
are
Token are
Feature activation+0.000
on
Token on
Feature activation+0.000
the
Token the
Feature activation+0.000
way
Token way
Feature activation+0.000
snow
Token snow
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.003
I
TokenI
Feature activation+1.966
have
Token have
Feature activation+1.361
helped
Token helped
Feature activation+0.000
push
Token push
Feature activation+0.000
a
Token a
Feature activation+0.000
stranger
Token stranger
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
had
Token had
Feature activation+0.000
stories
Token stories
Feature activation+0.000
where
Token where
Feature activation+0.000
Peter
Token Peter
Feature activation+1.023
had
Token had
Feature activation+0.594
to
Token to
Feature activation+1.334
raise
Token raise
Feature activation+0.000
money
Token money
Feature activation+0.000
to
Token to
Feature activation+0.000
help
Token help
Feature activation+0.270
keep
Token keep
Feature activation+0.172
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
G
TokenG
Feature activation+0.000
arn
Tokenarn
Feature activation+0.000
ett
Tokenett
Feature activation+1.334
was
Token was
Feature activation+0.277
with
Token with
Feature activation+0.000
the
Token the
Feature activation+0.000
team
Token team
Feature activation+0.000
in
Token in
Feature activation+0.000
the
Token the
Feature activation+0.000
Delaware
Token Delaware
Feature activation+0.000
Forest
Token Forest
Feature activation+0.000
Service
Token Service
Feature activation+0.378
is
Token is
Feature activation+0.159
sending
Token sending
Feature activation+1.307
help
Token help
Feature activation+0.000
in
Token in
Feature activation+0.000
the
Token the
Feature activation+0.000
form
Token form
Feature activation+0.000
of
Token of
Feature activation+0.939
to
Token to
Feature activation+0.000
endure
Token endure
Feature activation+0.000
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.000
help
Token help
Feature activation+0.607
and
Token and
Feature activation+1.301
generosity
Token generosity
Feature activation+0.456
made
Token made
Feature activation+0.135
it
Token it
Feature activation+0.011
much
Token much
Feature activation+0.000
more
Token more
Feature activation+0.000
,
Token,
Feature activation+0.000
I
Token I
Feature activation+0.000
have
Token have
Feature activation+0.000
a
Token a
Feature activation+0.000
right
Token right
Feature activation+0.000
to
Token to
Feature activation+1.257
your
Token your
Feature activation+0.000
help
Token help
Feature activation+0.000
"
Token"
Feature activation+0.000
Mar
Token Mar
Feature activation+0.000
at
Tokenat
Feature activation+0.000
by
Token by
Feature activation+0.000
people
Token people
Feature activation+0.015
whose
Token whose
Feature activation+0.025
businesses
Token businesses
Feature activation+0.000
have
Token have
Feature activation+0.000
received
Token received
Feature activation+1.240
taxpayer
Token taxpayer
Feature activation+0.000
money
Token money
Feature activation+0.000
but
Token but
Feature activation+0.000
are
Token are
Feature activation+0.000
so
Token so
Feature activation+0.000
Perhaps
Token Perhaps
Feature activation+0.000
the
Token the
Feature activation+0.000
Jama
Token Jama
Feature activation+0.000
ican
Tokenican
Feature activation+0.000
Olympic
Token Olympic
Feature activation+0.000
Association
Token Association
Feature activation+1.228
will
Token will
Feature activation+0.936
pony
Token pony
Feature activation+0.000
up
Token up
Feature activation+0.000
some
Token some
Feature activation+0.000
dough
Token dough
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
our
Tokenour
Feature activation+0.668
map
Token map
Feature activation+0.000
editors
Token editors
Feature activation+1.574
will
Token will
Feature activation+1.210
make
Token make
Feature activation+0.641
that
Token that
Feature activation+0.000
change
Token change
Feature activation+0.000
.
Token.
Feature activation+0.050
It
Token It
Feature activation+0.000
suggested
Token suggested
Feature activation+0.000
that
Token that
Feature activation+0.042
the
Token the
Feature activation+0.089
Indian
Token Indian
Feature activation+0.000
government
Token government
Feature activation+1.467
should
Token should
Feature activation+1.207
do
Token do
Feature activation+1.067
"
Token "
Feature activation+0.000
what
Tokenwhat
Feature activation+0.102
China
Token China
Feature activation+0.000
did
Token did
Feature activation+0.130
.
Token.
Feature activation+0.000
And
Token And
Feature activation+0.000
if
Token if
Feature activation+0.000
their
Token their
Feature activation+0.000
parents
Token parents
Feature activation+1.630
were
Token were
Feature activation+1.160
raised
Token raised
Feature activation+0.000
right
Token right
Feature activation+0.000
,
Token,
Feature activation+0.000
they
Token they
Feature activation+0.000
could
Token could
Feature activation+0.000

Top DFA by src position
MAX = 2.296

Often
Token Often
Feature activation-0.007
Top resid features:
Ċ
TokenĊ
Feature activation+0.001
Top resid features:
Ċ
TokenĊ
Feature activation-0.003
Top resid features:
I
TokenI
Feature activation+0.003
Top resid features:
have
Token have
Feature activation-0.057
Top resid features:
helped
Token helped
Feature activation+2.296
Top resid features:
push
Token push
Feature activation+0.070
Top resid features:
a
Token a
Feature activation+0.020
Top resid features:
stranger
Token stranger
Feature activation+0.011
Top resid features:
âĢ
TokenâĢ
Feature activation+0.017
Top resid features:
Ļ
TokenĻ
Feature activation+0.003
Top resid features:
how
Token how
Feature activation-0.004
Top resid features:
POS
Token POS
Feature activation-0.001
Top resid features:
RE
Token RE
Feature activation+0.001
Top resid features:
P
TokenP
Feature activation-0.008
Top resid features:
was
Token was
Feature activation-0.015
Top resid features:
instrumental
Token instrumental
Feature activation+1.395
Top resid features:
in
Token in
Feature activation+0.501
Top resid features:
a
Token a
Feature activation+0.029
Top resid features:
suicide
Token suicide
Feature activation-0.001
Top resid features:
intervention
Token intervention
Feature activation-0.000
Top resid features:
.
Token.
Feature activation-0.001
Top resid features:
Ċ
TokenĊ
Feature activation+0.002
Top resid features:
Ċ
TokenĊ
Feature activation+0.000
Top resid features:
Professor
TokenProfessor
Feature activation-0.017
Top resid features:
Rosen
Token Rosen
Feature activation-0.004
Top resid features:
has
Token has
Feature activation+0.066
Top resid features:
advised
Token advised
Feature activation+2.178
Top resid features:
the
Token the
Feature activation+0.017
Top resid features:
media
Token media
Feature activation-0.001
Top resid features:
to
Token to
Feature activation-0.008
Top resid features:
throw
Token throw
Feature activation+0.003
Top resid features:
out
Token out
Feature activation-0.000
Top resid features:
would
Token would
Feature activation-0.005
Top resid features:
have
Token have
Feature activation+0.001
Top resid features:
their
Token their
Feature activation+0.001
Top resid features:
parents
Token parents
Feature activation-0.008
Top resid features:
to
Token to
Feature activation+0.040
Top resid features:
guide
Token guide
Feature activation+1.973
Top resid features:
them
Token them
Feature activation+0.014
Top resid features:
along
Token along
Feature activation+0.011
Top resid features:
the
Token the
Feature activation+0.006
Top resid features:
way
Token way
Feature activation+0.008
Top resid features:
.
Token.
Feature activation+0.017
Top resid features:
E
TokenE
Feature activation-0.002
Top resid features:
lla
Tokenlla
Feature activation+0.020
Top resid features:
did
Token did
Feature activation+0.045
Top resid features:
a
Token a
Feature activation+0.040
Top resid features:
great
Token great
Feature activation+0.045
Top resid features:
job
Token job
Feature activation+1.755
Top resid features:
and
Token and
Feature activation+0.024
Top resid features:
we
Token we
Feature activation+0.006
Top resid features:
have
Token have
Feature activation+0.004
Top resid features:
already
Token already
Feature activation+0.007
Top resid features:
made
Token made
Feature activation+0.004
Top resid features:
of
Token of
Feature activation+0.003
Top resid features:
local
Token local
Feature activation+0.015
Top resid features:
volunteer
Token volunteer
Feature activation+0.073
Top resid features:
editors
Token editors
Feature activation+0.078
Top resid features:
who
Token who
Feature activation+0.074
Top resid features:
ensure
Token ensure
Feature activation+1.494
Top resid features:
that
Token that
Feature activation+0.125
Top resid features:
the
Token the
Feature activation+0.001
Top resid features:
maps
Token maps
Feature activation+0.000
Top resid features:
stay
Token stay
Feature activation-0.008
Top resid features:
up
Token up
Feature activation+0.004
Top resid features:
how
Token how
Feature activation-0.005
Top resid features:
POS
Token POS
Feature activation+0.007
Top resid features:
RE
Token RE
Feature activation+0.001
Top resid features:
P
TokenP
Feature activation-0.001
Top resid features:
was
Token was
Feature activation+0.003
Top resid features:
instrumental
Token instrumental
Feature activation+1.007
Top resid features:
in
Token in
Feature activation+0.857
Top resid features:
a
Token a
Feature activation+0.100
Top resid features:
suicide
Token suicide
Feature activation-0.004
Top resid features:
intervention
Token intervention
Feature activation-0.001
Top resid features:
.
Token.
Feature activation-0.002
Top resid features:
for
Token for
Feature activation-0.000
Top resid features:
the
Token the
Feature activation+0.000
Top resid features:
Indian
Token Indian
Feature activation-0.001
Top resid features:
government
Token government
Feature activation-0.002
Top resid features:
to
Token to
Feature activation-0.026
Top resid features:
intervene
Token intervene
Feature activation+1.216
Top resid features:
and
Token and
Feature activation+0.034
Top resid features:
protect
Token protect
Feature activation+0.638
Top resid features:
local
Token local
Feature activation+0.005
Top resid features:
companies
Token companies
Feature activation+0.002
Top resid features:
are
Token are
Feature activation-0.008
Top resid features:
no
Token no
Feature activation+0.000
Top resid features:
longer
Token longer
Feature activation+0.006
Top resid features:
rely
Token rely
Feature activation+0.014
Top resid features:
on
Token on
Feature activation+0.006
Top resid features:
America
Token America
Feature activation+0.164
Top resid features:
in
Token in
Feature activation+1.403
Top resid features:
allied
Token allied
Feature activation+0.089
Top resid features:
foreign
Token foreign
Feature activation+0.004
Top resid features:
diplomacy
Token diplomacy
Feature activation+0.028
Top resid features:
.
Token.
Feature activation-0.002
Top resid features:
âĢ
Token âĢ
Feature activation-0.000
Top resid features:
Often
Token Often
Feature activation-0.005
Top resid features:
Ċ
TokenĊ
Feature activation+0.001
Top resid features:
Ċ
TokenĊ
Feature activation-0.002
Top resid features:
I
TokenI
Feature activation+0.005
Top resid features:
have
Token have
Feature activation-0.002
Top resid features:
helped
Token helped
Feature activation+1.383
Top resid features:
push
Token push
Feature activation+0.183
Top resid features:
a
Token a
Feature activation+0.036
Top resid features:
stranger
Token stranger
Feature activation+0.015
Top resid features:
âĢ
TokenâĢ
Feature activation+0.016
Top resid features:
Ļ
TokenĻ
Feature activation+0.002
Top resid features:
to
Token to
Feature activation+0.001
Top resid features:
step
Token step
Feature activation+0.005
Top resid features:
up
Token up
Feature activation+0.160
Top resid features:
to
Token to
Feature activation+0.074
Top resid features:
take
Token take
Feature activation+0.135
Top resid features:
care
Token care
Feature activation+0.688
Top resid features:
of
Token of
Feature activation+0.301
Top resid features:
the
Token the
Feature activation+0.087
Top resid features:
woman
Token woman
Feature activation+0.011
Top resid features:
who
Token who
Feature activation+0.004
Top resid features:
had
Token had
Feature activation+0.005
Top resid features:
.
Token.
Feature activation-0.039
Top resid features:
Garn
Token Garn
Feature activation-0.000
Top resid features:
ett
Tokenett
Feature activation-0.002
Top resid features:
will
Token will
Feature activation-0.002
Top resid features:
occasionally
Token occasionally
Feature activation+0.018
Top resid features:
help
Token help
Feature activation+0.998
Top resid features:
out
Token out
Feature activation+0.083
Top resid features:
at
Token at
Feature activation+0.021
Top resid features:
practices
Token practices
Feature activation+0.004
Top resid features:
but
Token but
Feature activation-0.014
Top resid features:
will
Token will
Feature activation-0.012
Top resid features:
aware
Tokenaware
Feature activation+0.002
Top resid features:
Forest
Token Forest
Feature activation-0.001
Top resid features:
Service
Token Service
Feature activation+0.000
Top resid features:
to
Token to
Feature activation-0.029
Top resid features:
send
Token send
Feature activation+0.005
Top resid features:
help
Token help
Feature activation+0.848
Top resid features:
to
Token to
Feature activation+0.806
Top resid features:
Rocky
Token Rocky
Feature activation-0.002
Top resid features:
Mountains
Token Mountains
Feature activation+0.001
Top resid features:
Ċ
TokenĊ
Feature activation+0.034
Top resid features:
Ċ
TokenĊ
Feature activation+0.007
Top resid features:
surrounding
Token surrounding
Feature activation+0.008
Top resid features:
areas
Token areas
Feature activation+0.003
Top resid features:
was
Token was
Feature activation+0.002
Top resid features:
respectful
Token respectful
Feature activation+0.027
Top resid features:
and
Token and
Feature activation+0.006
Top resid features:
helpful
Token helpful
Feature activation+0.773
Top resid features:
,
Token,
Feature activation+0.078
Top resid features:
as
Token as
Feature activation+0.025
Top resid features:
well
Token well
Feature activation+0.009
Top resid features:
as
Token as
Feature activation+0.043
Top resid features:
generous
Token generous
Feature activation+0.560
Top resid features:
have
Token have
Feature activation+0.001
Top resid features:
a
Token a
Feature activation-0.007
Top resid features:
right
Token right
Feature activation+0.027
Top resid features:
to
Token to
Feature activation-0.005
Top resid features:
your
Token your
Feature activation-0.157
Top resid features:
help
Token help
Feature activation+1.773
Top resid features:
"
Token"
Feature activation+0.006
Top resid features:
Detail
Token Detail
Feature activation-0.001
Top resid features:
of
Token of
Feature activation-0.000
Top resid features:
sh
Tokensh
Feature activation-0.030
Top resid features:
owing
Tokenowing
Feature activation-0.002
Top resid features:
that
Token that
Feature activation-0.003
Top resid features:
had
Token had
Feature activation+0.000
Top resid features:
received
Token received
Feature activation-0.003
Top resid features:
very
Token very
Feature activation-0.035
Top resid features:
direct
Token direct
Feature activation+0.041
Top resid features:
help
Token help
Feature activation+1.127
Top resid features:
from
Token from
Feature activation+0.406
Top resid features:
the
Token the
Feature activation+0.114
Top resid features:
government
Token government
Feature activation+0.020
Top resid features:
.
Token .
Feature activation+0.025
Top resid features:
It
Token It
Feature activation-0.000
Top resid features:
Association
Token Association
Feature activation-0.003
Top resid features:
will
Token will
Feature activation+0.001
Top resid features:
step
Token step
Feature activation+0.243
Top resid features:
in
Token in
Feature activation+0.499
Top resid features:
and
Token and
Feature activation+0.018
Top resid features:
support
Token support
Feature activation+0.898
Top resid features:
us
Token us
Feature activation+0.008
Top resid features:
now
Token now
Feature activation+0.010
Top resid features:
we
Token we
Feature activation-0.001
Top resid features:
âĢ
TokenâĢ
Feature activation+0.002
Top resid features:
Ļ
TokenĻ
Feature activation+0.000
Top resid features:
of
Token of
Feature activation+0.004
Top resid features:
local
Token local
Feature activation+0.006
Top resid features:
volunteer
Token volunteer
Feature activation+0.160
Top resid features:
editors
Token editors
Feature activation+0.057
Top resid features:
who
Token who
Feature activation+0.026
Top resid features:
ensure
Token ensure
Feature activation+1.104
Top resid features:
that
Token that
Feature activation+0.119
Top resid features:
the
Token the
Feature activation+0.001
Top resid features:
maps
Token maps
Feature activation+0.001
Top resid features:
stay
Token stay
Feature activation+0.004
Top resid features:
up
Token up
Feature activation+0.007
Top resid features:
for
Token for
Feature activation-0.000
Top resid features:
the
Token the
Feature activation+0.000
Top resid features:
Indian
Token Indian
Feature activation-0.001
Top resid features:
government
Token government
Feature activation-0.001
Top resid features:
to
Token to
Feature activation-0.018
Top resid features:
intervene
Token intervene
Feature activation+0.993
Top resid features:
and
Token and
Feature activation+0.046
Top resid features:
protect
Token protect
Feature activation+0.555
Top resid features:
local
Token local
Feature activation+0.013
Top resid features:
companies
Token companies
Feature activation+0.002
Top resid features:
are
Token are
Feature activation-0.012
Top resid features:
would
Token would
Feature activation-0.000
Top resid features:
have
Token have
Feature activation+0.001
Top resid features:
their
Token their
Feature activation-0.002
Top resid features:
parents
Token parents
Feature activation-0.006
Top resid features:
to
Token to
Feature activation+0.079
Top resid features:
guide
Token guide
Feature activation+1.422
Top resid features:
them
Token them
Feature activation+0.010
Top resid features:
along
Token along
Feature activation+0.008
Top resid features:
the
Token the
Feature activation+0.002
Top resid features:
way
Token way
Feature activation+0.005
Top resid features:
.
Token.
Feature activation+0.043
Top resid features:

Decoder Weights Distribution

Head 0: 0.03

Head 1: 0.50

Head 2: 0.02

Head 3: 0.06

Head 4: 0.02

Head 5: 0.10

Head 6: 0.08

Head 7: 0.03

Head 8: 0.06

Head 9: 0.05

Head 10: 0.03

Head 11: 0.02

Positive logits

helping4.20

assisting3.88

helpers3.78

help3.59

helps3.52

helped3.45

rescuing3.41

helps3.38

encour3.34

Help3.33

Volunteers3.30

volunteering3.29

aiding3.23

invaluable3.22

volunteered3.20

volunteers3.14

aid3.14

interven3.14

teamwork3.12

Assist3.12

Negative logits

ocry-2.55

vari-2.46

classic-2.42

inflation-2.39

classification-2.37

trunc-2.37

Decay-2.37

Classification-2.34

intage-2.29

equival-2.25

usional-2.23

urst-2.21

istors-2.19

<=-2.18

terms-2.18

uberty-2.17

definition-2.17

mania-2.16

ieties-2.15

onyms-2.15

INTERVAL 1.769 - 1.966
CONTAINS 0.000%

the
Token the
Feature activation+0.000
snow
Token snow
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.003
I
TokenI
Feature activation+1.966
have
Token have
Feature activation+1.361
helped
Token helped
Feature activation+0.000
push
Token push
Feature activation+0.000
a
Token a
Feature activation+0.000
stranger
Token stranger
Feature activation+0.000
someone
Token someone
Feature activation+0.000
detailing
Token detailing
Feature activation+0.000
how
Token how
Feature activation+0.302
POS
Token POS
Feature activation+0.525
RE
Token RE
Feature activation+0.107
P
TokenP
Feature activation+1.891
was
Token was
Feature activation+1.561
instrumental
Token instrumental
Feature activation+0.930
in
Token in
Feature activation+0.000
a
Token a
Feature activation+0.000
suicide
Token suicide
Feature activation+0.000

INTERVAL 1.572 - 1.769
CONTAINS 0.000%

,
Token,
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
our
Tokenour
Feature activation+0.668
map
Token map
Feature activation+0.000
editors
Token editors
Feature activation+1.574
will
Token will
Feature activation+1.210
make
Token make
Feature activation+0.641
that
Token that
Feature activation+0.000
change
Token change
Feature activation+0.000
.
Token.
Feature activation+0.050
job
Token job
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Professor
TokenProfessor
Feature activation+0.000
Rosen
Token Rosen
Feature activation+1.677
wrote
Token wrote
Feature activation+0.167
a
Token a
Feature activation+0.000
compelling
Token compelling
Feature activation+0.000
ARTICLE
Token ARTICLE
Feature activation+0.000
in
Token in
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
Cook
TokenCook
Feature activation+0.000
ing
Tokening
Feature activation+0.000
with
Token with
Feature activation+0.000
Ell
Token Ell
Feature activation+0.000
a
Tokena
Feature activation+1.619
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
or
Token or
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
way
Token way
Feature activation+0.000
.
Token.
Feature activation+0.000
And
Token And
Feature activation+0.000
if
Token if
Feature activation+0.000
their
Token their
Feature activation+0.000
parents
Token parents
Feature activation+1.630
were
Token were
Feature activation+1.160
raised
Token raised
Feature activation+0.000
right
Token right
Feature activation+0.000
,
Token,
Feature activation+0.000
they
Token they
Feature activation+0.000

INTERVAL 1.376 - 1.572
CONTAINS 0.000%

,
Token,
Feature activation+0.000
suggested
Token suggested
Feature activation+0.000
that
Token that
Feature activation+0.042
the
Token the
Feature activation+0.089
Indian
Token Indian
Feature activation+0.000
government
Token government
Feature activation+1.467
should
Token should
Feature activation+1.207
do
Token do
Feature activation+1.067
"
Token "
Feature activation+0.000
what
Tokenwhat
Feature activation+0.102
China
Token China
Feature activation+0.000
detailing
Token detailing
Feature activation+0.000
how
Token how
Feature activation+0.302
POS
Token POS
Feature activation+0.525
RE
Token RE
Feature activation+0.107
P
TokenP
Feature activation+1.891
was
Token was
Feature activation+1.561
instrumental
Token instrumental
Feature activation+0.930
in
Token in
Feature activation+0.000
a
Token a
Feature activation+0.000
suicide
Token suicide
Feature activation+0.000
intervention
Token intervention
Feature activation+0.000

INTERVAL 1.179 - 1.376
CONTAINS 0.001%

by
Token by
Feature activation+0.000
people
Token people
Feature activation+0.015
whose
Token whose
Feature activation+0.025
businesses
Token businesses
Feature activation+0.000
have
Token have
Feature activation+0.000
received
Token received
Feature activation+1.240
taxpayer
Token taxpayer
Feature activation+0.000
money
Token money
Feature activation+0.000
but
Token but
Feature activation+0.000
are
Token are
Feature activation+0.000
so
Token so
Feature activation+0.000
Perhaps
Token Perhaps
Feature activation+0.000
the
Token the
Feature activation+0.000
Jama
Token Jama
Feature activation+0.000
ican
Tokenican
Feature activation+0.000
Olympic
Token Olympic
Feature activation+0.000
Association
Token Association
Feature activation+1.228
will
Token will
Feature activation+0.936
pony
Token pony
Feature activation+0.000
up
Token up
Feature activation+0.000
some
Token some
Feature activation+0.000
dough
Token dough
Feature activation+0.000
snow
Token snow
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.003
I
TokenI
Feature activation+1.966
have
Token have
Feature activation+1.361
helped
Token helped
Feature activation+0.000
push
Token push
Feature activation+0.000
a
Token a
Feature activation+0.000
stranger
Token stranger
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
to
Token to
Feature activation+0.000
endure
Token endure
Feature activation+0.000
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.000
help
Token help
Feature activation+0.607
and
Token and
Feature activation+1.301
generosity
Token generosity
Feature activation+0.456
made
Token made
Feature activation+0.135
it
Token it
Feature activation+0.011
much
Token much
Feature activation+0.000
more
Token more
Feature activation+0.000
suggested
Token suggested
Feature activation+0.000
that
Token that
Feature activation+0.042
the
Token the
Feature activation+0.089
Indian
Token Indian
Feature activation+0.000
government
Token government
Feature activation+1.467
should
Token should
Feature activation+1.207
do
Token do
Feature activation+1.067
"
Token "
Feature activation+0.000
what
Tokenwhat
Feature activation+0.102
China
Token China
Feature activation+0.000
did
Token did
Feature activation+0.130

INTERVAL 0.983 - 1.179
CONTAINS 0.002%

that
Token that
Feature activation+0.042
the
Token the
Feature activation+0.089
Indian
Token Indian
Feature activation+0.000
government
Token government
Feature activation+1.467
should
Token should
Feature activation+1.207
do
Token do
Feature activation+1.067
"
Token "
Feature activation+0.000
what
Tokenwhat
Feature activation+0.102
China
Token China
Feature activation+0.000
did
Token did
Feature activation+0.130
15
Token 15
Feature activation+0.000
I
Token I
Feature activation+0.000
want
Token want
Feature activation+0.000
to
Token to
Feature activation+0.000
thank
Token thank
Feature activation+0.133
you
Token you
Feature activation+0.541
for
Token for
Feature activation+1.042
what
Token what
Feature activation+0.483
you
Token you
Feature activation+0.877
have
Token have
Feature activation+0.612
done
Token done
Feature activation+0.382
,"
Token,"
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Because
TokenBecause
Feature activation+0.000
of
Token of
Feature activation+0.000
his
Token his
Feature activation+1.119
âĢ
Token âĢ
Feature activation+0.000
ĺ
Tokenĺ
Feature activation+0.000
warning
Tokenwarning
Feature activation+0.000
shot
Token shot
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
ruling
Token ruling
Feature activation+0.000
,
Token,
Feature activation+0.000
my
Token my
Feature activation+0.605
ability
Token ability
Feature activation+0.000
to
Token to
Feature activation+0.649
give
Token give
Feature activation+1.010
advice
Token advice
Feature activation+0.329
to
Token to
Feature activation+0.000
those
Token those
Feature activation+0.000
law
Token law
Feature activation+0.000
-
Token-
Feature activation+0.000
Error
TokenError
Feature activation+0.000
detection
Token detection
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
EC
TokenEC
Feature activation+0.392
C
TokenC
Feature activation+1.098
-
Token-
Feature activation+0.000
cap
Tokencap
Feature activation+0.000
able
Tokenable
Feature activation+0.000
RAM
Token RAM
Feature activation+0.000
modules
Token modules
Feature activation+0.000

INTERVAL 0.786 - 0.983
CONTAINS 0.005%

Monday
Token Monday
Feature activation+0.000
evening
Token evening
Feature activation+0.000
when
Token when
Feature activation+0.741
he
Token he
Feature activation+0.414
helped
Token helped
Feature activation+1.042
a
Token a
Feature activation+0.900
group
Token group
Feature activation+0.565
of
Token of
Feature activation+0.480
wind
Token wind
Feature activation+0.000
sur
Tokensur
Feature activation+0.000
f
Tokenf
Feature activation+0.000
uggish
Tokenuggish
Feature activation+0.000
dictatorship
Token dictatorship
Feature activation+0.000
.
Token.
Feature activation+0.000
U
Token U
Feature activation+0.000
.
Token.
Feature activation+0.000
S
TokenS
Feature activation+0.981
.
Token.
Feature activation+0.877
air
Token air
Feature activation+0.049
power
Token power
Feature activation+0.000
helped
Token helped
Feature activation+0.000
topple
Token topple
Feature activation+0.000
are
Token are
Feature activation+0.000
four
Token four
Feature activation+0.000
things
Token things
Feature activation+0.000
NATO
Token NATO
Feature activation+0.000
needs
Token needs
Feature activation+0.642
to
Token to
Feature activation+0.801
do
Token do
Feature activation+0.398
when
Token when
Feature activation+0.000
its
Token its
Feature activation+0.000
leaders
Token leaders
Feature activation+0.000
gather
Token gather
Feature activation+0.000
of
Token of
Feature activation+0.000
poor
Token poor
Feature activation+0.000
families
Token families
Feature activation+0.000
there
Token there
Feature activation+0.000
.
Token.
Feature activation+0.074
It
Token It
Feature activation+0.858
's
Token's
Feature activation+0.018
the
Token the
Feature activation+0.767
Low
Token Low
Feature activation+0.000
Income
Token Income
Feature activation+0.000
Home
Token Home
Feature activation+0.000
from
Token from
Feature activation+0.000
an
Token an
Feature activation+0.000
ownership
Token ownership
Feature activation+0.000
standpoint
Token standpoint
Feature activation+0.000
that
Token that
Feature activation+0.000
we
Token we
Feature activation+0.833
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.051
ve
Tokenve
Feature activation+0.234
never
Token never
Feature activation+0.357
lost
Token lost
Feature activation+0.000

INTERVAL 0.590 - 0.786
CONTAINS 0.014%

also
Token also
Feature activation+0.000
admitted
Token admitted
Feature activation+0.000
that
Token that
Feature activation+0.000
the
Token the
Feature activation+0.000
agency
Token agency
Feature activation+0.000
did
Token did
Feature activation+0.747
not
Token not
Feature activation+0.013
have
Token have
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
access
Tokenaccess
Feature activation+0.000
the
Token the
Feature activation+0.238
Pentagon
Token Pentagon
Feature activation+0.000
credits
Token credits
Feature activation+0.000
McCl
Token McCl
Feature activation+0.000
ough
Tokenough
Feature activation+0.000
an
Tokenan
Feature activation+0.660
with
Token with
Feature activation+0.595
saving
Token saving
Feature activation+0.288
the
Token the
Feature activation+0.001
lives
Token lives
Feature activation+0.000
of
Token of
Feature activation+0.184
of
Token of
Feature activation+0.000
America
Token America
Feature activation+0.000
,
Token,
Feature activation+0.000
admit
Token admit
Feature activation+0.000
to
Token to
Feature activation+0.000
giving
Token giving
Feature activation+0.748
cash
Token cash
Feature activation+0.000
and
Token and
Feature activation+0.000
theological
Token theological
Feature activation+0.000
support
Token support
Feature activation+0.000
to
Token to
Feature activation+0.000
/
Token/
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Th
TokenTh
Feature activation+0.000
an
Tokenan
Feature activation+0.000
atos
Tokenatos
Feature activation+0.633
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.312
Pictures
Token Pictures
Feature activation+0.000
:
Token:
Feature activation+0.000
You
Token You
Feature activation+0.000
may
Token may
Feature activation+0.231
with
Token with
Feature activation+0.093
reason
Token reason
Feature activation+0.000
share
Token share
Feature activation+0.520
in
Token in
Feature activation+0.662
his
Token his
Feature activation+0.000
treason
Token treason
Feature activation+0.000
,
Token,
Feature activation+0.000
and
Token and
Feature activation+0.000
in
Token in
Feature activation+0.000

INTERVAL 0.393 - 0.590
CONTAINS 0.031%

him
Token him
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
officers
Token officers
Feature activation+0.477
have
Token have
Feature activation+0.000
appealed
Token appealed
Feature activation+0.000
their
Token their
Feature activation+0.000
dismiss
Token dismiss
Feature activation+0.000
als
Tokenals
Feature activation+0.000
keep
Token keep
Feature activation+0.000
at
Token at
Feature activation+0.000
least
Token least
Feature activation+0.000
the
Token the
Feature activation+0.000
families
Token families
Feature activation+0.000
Jamie
Token Jamie
Feature activation+0.575
is
Token is
Feature activation+0.378
sworn
Token sworn
Feature activation+0.000
to
Token to
Feature activation+0.000
protect
Token protect
Feature activation+0.071
from
Token from
Feature activation+0.000
goal
Token goal
Feature activation+0.000
to
Token to
Feature activation+0.000
work
Token work
Feature activation+0.290
hol
Token hol
Feature activation+0.000
istically
Tokenistically
Feature activation+0.000
with
Token with
Feature activation+0.441
our
Token our
Feature activation+0.399
community
Token community
Feature activation+0.694
of
Token of
Feature activation+0.694
drivers
Token drivers
Feature activation+0.000
,
Token,
Feature activation+0.000
knowing
Token knowing
Feature activation+0.000
me
Token me
Feature activation+0.000
,
Token,
Feature activation+0.010
they
Token they
Feature activation+0.776
've
Token've
Feature activation+0.581
come
Token come
Feature activation+0.536
forward
Token forward
Feature activation+0.000
to
Token to
Feature activation+0.000
help
Token help
Feature activation+0.000
out
Token out
Feature activation+0.041
and
Token and
Feature activation+0.244
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
We
TokenWe
Feature activation+0.000
spoke
Token spoke
Feature activation+0.000
to
Token to
Feature activation+0.000
Pen
Token Pen
Feature activation+0.530
ello
Tokenello
Feature activation+0.310
earlier
Token earlier
Feature activation+0.000
this
Token this
Feature activation+0.000
afternoon
Token afternoon
Feature activation+0.000
at
Token at
Feature activation+0.000

INTERVAL 0.197 - 0.393
CONTAINS 0.096%

400
Token 400
Feature activation+0.000
,
Token,
Feature activation+0.000
000
Token000
Feature activation+0.000
Muslim
Token Muslim
Feature activation+0.000
soldiers
Token soldiers
Feature activation+0.000
who
Token who
Feature activation+0.251
fought
Token fought
Feature activation+0.000
for
Token for
Feature activation+0.000
Britain
Token Britain
Feature activation+0.000
is
Token is
Feature activation+0.000
one
Token one
Feature activation+0.000
were
Token were
Feature activation+0.000
evacuated
Token evacuated
Feature activation+0.000
from
Token from
Feature activation+0.000
the
Token the
Feature activation+0.000
Aqu
Token Aqu
Feature activation+0.000
arius
Tokenarius
Feature activation+0.224
by
Token by
Feature activation+0.000
helicopter
Token helicopter
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
lot
Token lot
Feature activation+0.022
.
Token.
Feature activation+0.000
Co
Token Co
Feature activation+0.000
aches
Tokenaches
Feature activation+0.000
had
Token had
Feature activation+0.000
me
Token me
Feature activation+0.248
working
Token working
Feature activation+0.245
on
Token on
Feature activation+0.000
my
Token my
Feature activation+0.000
ball
Token ball
Feature activation+0.000
handling
Token handling
Feature activation+0.000
centuries
Token centuries
Feature activation+0.060
our
Token our
Feature activation+0.000
brave
Token brave
Feature activation+0.000
men
Token men
Feature activation+0.138
and
Token and
Feature activation+0.000
women
Token women
Feature activation+0.233
in
Token in
Feature activation+0.000
uniform
Token uniform
Feature activation+0.000
have
Token have
Feature activation+0.242
overcome
Token overcome
Feature activation+0.083
tyranny
Token tyranny
Feature activation+0.000
ensure
Token ensure
Feature activation+0.000
he
Token he
Feature activation+0.171
continues
Token continues
Feature activation+0.288
to
Token to
Feature activation+0.714
work
Token work
Feature activation+0.623
with
Token with
Feature activation+0.277
the
Token the
Feature activation+0.000
Giants
Token Giants
Feature activation+0.000
'
Token'
Feature activation+0.000
AFL
Token AFL
Feature activation+0.000
W
TokenW
Feature activation+0.000

INTERVAL 0.000 - 0.197
CONTAINS 99.849%

the
Token the
Feature activation+0.000
show
Token show
Feature activation+0.000
's
Token's
Feature activation+0.000
central
Token central
Feature activation+0.000
alien
Token alien
Feature activation+0.000
protagonist
Token protagonist
Feature activation+0.000
,
Token,
Feature activation+0.000
he
Token he
Feature activation+0.000
is
Token is
Feature activation+0.000
a
Token a
Feature activation+0.000
very
Token very
Feature activation+0.000
if
Token if
Feature activation+0.000
your
Token your
Feature activation+0.000
theme
Token theme
Feature activation+0.000
is
Token is
Feature activation+0.000
not
Token not
Feature activation+0.000
compatible
Token compatible
Feature activation+0.000
with
Token with
Feature activation+0.000
OCD
Token OCD
Feature activation+0.000
I
TokenI
Feature activation+0.000
(
Token (
Feature activation+0.000
or
Tokenor
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
ll
Tokenll
Feature activation+0.000
continue
Token continue
Feature activation+0.000
to
Token to
Feature activation+0.000
meet
Token meet
Feature activation+0.110
with
Token with
Feature activation+0.000
community
Token community
Feature activation+0.000
members
Token members
Feature activation+0.000
âĢĵ
Token âĢĵ
Feature activation+0.000
that
Token that
Feature activation+0.000
with
Token with
Feature activation+0.000
this
Token this
Feature activation+0.000
new
Token new
Feature activation+0.000
character
Token character
Feature activation+0.000
because
Token because
Feature activation+0.000
it
Token it
Feature activation+0.000
was
Token was
Feature activation+0.000
not
Token not
Feature activation+0.000
interacting
Token interacting
Feature activation+0.000
with
Token with
Feature activation+0.000
the
Token the
Feature activation+0.000
how
Token how
Feature activation+0.000
landmark
Token landmark
Feature activation+0.000
a
Token a
Feature activation+0.000
mod
Token mod
Feature activation+0.000
ding
Tokending
Feature activation+0.000
platform
Token platform
Feature activation+0.000
this
Token this
Feature activation+0.000
game
Token game
Feature activation+0.000
was
Token was
Feature activation+0.000
and
Token and
Feature activation+0.000
is
Token is
Feature activation+0.000

BOTTOM ACTIVATIONS
MIN = 0.000

atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
Development
Token Development
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
<|endoftext|>
Token<|endoftext|>
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000

Top feature 2 in H7.1: (feature 38793)

TOP ACTIVATIONS
MAX = 4.174

Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Perfect
TokenPerfect
Feature activation+0.000
for
Token for
Feature activation+1.119
low
Token low
Feature activation+0.052
light
Token light
Feature activation+4.174
,
Token,
Feature activation+3.568
dawn
Token dawn
Feature activation+3.681
/
Token/
Feature activation+0.000
d
Tokend
Feature activation+0.267
usk
Tokenusk
Feature activation+2.159
Perfect
TokenPerfect
Feature activation+0.000
for
Token for
Feature activation+1.119
low
Token low
Feature activation+0.052
light
Token light
Feature activation+4.174
,
Token,
Feature activation+3.568
dawn
Token dawn
Feature activation+3.681
/
Token/
Feature activation+0.000
d
Tokend
Feature activation+0.267
usk
Tokenusk
Feature activation+2.159
and
Token and
Feature activation+0.000
nighttime
Token nighttime
Feature activation+0.431
gear
Token gear
Feature activation+0.000
made
Token made
Feature activation+0.000
for
Token for
Feature activation+0.000
Starship
Token Starship
Feature activation+0.589
Tro
Token Tro
Feature activation+0.643
opers
Tokenopers
Feature activation+3.677
was
Token was
Feature activation+0.000
re
Token re
Feature activation+0.000
-
Token-
Feature activation+0.000
painted
Tokenpainted
Feature activation+0.000
and
Token and
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Perfect
TokenPerfect
Feature activation+0.000
for
Token for
Feature activation+1.119
low
Token low
Feature activation+0.052
light
Token light
Feature activation+4.174
,
Token,
Feature activation+3.568
dawn
Token dawn
Feature activation+3.681
/
Token/
Feature activation+0.000
d
Tokend
Feature activation+0.267
usk
Tokenusk
Feature activation+2.159
and
Token and
Feature activation+0.000
fun
Token fun
Feature activation+0.000
color
Token color
Feature activation+0.000
g
Token g
Feature activation+0.000
els
Tokenels
Feature activation+0.000
in
Token in
Feature activation+0.000
blue
Token blue
Feature activation+3.319
,
Token,
Feature activation+2.239
red
Token red
Feature activation+2.491
,
Token,
Feature activation+0.374
purple
Token purple
Feature activation+0.000
and
Token and
Feature activation+0.000
and
Token and
Feature activation+0.000
customer
Token customer
Feature activation+0.000
service
Token service
Feature activation+0.000
.
Token.
Feature activation+0.000
The
Token The
Feature activation+0.000
TV
Token TV
Feature activation+3.275
study
Token study
Feature activation+0.000
measures
Token measures
Feature activation+0.000
satisfaction
Token satisfaction
Feature activation+0.000
in
Token in
Feature activation+0.000
those
Token those
Feature activation+0.000
including
Token including
Feature activation+0.000
EM
Token EM
Feature activation+0.000
V
TokenV
Feature activation+0.720
(
Token (
Feature activation+0.631
Euro
TokenEuro
Feature activation+0.215
pay
Tokenpay
Feature activation+2.943
,
Token,
Feature activation+2.016
Master
Token Master
Feature activation+0.656
card
Tokencard
Feature activation+0.215
and
Token and
Feature activation+0.136
Visa
Token Visa
Feature activation+0.711
have
Token have
Feature activation+0.000
business
Token business
Feature activation+0.000
arrangements
Token arrangements
Feature activation+0.000
with
Token with
Feature activation+0.000
Marshall
Token Marshall
Feature activation+1.381
,
Token,
Feature activation+2.775
Carter
Token Carter
Feature activation+0.485
and
Token and
Feature activation+0.000
Dark
Token Dark
Feature activation+0.000
to
Token to
Feature activation+0.000
regularly
Token regularly
Feature activation+0.000
applications
Token applications
Feature activation+0.133
like
Token like
Feature activation+1.157
Firefox
Token Firefox
Feature activation+2.157
,
Token,
Feature activation+1.936
Thunder
Token Thunder
Feature activation+0.463
bird
Tokenbird
Feature activation+2.763
,
Token,
Feature activation+1.483
Google
Token Google
Feature activation+0.299
Chrome
Token Chrome
Feature activation+1.816
,
Token,
Feature activation+0.826
Libre
Token Libre
Feature activation+0.000
characters
Token characters
Feature activation+0.000
on
Token on
Feature activation+0.000
your
Token your
Feature activation+0.000
team
Token team
Feature activation+0.000
.
Token.
Feature activation+0.000
Hero
Token Hero
Feature activation+2.571
and
Token and
Feature activation+0.606
Vill
Token Vill
Feature activation+0.496
ain
Tokenain
Feature activation+0.139
cards
Token cards
Feature activation+0.000
can
Token can
Feature activation+0.000
g
Token g
Feature activation+0.000
els
Tokenels
Feature activation+0.000
in
Token in
Feature activation+0.000
blue
Token blue
Feature activation+3.319
,
Token,
Feature activation+2.239
red
Token red
Feature activation+2.491
,
Token,
Feature activation+0.374
purple
Token purple
Feature activation+0.000
and
Token and
Feature activation+0.000
yellow
Token yellow
Feature activation+0.000
!
Token!
Feature activation+0.000
Store
Token Store
Feature activation+1.693
content
Token content
Feature activation+0.000
.
Token.
Feature activation+0.000
The
TokenThe
Feature activation+0.000
App
Token App
Feature activation+0.000
Store
Token Store
Feature activation+2.473
took
Token took
Feature activation+0.000
a
Token a
Feature activation+0.000
little
Token little
Feature activation+0.000
under
Token under
Feature activation+0.000
two
Token two
Feature activation+0.000
the
Token the
Feature activation+0.000
Steelers
Token Steelers
Feature activation+0.948
over
Token over
Feature activation+0.000
the
Token the
Feature activation+0.000
Rams
Token Rams
Feature activation+0.926
,
Token,
Feature activation+2.447
for
Token for
Feature activation+0.000
whom
Token whom
Feature activation+0.000
he
Token he
Feature activation+0.000
played
Token played
Feature activation+0.000
his
Token his
Feature activation+0.000
of
Token of
Feature activation+0.000
his
Token his
Feature activation+0.163
music
Token music
Feature activation+0.000
.
Token.
Feature activation+0.000
The
Token The
Feature activation+0.000
anger
Token anger
Feature activation+2.346
turned
Token turned
Feature activation+0.000
to
Token to
Feature activation+0.000
disbelief
Token disbelief
Feature activation+0.000
when
Token when
Feature activation+0.000
the
Token the
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Secret
TokenSecret
Feature activation+0.000
aries
Tokenaries
Feature activation+1.965
Of
Token Of
Feature activation+0.226
Defense
Token Defense
Feature activation+2.307
And
Token And
Feature activation+1.229
State
Token State
Feature activation+1.835
,
Token,
Feature activation+2.393
Military
Token Military
Feature activation+0.550
Command
Token Command
Feature activation+0.000
.
Token.
Feature activation+0.000
This
Token This
Feature activation+0.000
time
Token time
Feature activation+0.000
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.000
Row
Token Row
Feature activation+2.294
dies
Tokendies
Feature activation+0.893
faced
Token faced
Feature activation+0.000
their
Token their
Feature activation+0.146
first
Token first
Feature activation+0.000
real
Token real
Feature activation+0.000
,
Token,
Feature activation+1.082
Seven
Token Seven
Feature activation+0.683
-
Token-
Feature activation+0.513
S
TokenS
Feature activation+0.000
ided
Tokenided
Feature activation+0.964
Strike
Token Strike
Feature activation+2.279
,
Token,
Feature activation+0.963
Se
Token Se
Feature activation+1.196
ism
Tokenism
Feature activation+0.177
ic
Tokenic
Feature activation+1.299
Slam
Token Slam
Feature activation+1.990
color
Token color
Feature activation+0.000
g
Token g
Feature activation+0.000
els
Tokenels
Feature activation+0.000
in
Token in
Feature activation+0.000
blue
Token blue
Feature activation+3.319
,
Token,
Feature activation+2.239
red
Token red
Feature activation+2.491
,
Token,
Feature activation+0.374
purple
Token purple
Feature activation+0.000
and
Token and
Feature activation+0.000
yellow
Token yellow
Feature activation+0.000
oss
Tokenoss
Feature activation+0.000
Whedon
Token Whedon
Feature activation+0.000
recycled
Token recycled
Feature activation+0.000
Starship
Token Starship
Feature activation+0.000
Tro
Token Tro
Feature activation+0.000
opers
Tokenopers
Feature activation+2.236
'
Token'
Feature activation+0.000
Federation
Token Federation
Feature activation+0.000
body
Token body
Feature activation+0.000
armor
Token armor
Feature activation+0.000
for
Token for
Feature activation+0.000
of
Token of
Feature activation+0.320
S
Token S
Feature activation+0.974
OD
TokenOD
Feature activation+1.463
1
Token1
Feature activation+1.923
,
Token,
Feature activation+1.733
CAT
Token CAT
Feature activation+2.218
and
Token and
Feature activation+0.888
G
Token G
Feature activation+0.819
SH
TokenSH
Feature activation+0.224
P
TokenP
Feature activation+0.051
-
Token-
Feature activation+0.109

Top DFA by src position
MAX = 4.793

,
Token,
Feature activation-0.137
Top resid features:
dawn
Token dawn
Feature activation+0.212
Top resid features:
/
Token/
Feature activation+0.043
Top resid features:
d
Tokend
Feature activation-0.001
Top resid features:
usk
Tokenusk
Feature activation+0.000
Top resid features:
and
Token and
Feature activation+4.793
Top resid features:
nighttime
Token nighttime
Feature activation-0.002
Top resid features:
shooting
Token shooting
Feature activation-0.025
Top resid features:
.
Token.
Feature activation-0.010
Top resid features:
Get
Token Get
Feature activation+0.001
Top resid features:
artistic
Token artistic
Feature activation-0.003
Top resid features:
,
Token,
Feature activation-0.001
Top resid features:
dawn
Token dawn
Feature activation+0.025
Top resid features:
/
Token/
Feature activation+0.049
Top resid features:
d
Tokend
Feature activation+0.038
Top resid features:
usk
Tokenusk
Feature activation+0.066
Top resid features:
and
Token and
Feature activation+4.118
Top resid features:
nighttime
Token nighttime
Feature activation+0.011
Top resid features:
shooting
Token shooting
Feature activation+0.028
Top resid features:
.
Token.
Feature activation-0.003
Top resid features:
Get
Token Get
Feature activation+0.000
Top resid features:
artistic
Token artistic
Feature activation-0.001
Top resid features:
Ap
Token Ap
Feature activation-0.001
Top resid features:
es
Tokenes
Feature activation+0.111
Top resid features:
(
Token (
Feature activation+0.011
Top resid features:
2001
Token2001
Feature activation-0.033
Top resid features:
),
Token),
Feature activation+0.135
Top resid features:
and
Token and
Feature activation+3.721
Top resid features:
Firefly
Token Firefly
Feature activation+0.100
Top resid features:
Ċ
TokenĊ
Feature activation+0.001
Top resid features:
Ċ
TokenĊ
Feature activation+0.001
Top resid features:
Most
TokenMost
Feature activation-0.012
Top resid features:
of
Token of
Feature activation+0.003
Top resid features:
,
Token,
Feature activation-0.003
Top resid features:
dawn
Token dawn
Feature activation+0.214
Top resid features:
/
Token/
Feature activation+0.267
Top resid features:
d
Tokend
Feature activation-0.008
Top resid features:
usk
Tokenusk
Feature activation-0.017
Top resid features:
and
Token and
Feature activation+3.892
Top resid features:
nighttime
Token nighttime
Feature activation-0.014
Top resid features:
shooting
Token shooting
Feature activation-0.010
Top resid features:
.
Token.
Feature activation-0.025
Top resid features:
Get
Token Get
Feature activation+0.001
Top resid features:
artistic
Token artistic
Feature activation-0.003
Top resid features:
blue
Token blue
Feature activation-0.002
Top resid features:
,
Token,
Feature activation-0.043
Top resid features:
red
Token red
Feature activation+0.051
Top resid features:
,
Token,
Feature activation+0.642
Top resid features:
purple
Token purple
Feature activation+0.014
Top resid features:
and
Token and
Feature activation+3.358
Top resid features:
yellow
Token yellow
Feature activation-0.008
Top resid features:
!
Token!
Feature activation-0.011
Top resid features:
Ċ
TokenĊ
Feature activation+0.001
Top resid features:
Ċ
TokenĊ
Feature activation-0.011
Top resid features:
Transform
TokenTransform
Feature activation-0.006
Top resid features:
experiences
Token experiences
Feature activation-0.002
Top resid features:
with
Token with
Feature activation-0.001
Top resid features:
TV
Token TV
Feature activation-0.004
Top resid features:
,
Token,
Feature activation+0.056
Top resid features:
internet
Token internet
Feature activation+0.175
Top resid features:
and
Token and
Feature activation+3.330
Top resid features:
phone
Token phone
Feature activation+0.044
Top resid features:
services
Token services
Feature activation+0.037
Top resid features:
in
Token in
Feature activation+0.001
Top resid features:
four
Token four
Feature activation+0.002
Top resid features:
geographical
Token geographical
Feature activation+0.003
Top resid features:
Euro
TokenEuro
Feature activation+0.004
Top resid features:
pay
Tokenpay
Feature activation-0.000
Top resid features:
,
Token,
Feature activation+0.197
Top resid features:
Master
Token Master
Feature activation+0.204
Top resid features:
Card
TokenCard
Feature activation+0.172
Top resid features:
and
Token and
Feature activation+2.485
Top resid features:
Visa
Token Visa
Feature activation+0.041
Top resid features:
)
Token)
Feature activation+0.071
Top resid features:
credit
Token credit
Feature activation+0.003
Top resid features:
card
Token card
Feature activation+0.009
Top resid features:
authorization
Token authorization
Feature activation+0.009
Top resid features:
of
Token of
Feature activation-0.005
Top resid features:
interest
Token interest
Feature activation-0.003
Top resid features:
Marshall
Token Marshall
Feature activation-0.004
Top resid features:
,
Token,
Feature activation+0.018
Top resid features:
Carter
Token Carter
Feature activation-0.015
Top resid features:
and
Token and
Feature activation+3.283
Top resid features:
Dark
Token Dark
Feature activation+0.020
Top resid features:
.
Token.
Feature activation+0.002
Top resid features:
Cont
Token Cont
Feature activation-0.001
Top resid features:
ainment
Tokenainment
Feature activation-0.000
Top resid features:
procedures
Token procedures
Feature activation+0.000
Top resid features:
P
Token P
Feature activation+0.019
Top resid features:
id
Tokenid
Feature activation+0.002
Top resid features:
gin
Tokengin
Feature activation+0.001
Top resid features:
Instant
Token Instant
Feature activation+0.002
Top resid features:
Messenger
Token Messenger
Feature activation+0.004
Top resid features:
and
Token and
Feature activation+0.932
Top resid features:
Spotify
Token Spotify
Feature activation+0.031
Top resid features:
.
Token.
Feature activation+0.031
Top resid features:
Ċ
TokenĊ
Feature activation+0.008
Top resid features:
Ċ
TokenĊ
Feature activation+0.008
Top resid features:
They
TokenThey
Feature activation-0.002
Top resid features:
of
Token of
Feature activation+0.018
Top resid features:
Hero
Token Hero
Feature activation+0.000
Top resid features:
,
Token,
Feature activation+0.050
Top resid features:
Vill
Token Vill
Feature activation+0.080
Top resid features:
ain
Tokenain
Feature activation+0.035
Top resid features:
and
Token and
Feature activation+2.777
Top resid features:
Neutral
Token Neutral
Feature activation+0.009
Top resid features:
characters
Token characters
Feature activation+0.014
Top resid features:
on
Token on
Feature activation+0.002
Top resid features:
your
Token your
Feature activation-0.000
Top resid features:
team
Token team
Feature activation-0.001
Top resid features:
blue
Token blue
Feature activation-0.001
Top resid features:
,
Token,
Feature activation-0.012
Top resid features:
red
Token red
Feature activation+0.013
Top resid features:
,
Token,
Feature activation+0.266
Top resid features:
purple
Token purple
Feature activation+0.080
Top resid features:
and
Token and
Feature activation+2.843
Top resid features:
yellow
Token yellow
Feature activation-0.000
Top resid features:
!
Token!
Feature activation+0.000
Top resid features:
Ċ
TokenĊ
Feature activation-0.001
Top resid features:
Ċ
TokenĊ
Feature activation-0.002
Top resid features:
Transform
TokenTransform
Feature activation-0.001
Top resid features:
,
Token,
Feature activation+0.014
Top resid features:
Mac
Token Mac
Feature activation+0.032
Top resid features:
App
Token App
Feature activation+0.002
Top resid features:
Store
Token Store
Feature activation+0.080
Top resid features:
,
Token,
Feature activation+0.720
Top resid features:
or
Token or
Feature activation+2.073
Top resid features:
any
Token any
Feature activation+0.047
Top resid features:
other
Token other
Feature activation+0.032
Top resid features:
iTunes
Token iTunes
Feature activation-0.002
Top resid features:
Store
Token Store
Feature activation+0.001
Top resid features:
content
Token content
Feature activation+0.012
Top resid features:
Rams
Token Rams
Feature activation-0.003
Top resid features:
,
Token,
Feature activation-0.000
Top resid features:
Steelers
Token Steelers
Feature activation+0.337
Top resid features:
,
Token,
Feature activation+0.259
Top resid features:
Panthers
Token Panthers
Feature activation+0.635
Top resid features:
and
Token and
Feature activation+1.362
Top resid features:
49
Token 49
Feature activation+0.315
Top resid features:
ers
Tokeners
Feature activation+0.017
Top resid features:
over
Token over
Feature activation-0.000
Top resid features:
the
Token the
Feature activation-0.000
Top resid features:
course
Token course
Feature activation-0.001
Top resid features:
reacted
Token reacted
Feature activation-0.005
Top resid features:
with
Token with
Feature activation-0.005
Top resid features:
remorse
Token remorse
Feature activation-0.013
Top resid features:
,
Token,
Feature activation-0.013
Top resid features:
outrage
Token outrage
Feature activation+0.105
Top resid features:
and
Token and
Feature activation+2.561
Top resid features:
even
Token even
Feature activation+0.119
Top resid features:
the
Token the
Feature activation+0.069
Top resid features:
rare
Token rare
Feature activation-0.001
Top resid features:
threat
Token threat
Feature activation+0.001
Top resid features:
of
Token of
Feature activation+0.006
Top resid features:
of
Token of
Feature activation+0.003
Top resid features:
Defense
Token Defense
Feature activation+0.015
Top resid features:
Leon
Token Leon
Feature activation+0.190
Top resid features:
Pan
Token Pan
Feature activation+0.068
Top resid features:
etta
Tokenetta
Feature activation+0.745
Top resid features:
,
Token,
Feature activation+0.845
Top resid features:
U
Token U
Feature activation+0.017
Top resid features:
.
Token.
Feature activation+0.000
Top resid features:
S
TokenS
Feature activation+0.001
Top resid features:
.
Token.
Feature activation+0.001
Top resid features:
military
Token military
Feature activation+0.006
Top resid features:
Rays
Token Rays
Feature activation-0.011
Top resid features:
,
Token,
Feature activation+0.002
Top resid features:
Row
Token Row
Feature activation+0.020
Top resid features:
dies
Tokendies
Feature activation+0.931
Top resid features:
,
Token,
Feature activation+0.187
Top resid features:
and
Token and
Feature activation+1.633
Top resid features:
Lightning
Token Lightning
Feature activation+0.046
Top resid features:
all
Token all
Feature activation+0.001
Top resid features:
have
Token have
Feature activation+0.002
Top resid features:
in
Token in
Feature activation-0.000
Top resid features:
common
Token common
Feature activation-0.000
Top resid features:
Seven
Token Seven
Feature activation+0.041
Top resid features:
-
Token-
Feature activation+0.000
Top resid features:
S
TokenS
Feature activation-0.001
Top resid features:
ided
Tokenided
Feature activation+0.001
Top resid features:
Strike
Token Strike
Feature activation+0.013
Top resid features:
,
Token,
Feature activation+1.019
Top resid features:
Se
Token Se
Feature activation+0.208
Top resid features:
ism
Tokenism
Feature activation+0.010
Top resid features:
ic
Tokenic
Feature activation+0.002
Top resid features:
Slam
Token Slam
Feature activation+0.013
Top resid features:
,
Token,
Feature activation+0.844
Top resid features:
blue
Token blue
Feature activation-0.003
Top resid features:
,
Token,
Feature activation-0.009
Top resid features:
red
Token red
Feature activation+0.213
Top resid features:
,
Token,
Feature activation+1.112
Top resid features:
purple
Token purple
Feature activation+0.029
Top resid features:
and
Token and
Feature activation+1.646
Top resid features:
yellow
Token yellow
Feature activation-0.003
Top resid features:
!
Token!
Feature activation-0.005
Top resid features:
Ċ
TokenĊ
Feature activation-0.004
Top resid features:
Ċ
TokenĊ
Feature activation-0.011
Top resid features:
Transform
TokenTransform
Feature activation-0.003
Top resid features:
Ap
Token Ap
Feature activation+0.000
Top resid features:
es
Tokenes
Feature activation+0.033
Top resid features:
(
Token (
Feature activation+0.007
Top resid features:
2001
Token2001
Feature activation-0.044
Top resid features:
),
Token),
Feature activation+0.127
Top resid features:
and
Token and
Feature activation+2.590
Top resid features:
Firefly
Token Firefly
Feature activation+0.044
Top resid features:
Ċ
TokenĊ
Feature activation+0.005
Top resid features:
Ċ
TokenĊ
Feature activation+0.003
Top resid features:
Most
TokenMost
Feature activation-0.001
Top resid features:
of
Token of
Feature activation-0.001
Top resid features:
PS
TokenPS
Feature activation+0.013
Top resid features:
H
TokenH
Feature activation+0.002
Top resid features:
-
Token-
Feature activation+0.003
Top resid features:
x
Tokenx
Feature activation+0.001
Top resid features:
)
Token)
Feature activation+0.033
Top resid features:
and
Token and
Feature activation+1.020
Top resid features:
the
Token the
Feature activation+0.012
Top resid features:
concentration
Token concentration
Feature activation+0.009
Top resid features:
of
Token of
Feature activation+0.001
Top resid features:
mal
Token mal
Feature activation+0.002
Top resid features:
ony
Tokenony
Feature activation-0.001
Top resid features:

Decoder Weights Distribution

Head 0: 0.05

Head 1: 0.50

Head 2: 0.04

Head 3: 0.05

Head 4: 0.04

Head 5: 0.03

Head 6: 0.08

Head 7: 0.03

Head 8: 0.03

Head 9: 0.04

Head 10: 0.09

Head 11: 0.03

Positive logits

&2.12

and2.07

etc2.06

Trend2.06

plus1.99

ggles1.97

or1.93

Ascension1.89

-,1.89

et1.88

aults1.88

:,1.86

Trends1.86

&1.81

nces1.81

1.81

OU1.80

AND1.79

orbit1.78

||1.75

Negative logits

netic-2.10

bos-2.03

symp-2.01

teasp-1.97

iem-1.86

Buckley-1.81

cull-1.80

��-1.76

abdom-1.76

Filipino-1.75

FontSize-1.75

McN-1.73

Moff-1.71

oqu-1.71

disrespect-1.71

disav-1.71

plagiar-1.71

owsky-1.69

wra-1.66

ipation-1.66

INTERVAL 3.757 - 4.174
CONTAINS 0.000%

Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Perfect
TokenPerfect
Feature activation+0.000
for
Token for
Feature activation+1.119
low
Token low
Feature activation+0.052
light
Token light
Feature activation+4.174
,
Token,
Feature activation+3.568
dawn
Token dawn
Feature activation+3.681
/
Token/
Feature activation+0.000
d
Tokend
Feature activation+0.267
usk
Tokenusk
Feature activation+2.159

INTERVAL 3.339 - 3.757
CONTAINS 0.000%

Ċ
TokenĊ
Feature activation+0.000
Perfect
TokenPerfect
Feature activation+0.000
for
Token for
Feature activation+1.119
low
Token low
Feature activation+0.052
light
Token light
Feature activation+4.174
,
Token,
Feature activation+3.568
dawn
Token dawn
Feature activation+3.681
/
Token/
Feature activation+0.000
d
Tokend
Feature activation+0.267
usk
Tokenusk
Feature activation+2.159
and
Token and
Feature activation+0.000
Perfect
TokenPerfect
Feature activation+0.000
for
Token for
Feature activation+1.119
low
Token low
Feature activation+0.052
light
Token light
Feature activation+4.174
,
Token,
Feature activation+3.568
dawn
Token dawn
Feature activation+3.681
/
Token/
Feature activation+0.000
d
Tokend
Feature activation+0.267
usk
Tokenusk
Feature activation+2.159
and
Token and
Feature activation+0.000
nighttime
Token nighttime
Feature activation+0.431
gear
Token gear
Feature activation+0.000
made
Token made
Feature activation+0.000
for
Token for
Feature activation+0.000
Starship
Token Starship
Feature activation+0.589
Tro
Token Tro
Feature activation+0.643
opers
Tokenopers
Feature activation+3.677
was
Token was
Feature activation+0.000
re
Token re
Feature activation+0.000
-
Token-
Feature activation+0.000
painted
Tokenpainted
Feature activation+0.000
and
Token and
Feature activation+0.000

INTERVAL 2.922 - 3.339
CONTAINS 0.000%

and
Token and
Feature activation+0.000
customer
Token customer
Feature activation+0.000
service
Token service
Feature activation+0.000
.
Token.
Feature activation+0.000
The
Token The
Feature activation+0.000
TV
Token TV
Feature activation+3.275
study
Token study
Feature activation+0.000
measures
Token measures
Feature activation+0.000
satisfaction
Token satisfaction
Feature activation+0.000
in
Token in
Feature activation+0.000
those
Token those
Feature activation+0.000
including
Token including
Feature activation+0.000
EM
Token EM
Feature activation+0.000
V
TokenV
Feature activation+0.720
(
Token (
Feature activation+0.631
Euro
TokenEuro
Feature activation+0.215
pay
Tokenpay
Feature activation+2.943
,
Token,
Feature activation+2.016
Master
Token Master
Feature activation+0.656
card
Tokencard
Feature activation+0.215
and
Token and
Feature activation+0.136
Visa
Token Visa
Feature activation+0.711
fun
Token fun
Feature activation+0.000
color
Token color
Feature activation+0.000
g
Token g
Feature activation+0.000
els
Tokenels
Feature activation+0.000
in
Token in
Feature activation+0.000
blue
Token blue
Feature activation+3.319
,
Token,
Feature activation+2.239
red
Token red
Feature activation+2.491
,
Token,
Feature activation+0.374
purple
Token purple
Feature activation+0.000
and
Token and
Feature activation+0.000

INTERVAL 2.504 - 2.922
CONTAINS 0.000%

applications
Token applications
Feature activation+0.133
like
Token like
Feature activation+1.157
Firefox
Token Firefox
Feature activation+2.157
,
Token,
Feature activation+1.936
Thunder
Token Thunder
Feature activation+0.463
bird
Tokenbird
Feature activation+2.763
,
Token,
Feature activation+1.483
Google
Token Google
Feature activation+0.299
Chrome
Token Chrome
Feature activation+1.816
,
Token,
Feature activation+0.826
Libre
Token Libre
Feature activation+0.000
characters
Token characters
Feature activation+0.000
on
Token on
Feature activation+0.000
your
Token your
Feature activation+0.000
team
Token team
Feature activation+0.000
.
Token.
Feature activation+0.000
Hero
Token Hero
Feature activation+2.571
and
Token and
Feature activation+0.606
Vill
Token Vill
Feature activation+0.496
ain
Tokenain
Feature activation+0.139
cards
Token cards
Feature activation+0.000
can
Token can
Feature activation+0.000
have
Token have
Feature activation+0.000
business
Token business
Feature activation+0.000
arrangements
Token arrangements
Feature activation+0.000
with
Token with
Feature activation+0.000
Marshall
Token Marshall
Feature activation+1.381
,
Token,
Feature activation+2.775
Carter
Token Carter
Feature activation+0.485
and
Token and
Feature activation+0.000
Dark
Token Dark
Feature activation+0.000
to
Token to
Feature activation+0.000
regularly
Token regularly
Feature activation+0.000

INTERVAL 2.087 - 2.504
CONTAINS 0.002%

lungs
Token lungs
Feature activation+0.000
.
Token.
Feature activation+0.000
Doctors
Token Doctors
Feature activation+0.000
have
Token have
Feature activation+0.000
stopped
Token stopped
Feature activation+0.000
chem
Token chem
Feature activation+2.093
o
Tokeno
Feature activation+1.416
and
Token and
Feature activation+1.166
placed
Token placed
Feature activation+0.000
him
Token him
Feature activation+0.000
on
Token on
Feature activation+0.000
oss
Tokenoss
Feature activation+0.000
Whedon
Token Whedon
Feature activation+0.000
recycled
Token recycled
Feature activation+0.000
Starship
Token Starship
Feature activation+0.000
Tro
Token Tro
Feature activation+0.000
opers
Tokenopers
Feature activation+2.236
'
Token'
Feature activation+0.000
Federation
Token Federation
Feature activation+0.000
body
Token body
Feature activation+0.000
armor
Token armor
Feature activation+0.000
for
Token for
Feature activation+0.000
The
Token The
Feature activation+0.000
shore
Token shore
Feature activation+0.000
line
Token line
Feature activation+0.000
of
Token of
Feature activation+0.000
North
Token North
Feature activation+0.918
Pond
Token Pond
Feature activation+2.182
is
Token is
Feature activation+0.184
5
Token 5
Feature activation+0.000
.
Token.
Feature activation+0.000
78
Token78
Feature activation+0.000
miles
Token miles
Feature activation+0.000
of
Token of
Feature activation+0.320
S
Token S
Feature activation+0.974
OD
TokenOD
Feature activation+1.463
1
Token1
Feature activation+1.923
,
Token,
Feature activation+1.733
CAT
Token CAT
Feature activation+2.218
and
Token and
Feature activation+0.888
G
Token G
Feature activation+0.819
SH
TokenSH
Feature activation+0.224
P
TokenP
Feature activation+0.051
-
Token-
Feature activation+0.109
the
Token the
Feature activation+0.000
Steelers
Token Steelers
Feature activation+0.948
over
Token over
Feature activation+0.000
the
Token the
Feature activation+0.000
Rams
Token Rams
Feature activation+0.926
,
Token,
Feature activation+2.447
for
Token for
Feature activation+0.000
whom
Token whom
Feature activation+0.000
he
Token he
Feature activation+0.000
played
Token played
Feature activation+0.000
his
Token his
Feature activation+0.000

INTERVAL 1.670 - 2.087
CONTAINS 0.004%

to
Token to
Feature activation+0.000
say
Token say
Feature activation+0.000
it
Token it
Feature activation+0.000
but
Token but
Feature activation+0.000
Meyer
Token Meyer
Feature activation+1.892
,
Token,
Feature activation+2.060
Burns
Token Burns
Feature activation+1.624
,
Token,
Feature activation+1.171
and
Token and
Feature activation+0.731
Miller
Token Miller
Feature activation+1.000
are
Token are
Feature activation+0.000
official
Token official
Feature activation+0.000
endorsement
Token endorsement
Feature activation+0.000
for
Token for
Feature activation+0.000
the
Token the
Feature activation+0.000
new
Token new
Feature activation+0.000
digital
Token digital
Feature activation+1.707
age
Token age
Feature activation+0.281
and
Token and
Feature activation+0.000
a
Token a
Feature activation+0.000
big
Token big
Feature activation+0.000
boost
Token boost
Feature activation+0.000
found
Token found
Feature activation+0.000
that
Token that
Feature activation+0.000
the
Token the
Feature activation+0.000
sub
Token sub
Feature activation+0.557
son
Tokenson
Feature activation+0.536
ic
Tokenic
Feature activation+1.851
and
Token and
Feature activation+1.021
trans
Token trans
Feature activation+0.818
onic
Tokenonic
Feature activation+1.085
results
Token results
Feature activation+0.000
better
Token better
Feature activation+0.000
its
Token its
Feature activation+0.000
widest
Token widest
Feature activation+0.000
point
Token point
Feature activation+0.000
in
Token in
Feature activation+0.000
Middle
Token Middle
Feature activation+0.799
Pond
Token Pond
Feature activation+1.953
,
Token,
Feature activation+1.043
the
Token the
Feature activation+0.000
distance
Token distance
Feature activation+0.000
is
Token is
Feature activation+0.000
1
Token 1
Feature activation+0.000
pocket
Token pocket
Feature activation+0.026
is
Token is
Feature activation+0.000
deep
Token deep
Feature activation+0.000
enough
Token enough
Feature activation+0.000
for
Token for
Feature activation+0.000
sticks
Token sticks
Feature activation+1.841
and
Token and
Feature activation+1.541
leaves
Token leaves
Feature activation+0.893
to
Token to
Feature activation+0.129
catch
Token catch
Feature activation+0.000
,
Token,
Feature activation+0.000

INTERVAL 1.252 - 1.670
CONTAINS 0.009%

fucking
Token fucking
Feature activation+0.000
sick
Token sick
Feature activation+0.000
.
Token.
Feature activation+0.000
Self
Token Self
Feature activation+0.000
-
Token-
Feature activation+0.000
contained
Tokencontained
Feature activation+1.321
,
Token,
Feature activation+0.255
sleek
Token sleek
Feature activation+0.000
and
Token and
Feature activation+0.000
sexy
Token sexy
Feature activation+0.000
.
Token.
Feature activation+0.000
skills
Token skills
Feature activation+0.746
like
Token like
Feature activation+1.331
Arch
Token Arch
Feature activation+0.366
on
Tokenon
Feature activation+2.174
,
Token,
Feature activation+1.971
Stra
Token Stra
Feature activation+1.373
fe
Tokenfe
Feature activation+1.907
,
Token,
Feature activation+1.082
Seven
Token Seven
Feature activation+0.683
-
Token-
Feature activation+0.513
S
TokenS
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
ll
Tokenll
Feature activation+0.000
work
Token work
Feature activation+0.000
with
Token with
Feature activation+0.000
Irvine
Token Irvine
Feature activation+1.542
,
Token,
Feature activation+1.340
Louisville
Token Louisville
Feature activation+1.799
and
Token and
Feature activation+0.775
San
Token San
Feature activation+0.028
Diego
Token Diego
Feature activation+0.386
to
Token to
Feature activation+0.000
,
Token,
Feature activation+0.000
20
Token 20
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.158
June
TokenJune
Feature activation+0.402
22
Token 22
Feature activation+1.475
9
Token 9
Feature activation+0.394
pm
Tokenpm
Feature activation+0.000
The
Token The
Feature activation+0.000
Red
Token Red
Feature activation+0.000
Serpent
Token Serpent
Feature activation+0.000
art
Token art
Feature activation+0.000
history
Token history
Feature activation+0.000
,
Token,
Feature activation+0.004
and
Token and
Feature activation+0.000
philosoph
Token philosoph
Feature activation+0.000
ology
Tokenology
Feature activation+1.316
thrive
Token thrive
Feature activation+0.000
in
Token in
Feature activation+0.000
academic
Token academic
Feature activation+0.000
institutions
Token institutions
Feature activation+0.000
because
Token because
Feature activation+0.000

INTERVAL 0.835 - 1.252
CONTAINS 0.023%

City
Token City
Feature activation+0.000
of
Token of
Feature activation+0.000
Bel
Token Bel
Feature activation+0.000
ur
Tokenur
Feature activation+0.000
via
Token via
Feature activation+0.000
Greek
Token Greek
Feature activation+1.177
β
Token β
Feature activation+0.000
Î
TokenÎ
Feature activation+0.000
®
Token®
Feature activation+0.000
Ïģ
TokenÏģ
Feature activation+0.000
Ï
TokenÏ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
f
Tokenf
Feature activation+0.524
aster
Tokenaster
Feature activation+0.726
performance
Token performance
Feature activation+1.921
,
Token,
Feature activation+2.013
improved
Token improved
Feature activation+0.998
security
Token security
Feature activation+0.713
,
Token,
Feature activation+0.403
convenient
Token convenient
Feature activation+0.013
updates
Token updates
Feature activation+1.443
,
Token,
Feature activation+0.000
ia
Tokenia
Feature activation+0.000
(
Token (
Feature activation+0.000
8
Token8
Feature activation+0.000
c
Tokenc
Feature activation+0.056
+
Token+
Feature activation+0.000
),
Token),
Feature activation+0.953
and
Token and
Feature activation+0.000
last
Token last
Feature activation+0.000
year
Token year
Feature activation+0.000
Gh
Token Gh
Feature activation+0.000
is
Tokenis
Feature activation+0.000
TV
Token TV
Feature activation+0.000
series
Token series
Feature activation+0.000
,
Token,
Feature activation+0.351
video
Token video
Feature activation+0.000
games
Token games
Feature activation+0.932
,
Token,
Feature activation+1.000
movies
Token movies
Feature activation+0.000
,
Token,
Feature activation+0.208
literature
Token literature
Feature activation+0.000
and
Token and
Feature activation+0.000
more
Token more
Feature activation+0.000
e
Tokene
Feature activation+0.000
is
Token is
Feature activation+0.000
simply
Token simply
Feature activation+0.000
choosing
Token choosing
Feature activation+0.000
the
Token the
Feature activation+0.000
Steelers
Token Steelers
Feature activation+0.948
over
Token over
Feature activation+0.000
the
Token the
Feature activation+0.000
Rams
Token Rams
Feature activation+0.926
,
Token,
Feature activation+2.447
for
Token for
Feature activation+0.000

INTERVAL 0.417 - 0.835
CONTAINS 0.079%

ace
Tokenace
Feature activation+2.100
;
Token;
Feature activation+1.070
Star
Token Star
Feature activation+0.412
Wars
Token Wars
Feature activation+0.423
:
Token:
Feature activation+1.228
Attack
Token Attack
Feature activation+0.619
of
Token of
Feature activation+0.022
the
Token the
Feature activation+0.000
Cl
Token Cl
Feature activation+0.098
ones
Tokenones
Feature activation+1.702
;
Token;
Feature activation+0.888
accident
Token accident
Feature activation+0.000
.
Token.
Feature activation+0.000
I
Token I
Feature activation+0.000
saw
Token saw
Feature activation+0.000
his
Token his
Feature activation+0.000
photo
Token photo
Feature activation+0.509
in
Token in
Feature activation+0.000
a
Token a
Feature activation+0.000
newspaper
Token newspaper
Feature activation+0.000
and
Token and
Feature activation+0.000
he
Token he
Feature activation+0.000
away
Token away
Feature activation+0.000
from
Token from
Feature activation+0.000
Russia
Token Russia
Feature activation+0.000
,
Token,
Feature activation+0.000
and
Token and
Feature activation+0.000
Europe
Token Europe
Feature activation+0.495
.
Token.
Feature activation+0.000
Spot
Token Spot
Feature activation+0.000
ters
Tokenters
Feature activation+0.000
suggest
Token suggest
Feature activation+0.000
that
Token that
Feature activation+0.000
water
Token water
Feature activation+0.000
flux
Token flux
Feature activation+0.000
es
Tokenes
Feature activation+0.000
through
Token through
Feature activation+0.000
Mo
Token Mo
Feature activation+0.958
S
TokenS
Feature activation+0.505
2
Token 2
Feature activation+0.088
nanop
Token nanop
Feature activation+0.000
ores
Tokenores
Feature activation+0.000
compared
Token compared
Feature activation+0.000
with
Token with
Feature activation+0.000
de
Token de
Feature activation+0.000
arth
Tokenarth
Feature activation+0.000
of
Token of
Feature activation+0.000
mart
Token mart
Feature activation+0.033
en
Tokenen
Feature activation+0.072
tracks
Token tracks
Feature activation+0.505
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.000
main
Token main
Feature activation+0.000
fur
Token fur
Feature activation+0.000
be
Tokenbe
Feature activation+0.000

INTERVAL 0.000 - 0.417
CONTAINS 99.883%

wash
Tokenwash
Feature activation+0.000
y
Tokeny
Feature activation+0.000
list
Token list
Feature activation+0.000
of
Token of
Feature activation+0.209
touch
Token touch
Feature activation+0.000
y
Tokeny
Feature activation+0.000
-
Token-
Feature activation+0.000
fee
Tokenfee
Feature activation+0.000
ly
Tokenly
Feature activation+0.000
priorities
Token priorities
Feature activation+0.001
âĢĶ
TokenâĢĶ
Feature activation+0.000
anti
Token anti
Feature activation+0.000
-
Token-
Feature activation+0.000
Uber
TokenUber
Feature activation+0.000
protest
Token protest
Feature activation+0.000
in
Token in
Feature activation+0.000
downtown
Token downtown
Feature activation+0.000
Toronto
Token Toronto
Feature activation+0.000
Wednesday
Token Wednesday
Feature activation+0.000
morning
Token morning
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
straight
Token straight
Feature activation+0.000
people
Token people
Feature activation+0.000
who
Token who
Feature activation+0.000
assume
Token assume
Feature activation+0.000
that
Token that
Feature activation+0.000
anti
Token anti
Feature activation+0.000
-
Token-
Feature activation+0.000
LGBT
TokenLGBT
Feature activation+0.000
employment
Token employment
Feature activation+0.000
or
Token or
Feature activation+0.000
housing
Token housing
Feature activation+0.000
izing
Tokenizing
Feature activation+0.000
and
Token and
Feature activation+0.000
perfect
Token perfect
Feature activation+0.000
for
Token for
Feature activation+0.000
ble
Token ble
Feature activation+0.000
ached
Tokenached
Feature activation+0.000
hair
Token hair
Feature activation+0.000
,
Token,
Feature activation+0.000
and
Token and
Feature activation+0.000
you
Token you
Feature activation+0.000
can
Token can
Feature activation+0.000
not
Token not
Feature activation+0.000
so
Token so
Feature activation+0.000
much
Token much
Feature activation+0.000
welfare
Token welfare
Feature activation+0.000
queens
Token queens
Feature activation+0.000
as
Token as
Feature activation+0.000
mat
Token mat
Feature activation+0.000
ri
Tokenri
Feature activation+0.000
arch
Tokenarch
Feature activation+0.000
al
Tokenal
Feature activation+0.000
dyn
Token dyn
Feature activation+0.000

BOTTOM ACTIVATIONS
MIN = 0.000

atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
Development
Token Development
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
<|endoftext|>
Token<|endoftext|>
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000

Top feature 3 in H7.1: (feature 6202)

TOP ACTIVATIONS
MAX = 4.106

the
Token the
Feature activation+0.000
ruling
Token ruling
Feature activation+0.000
an
Token an
Feature activation+0.953
"
Token "
Feature activation+1.155
important
Tokenimportant
Feature activation+3.505
defeat
Token defeat
Feature activation+4.106
for
Token for
Feature activation+3.610
the
Token the
Feature activation+1.401
tobacco
Token tobacco
Feature activation+0.000
industry
Token industry
Feature activation+1.557
."
Token."
Feature activation+0.000
needs
Token needs
Feature activation+0.000
to
Token to
Feature activation+0.000
address
Token address
Feature activation+0.000
"
Token "
Feature activation+1.017
deep
Tokendeep
Feature activation+3.505
and
Token and
Feature activation+3.886
profound
Token profound
Feature activation+3.496
issues
Token issues
Feature activation+1.296
"
Token"
Feature activation+0.000
to
Token to
Feature activation+0.000
keep
Token keep
Feature activation+0.000
must
Token must
Feature activation+0.000
be
Token be
Feature activation+0.000
a
Token a
Feature activation+0.000
"
Token "
Feature activation+1.358
cont
Tokencont
Feature activation+3.208
iguous
Tokeniguous
Feature activation+3.634
territory
Token territory
Feature activation+1.218
"
Token"
Feature activation+0.000
rather
Token rather
Feature activation+0.000
than
Token than
Feature activation+0.000
a
Token a
Feature activation+0.000
ruling
Token ruling
Feature activation+0.000
an
Token an
Feature activation+0.953
"
Token "
Feature activation+1.155
important
Tokenimportant
Feature activation+3.505
defeat
Token defeat
Feature activation+4.106
for
Token for
Feature activation+3.610
the
Token the
Feature activation+1.401
tobacco
Token tobacco
Feature activation+0.000
industry
Token industry
Feature activation+1.557
."
Token."
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
government
Token government
Feature activation+0.000
needs
Token needs
Feature activation+0.000
to
Token to
Feature activation+0.000
address
Token address
Feature activation+0.000
"
Token "
Feature activation+1.017
deep
Tokendeep
Feature activation+3.505
and
Token and
Feature activation+3.886
profound
Token profound
Feature activation+3.496
issues
Token issues
Feature activation+1.296
"
Token"
Feature activation+0.000
to
Token to
Feature activation+0.000
called
Token called
Feature activation+0.861
the
Token the
Feature activation+0.000
ruling
Token ruling
Feature activation+0.000
an
Token an
Feature activation+0.953
"
Token "
Feature activation+1.155
important
Tokenimportant
Feature activation+3.505
defeat
Token defeat
Feature activation+4.106
for
Token for
Feature activation+3.610
the
Token the
Feature activation+1.401
tobacco
Token tobacco
Feature activation+0.000
industry
Token industry
Feature activation+1.557
to
Token to
Feature activation+0.000
address
Token address
Feature activation+0.000
"
Token "
Feature activation+1.017
deep
Tokendeep
Feature activation+3.505
and
Token and
Feature activation+3.886
profound
Token profound
Feature activation+3.496
issues
Token issues
Feature activation+1.296
"
Token"
Feature activation+0.000
to
Token to
Feature activation+0.000
keep
Token keep
Feature activation+0.000
insurance
Token insurance
Feature activation+0.000
,"
Token,"
Feature activation+0.000
said
Token said
Feature activation+2.377
David
Token David
Feature activation+3.173
Bl
Token Bl
Feature activation+2.318
anch
Tokenanch
Feature activation+2.071
ard
Tokenard
Feature activation+3.447
,
Token,
Feature activation+0.827
who
Token who
Feature activation+0.356
is
Token is
Feature activation+0.000
a
Token a
Feature activation+0.000
partner
Token partner
Feature activation+0.000
state
Token state
Feature activation+0.000
must
Token must
Feature activation+0.000
be
Token be
Feature activation+0.000
a
Token a
Feature activation+0.000
"
Token "
Feature activation+1.358
cont
Tokencont
Feature activation+3.208
iguous
Tokeniguous
Feature activation+3.634
territory
Token territory
Feature activation+1.218
"
Token"
Feature activation+0.000
rather
Token rather
Feature activation+0.000
than
Token than
Feature activation+0.000
as
Token as
Feature activation+0.000
we
Token we
Feature activation+0.000
speak
Token speak
Feature activation+0.000
,"
Token,"
Feature activation+0.000
said
Token said
Feature activation+2.377
David
Token David
Feature activation+3.173
Bl
Token Bl
Feature activation+2.318
anch
Tokenanch
Feature activation+2.071
ard
Tokenard
Feature activation+3.447
,
Token,
Feature activation+0.827
who
Token who
Feature activation+0.356
"
Token "
Feature activation+2.032
multi
Tokenmulti
Feature activation+2.515
-
Token-
Feature activation+0.000
p
Tokenp
Feature activation+0.000
olar
Tokenolar
Feature activation+2.418
world
Token world
Feature activation+3.129
"
Token"
Feature activation+1.190
in
Token in
Feature activation+0.000
which
Token which
Feature activation+0.507
Latin
Token Latin
Feature activation+0.000
American
Token American
Feature activation+0.000
that
Token that
Feature activation+0.155
the
Token the
Feature activation+0.000
slots
Token slots
Feature activation+0.000
hall
Token hall
Feature activation+0.000
was
Token was
Feature activation+0.366
moved
Token moved
Feature activation+2.858
from
Token from
Feature activation+1.080
the
Token the
Feature activation+0.200
Casino
Token Casino
Feature activation+0.000
A
Token A
Feature activation+0.000
ac
Tokenac
Feature activation+0.000
securities
Token securities
Feature activation+0.000
regulator
Token regulator
Feature activation+0.000
offers
Token offers
Feature activation+0.653
rewards
Token rewards
Feature activation+2.432
of
Token of
Feature activation+2.752
up
Token up
Feature activation+2.780
to
Token to
Feature activation+2.924
$
Token $
Feature activation+3.227
5
Token5
Feature activation+2.368
million
Token million
Feature activation+1.818
for
Token for
Feature activation+0.428
the
Token the
Feature activation+0.128
securities
Token securities
Feature activation+0.000
regulator
Token regulator
Feature activation+0.000
offers
Token offers
Feature activation+0.653
rewards
Token rewards
Feature activation+2.432
of
Token of
Feature activation+2.752
up
Token up
Feature activation+2.780
to
Token to
Feature activation+2.924
$
Token $
Feature activation+3.227
5
Token5
Feature activation+2.368
million
Token million
Feature activation+1.818
his
Token his
Feature activation+0.000
guilt
Token guilt
Feature activation+0.000
,
Token,
Feature activation+0.090
but
Token but
Feature activation+0.405
fought
Token fought
Feature activation+2.538
to
Token to
Feature activation+2.737
have
Token have
Feature activation+2.398
his
Token his
Feature activation+0.000
sentence
Token sentence
Feature activation+0.926
transferred
Token transferred
Feature activation+2.254
so
Token so
Feature activation+0.391
Ċ
TokenĊ
Feature activation+0.000
"
Token"
Feature activation+0.265
This
TokenThis
Feature activation+0.109
arrest
Token arrest
Feature activation+0.674
is
Token is
Feature activation+2.221
significant
Token significant
Feature activation+2.732
,"
Token,"
Feature activation+0.000
said
Token said
Feature activation+0.000
Sgt
Token Sgt
Feature activation+0.000
.
Token.
Feature activation+0.000
Todd
Token Todd
Feature activation+0.000
the
Token the
Feature activation+0.000
hospital
Token hospital
Feature activation+0.000
this
Token this
Feature activation+0.104
morning
Token morning
Feature activation+0.114
every
Token every
Feature activation+1.814
ward
Token ward
Feature activation+2.705
was
Token was
Feature activation+2.429
full
Token full
Feature activation+2.273
,
Token,
Feature activation+1.277
every
Token every
Feature activation+1.126
escalation
Token escalation
Feature activation+0.000
sets
Token sets
Feature activation+0.173
out
Token out
Feature activation+0.830
an
Token an
Feature activation+0.270
alternative
Token alternative
Feature activation+0.000
menu
Token menu
Feature activation+2.366
of
Token of
Feature activation+2.622
options
Token options
Feature activation+0.375
which
Token which
Feature activation+0.236
he
Token he
Feature activation+0.000
thinks
Token thinks
Feature activation+0.000
might
Token might
Feature activation+0.100
admitted
Token admitted
Feature activation+0.000
his
Token his
Feature activation+0.000
guilt
Token guilt
Feature activation+0.000
,
Token,
Feature activation+0.090
but
Token but
Feature activation+0.405
fought
Token fought
Feature activation+2.538
to
Token to
Feature activation+2.737
have
Token have
Feature activation+2.398
his
Token his
Feature activation+0.000
sentence
Token sentence
Feature activation+0.926
transferred
Token transferred
Feature activation+2.254
what
Token what
Feature activation+0.000
he
Token he
Feature activation+0.000
calls
Token calls
Feature activation+0.000
a
Token a
Feature activation+0.872
"
Token "
Feature activation+2.032
multi
Tokenmulti
Feature activation+2.515
-
Token-
Feature activation+0.000
p
Tokenp
Feature activation+0.000
olar
Tokenolar
Feature activation+2.418
world
Token world
Feature activation+3.129
"
Token"
Feature activation+1.190

Top DFA by src position
MAX = 4.435

defeat
Token defeat
Feature activation+0.005
Top resid features:
for
Token for
Feature activation+0.018
Top resid features:
the
Token the
Feature activation+0.009
Top resid features:
tobacco
Token tobacco
Feature activation-0.003
Top resid features:
industry
Token industry
Feature activation-0.007
Top resid features:
'
Token'
Feature activation+4.335
Top resid features:
Ċ
TokenĊ
Feature activation+0.294
Top resid features:
Ċ
TokenĊ
Feature activation+0.042
Top resid features:
Rob
TokenRob
Feature activation+0.002
Top resid features:
Cunningham
Token Cunningham
Feature activation+0.002
Top resid features:
,
Token,
Feature activation+0.000
Top resid features:
'
Token'
Feature activation-0.000
Top resid features:
Deep
TokenDeep
Feature activation+0.006
Top resid features:
and
Token and
Feature activation+0.002
Top resid features:
profound
Token profound
Feature activation-0.049
Top resid features:
issues
Token issues
Feature activation-0.002
Top resid features:
'
Token'
Feature activation+4.435
Top resid features:
Ċ
TokenĊ
Feature activation+0.158
Top resid features:
Ċ
TokenĊ
Feature activation+0.001
Top resid features:
In
TokenIn
Feature activation+0.000
Top resid features:
the
Token the
Feature activation+0.000
Top resid features:
last
Token last
Feature activation+0.000
Top resid features:
Ċ
TokenĊ
Feature activation-0.000
Top resid features:
'
Token'
Feature activation+0.001
Top resid features:
Cont
TokenCont
Feature activation+0.001
Top resid features:
iguous
Tokeniguous
Feature activation+0.021
Top resid features:
territory
Token territory
Feature activation-0.008
Top resid features:
'
Token'
Feature activation+2.363
Top resid features:
Ċ
TokenĊ
Feature activation+1.828
Top resid features:
Ċ
TokenĊ
Feature activation+0.005
Top resid features:
The
TokenThe
Feature activation-0.000
Top resid features:
US
Token US
Feature activation-0.001
Top resid features:
president
Token president
Feature activation+0.002
Top resid features:
defeat
Token defeat
Feature activation+0.005
Top resid features:
for
Token for
Feature activation+0.003
Top resid features:
the
Token the
Feature activation-0.039
Top resid features:
tobacco
Token tobacco
Feature activation+0.041
Top resid features:
industry
Token industry
Feature activation+0.039
Top resid features:
'
Token'
Feature activation+3.827
Top resid features:
Ċ
TokenĊ
Feature activation+0.287
Top resid features:
Ċ
TokenĊ
Feature activation+0.024
Top resid features:
Rob
TokenRob
Feature activation+0.005
Top resid features:
Cunningham
Token Cunningham
Feature activation+0.003
Top resid features:
,
Token,
Feature activation+0.001
Top resid features:
'
Token'
Feature activation+0.002
Top resid features:
Deep
TokenDeep
Feature activation+0.008
Top resid features:
and
Token and
Feature activation+0.031
Top resid features:
profound
Token profound
Feature activation+0.025
Top resid features:
issues
Token issues
Feature activation-0.000
Top resid features:
'
Token'
Feature activation+3.032
Top resid features:
Ċ
TokenĊ
Feature activation+1.005
Top resid features:
Ċ
TokenĊ
Feature activation+0.005
Top resid features:
In
TokenIn
Feature activation+0.001
Top resid features:
the
Token the
Feature activation+0.000
Top resid features:
last
Token last
Feature activation+0.001
Top resid features:
defeat
Token defeat
Feature activation-0.244
Top resid features:
for
Token for
Feature activation+0.049
Top resid features:
the
Token the
Feature activation+0.006
Top resid features:
tobacco
Token tobacco
Feature activation+0.002
Top resid features:
industry
Token industry
Feature activation+0.000
Top resid features:
'
Token'
Feature activation+3.816
Top resid features:
Ċ
TokenĊ
Feature activation+0.283
Top resid features:
Ċ
TokenĊ
Feature activation+0.058
Top resid features:
Rob
TokenRob
Feature activation+0.001
Top resid features:
Cunningham
Token Cunningham
Feature activation+0.001
Top resid features:
,
Token,
Feature activation+0.000
Top resid features:
'
Token'
Feature activation+0.001
Top resid features:
Deep
TokenDeep
Feature activation+0.006
Top resid features:
and
Token and
Feature activation-0.000
Top resid features:
profound
Token profound
Feature activation-0.002
Top resid features:
issues
Token issues
Feature activation-0.098
Top resid features:
'
Token'
Feature activation+3.391
Top resid features:
Ċ
TokenĊ
Feature activation+0.806
Top resid features:
Ċ
TokenĊ
Feature activation+0.008
Top resid features:
In
TokenIn
Feature activation+0.002
Top resid features:
the
Token the
Feature activation+0.000
Top resid features:
last
Token last
Feature activation+0.001
Top resid features:
-
Token -
Feature activation+0.013
Top resid features:
Dave
Token Dave
Feature activation+0.045
Top resid features:
Bl
Token Bl
Feature activation+0.002
Top resid features:
anch
Tokenanch
Feature activation+0.001
Top resid features:
ard
Tokenard
Feature activation+0.006
Top resid features:
Ċ
TokenĊ
Feature activation+4.009
Top resid features:
Ċ
TokenĊ
Feature activation+0.049
Top resid features:
"
Token"
Feature activation+0.006
Top resid features:
The
TokenThe
Feature activation+0.000
Top resid features:
buildings
Token buildings
Feature activation+0.003
Top resid features:
are
Token are
Feature activation+0.003
Top resid features:
Ċ
TokenĊ
Feature activation-0.000
Top resid features:
'
Token'
Feature activation+0.001
Top resid features:
Cont
TokenCont
Feature activation+0.005
Top resid features:
iguous
Tokeniguous
Feature activation+0.073
Top resid features:
territory
Token territory
Feature activation+0.060
Top resid features:
'
Token'
Feature activation+3.310
Top resid features:
Ċ
TokenĊ
Feature activation+0.394
Top resid features:
Ċ
TokenĊ
Feature activation+0.004
Top resid features:
The
TokenThe
Feature activation-0.000
Top resid features:
US
Token US
Feature activation+0.000
Top resid features:
president
Token president
Feature activation+0.001
Top resid features:
-
Token -
Feature activation+0.003
Top resid features:
Dave
Token Dave
Feature activation+0.008
Top resid features:
Bl
Token Bl
Feature activation+0.010
Top resid features:
anch
Tokenanch
Feature activation+0.005
Top resid features:
ard
Tokenard
Feature activation+0.014
Top resid features:
Ċ
TokenĊ
Feature activation+3.743
Top resid features:
Ċ
TokenĊ
Feature activation+0.013
Top resid features:
"
Token"
Feature activation+0.002
Top resid features:
The
TokenThe
Feature activation-0.002
Top resid features:
buildings
Token buildings
Feature activation+0.002
Top resid features:
are
Token are
Feature activation+0.002
Top resid features:
Multi
TokenMulti
Feature activation-0.000
Top resid features:
-
Token-
Feature activation-0.001
Top resid features:
p
Tokenp
Feature activation-0.000
Top resid features:
olar
Tokenolar
Feature activation+0.001
Top resid features:
world
Token world
Feature activation+0.023
Top resid features:
Ċ
TokenĊ
Feature activation+3.515
Top resid features:
Ċ
TokenĊ
Feature activation+0.116
Top resid features:
Speaking
TokenSpeaking
Feature activation-0.001
Top resid features:
on
Token on
Feature activation-0.000
Top resid features:
state
Token state
Feature activation-0.001
Top resid features:
television
Token television
Feature activation-0.000
Top resid features:
moved
Token moved
Feature activation+0.006
Top resid features:
to
Token to
Feature activation-0.012
Top resid features:
the
Token the
Feature activation-0.010
Top resid features:
city
Token city
Feature activation-0.001
Top resid features:
center
Token center
Feature activation+0.005
Top resid features:
Ċ
TokenĊ
Feature activation+3.359
Top resid features:
Ċ
TokenĊ
Feature activation+0.198
Top resid features:
The
TokenThe
Feature activation-0.003
Top resid features:
first
Token first
Feature activation-0.004
Top resid features:
thing
Token thing
Feature activation+0.001
Top resid features:
I
Token I
Feature activation+0.000
Top resid features:
up
Token up
Feature activation-0.001
Top resid features:
to
Token to
Feature activation+0.031
Top resid features:
$
Token $
Feature activation-0.013
Top resid features:
5
Token5
Feature activation-0.006
Top resid features:
million
Token million
Feature activation+0.022
Top resid features:
Ċ
TokenĊ
Feature activation+3.264
Top resid features:
Ċ
TokenĊ
Feature activation+0.135
Top resid features:
Since
TokenSince
Feature activation-0.001
Top resid features:
its
Token its
Feature activation+0.000
Top resid features:
launch
Token launch
Feature activation+0.001
Top resid features:
in
Token in
Feature activation+0.000
Top resid features:
up
Token up
Feature activation-0.021
Top resid features:
to
Token to
Feature activation-0.011
Top resid features:
$
Token $
Feature activation-0.015
Top resid features:
5
Token5
Feature activation-0.007
Top resid features:
million
Token million
Feature activation+0.099
Top resid features:
Ċ
TokenĊ
Feature activation+3.295
Top resid features:
Ċ
TokenĊ
Feature activation+0.070
Top resid features:
Since
TokenSince
Feature activation-0.000
Top resid features:
its
Token its
Feature activation+0.000
Top resid features:
launch
Token launch
Feature activation+0.001
Top resid features:
in
Token in
Feature activation+0.000
Top resid features:
ought
Tokenought
Feature activation+0.007
Top resid features:
to
Token to
Feature activation+0.003
Top resid features:
have
Token have
Feature activation-0.068
Top resid features:
sentence
Token sentence
Feature activation-0.014
Top resid features:
transferred
Token transferred
Feature activation+0.012
Top resid features:
Ċ
TokenĊ
Feature activation+2.716
Top resid features:
Ċ
TokenĊ
Feature activation+0.484
Top resid features:
From
TokenFrom
Feature activation+0.002
Top resid features:
the
Token the
Feature activation-0.002
Top resid features:
start
Token start
Feature activation+0.020
Top resid features:
,
Token,
Feature activation+0.006
Top resid features:
Ċ
TokenĊ
Feature activation+0.003
Top resid features:
Ċ
TokenĊ
Feature activation-0.001
Top resid features:
'
Token'
Feature activation+0.005
Top resid features:
Sign
TokenSign
Feature activation+0.004
Top resid features:
ificant
Tokenificant
Feature activation+0.017
Top resid features:
'
Token'
Feature activation+2.935
Top resid features:
arrest
Token arrest
Feature activation+0.017
Top resid features:
:
Token:
Feature activation+0.144
Top resid features:
RCMP
Token RCMP
Feature activation-0.003
Top resid features:
Ċ
TokenĊ
Feature activation+0.091
Top resid features:
Ċ
TokenĊ
Feature activation+0.020
Top resid features:
Ċ
TokenĊ
Feature activation+0.000
Top resid features:
'
Token'
Feature activation+0.001
Top resid features:
Every
TokenEvery
Feature activation+0.001
Top resid features:
ward
Token ward
Feature activation+0.010
Top resid features:
full
Token full
Feature activation+0.467
Top resid features:
'
Token'
Feature activation+1.699
Top resid features:
Ċ
TokenĊ
Feature activation+0.959
Top resid features:
Ċ
TokenĊ
Feature activation+0.009
Top resid features:
Mike
TokenMike
Feature activation+0.001
Top resid features:
Pro
Token Pro
Feature activation-0.000
Top resid features:
ctor
Tokenctor
Feature activation+0.000
Top resid features:
Ċ
TokenĊ
Feature activation+0.000
Top resid features:
'
Token'
Feature activation+0.004
Top resid features:
Menu
TokenMenu
Feature activation+0.001
Top resid features:
of
Token of
Feature activation-0.008
Top resid features:
options
Token options
Feature activation+0.433
Top resid features:
'
Token'
Feature activation+2.309
Top resid features:
Ċ
TokenĊ
Feature activation+0.379
Top resid features:
Ċ
TokenĊ
Feature activation+0.007
Top resid features:
As
TokenAs
Feature activation+0.000
Top resid features:
well
Token well
Feature activation+0.001
Top resid features:
as
Token as
Feature activation+0.002
Top resid features:
ought
Tokenought
Feature activation+0.002
Top resid features:
to
Token to
Feature activation+0.015
Top resid features:
have
Token have
Feature activation-0.060
Top resid features:
sentence
Token sentence
Feature activation-0.006
Top resid features:
transferred
Token transferred
Feature activation+0.007
Top resid features:
Ċ
TokenĊ
Feature activation+2.934
Top resid features:
Ċ
TokenĊ
Feature activation+0.206
Top resid features:
From
TokenFrom
Feature activation+0.002
Top resid features:
the
Token the
Feature activation+0.000
Top resid features:
start
Token start
Feature activation+0.016
Top resid features:
,
Token,
Feature activation+0.008
Top resid features:
Multi
TokenMulti
Feature activation+0.000
Top resid features:
-
Token-
Feature activation+0.000
Top resid features:
p
Tokenp
Feature activation-0.023
Top resid features:
olar
Tokenolar
Feature activation+0.003
Top resid features:
world
Token world
Feature activation+0.027
Top resid features:
Ċ
TokenĊ
Feature activation+3.037
Top resid features:
Ċ
TokenĊ
Feature activation+0.039
Top resid features:
Speaking
TokenSpeaking
Feature activation-0.000
Top resid features:
on
Token on
Feature activation-0.000
Top resid features:
state
Token state
Feature activation-0.000
Top resid features:
television
Token television
Feature activation-0.000
Top resid features:

Decoder Weights Distribution

Head 0: 0.05

Head 1: 0.50

Head 2: 0.03

Head 3: 0.06

Head 4: 0.04

Head 5: 0.04

Head 6: 0.06

Head 7: 0.03

Head 8: 0.04

Head 9: 0.08

Head 10: 0.06

Head 11: 0.02

Positive logits

advoc2.20

charact2.08

isSpecialOrderable2.03

encount2.00

territorial1.82

toile1.80

doct1.79

nomine1.79

nutshell1.79

plet1.76

razil1.76

fundament1.68

irrel1.63

defe1.62

understatement1.62

artif1.61

ansk1.61

cod1.58

counterpart1.56

somet1.56

Negative logits

MU-2.04

Hau-1.94

fu-1.86

HUD-1.86

Reborn-1.84

Clin-1.84

Shutdown-1.83

Seller-1.81

Suzuki-1.78

Ru-1.73

Merge-1.71

Cu-1.71

Wast-1.71

bike-1.70

SLI-1.69

Ga-1.69

Bender-1.68

Airl-1.68

vari-1.67

Loan-1.67

INTERVAL 3.696 - 4.106
CONTAINS 0.000%

needs
Token needs
Feature activation+0.000
to
Token to
Feature activation+0.000
address
Token address
Feature activation+0.000
"
Token "
Feature activation+1.017
deep
Tokendeep
Feature activation+3.505
and
Token and
Feature activation+3.886
profound
Token profound
Feature activation+3.496
issues
Token issues
Feature activation+1.296
"
Token"
Feature activation+0.000
to
Token to
Feature activation+0.000
keep
Token keep
Feature activation+0.000
the
Token the
Feature activation+0.000
ruling
Token ruling
Feature activation+0.000
an
Token an
Feature activation+0.953
"
Token "
Feature activation+1.155
important
Tokenimportant
Feature activation+3.505
defeat
Token defeat
Feature activation+4.106
for
Token for
Feature activation+3.610
the
Token the
Feature activation+1.401
tobacco
Token tobacco
Feature activation+0.000
industry
Token industry
Feature activation+1.557
."
Token."
Feature activation+0.000

INTERVAL 3.285 - 3.696
CONTAINS 0.001%

called
Token called
Feature activation+0.861
the
Token the
Feature activation+0.000
ruling
Token ruling
Feature activation+0.000
an
Token an
Feature activation+0.953
"
Token "
Feature activation+1.155
important
Tokenimportant
Feature activation+3.505
defeat
Token defeat
Feature activation+4.106
for
Token for
Feature activation+3.610
the
Token the
Feature activation+1.401
tobacco
Token tobacco
Feature activation+0.000
industry
Token industry
Feature activation+1.557
,"
Token,"
Feature activation+0.000
said
Token said
Feature activation+2.377
David
Token David
Feature activation+3.173
Bl
Token Bl
Feature activation+2.318
anch
Tokenanch
Feature activation+2.071
ard
Tokenard
Feature activation+3.447
,
Token,
Feature activation+0.827
who
Token who
Feature activation+0.356
is
Token is
Feature activation+0.000
a
Token a
Feature activation+0.000
partner
Token partner
Feature activation+0.000
government
Token government
Feature activation+0.000
needs
Token needs
Feature activation+0.000
to
Token to
Feature activation+0.000
address
Token address
Feature activation+0.000
"
Token "
Feature activation+1.017
deep
Tokendeep
Feature activation+3.505
and
Token and
Feature activation+3.886
profound
Token profound
Feature activation+3.496
issues
Token issues
Feature activation+1.296
"
Token"
Feature activation+0.000
to
Token to
Feature activation+0.000
must
Token must
Feature activation+0.000
be
Token be
Feature activation+0.000
a
Token a
Feature activation+0.000
"
Token "
Feature activation+1.358
cont
Tokencont
Feature activation+3.208
iguous
Tokeniguous
Feature activation+3.634
territory
Token territory
Feature activation+1.218
"
Token"
Feature activation+0.000
rather
Token rather
Feature activation+0.000
than
Token than
Feature activation+0.000
a
Token a
Feature activation+0.000
to
Token to
Feature activation+0.000
address
Token address
Feature activation+0.000
"
Token "
Feature activation+1.017
deep
Tokendeep
Feature activation+3.505
and
Token and
Feature activation+3.886
profound
Token profound
Feature activation+3.496
issues
Token issues
Feature activation+1.296
"
Token"
Feature activation+0.000
to
Token to
Feature activation+0.000
keep
Token keep
Feature activation+0.000
insurance
Token insurance
Feature activation+0.000

INTERVAL 2.874 - 3.285
CONTAINS 0.001%

state
Token state
Feature activation+0.000
must
Token must
Feature activation+0.000
be
Token be
Feature activation+0.000
a
Token a
Feature activation+0.000
"
Token "
Feature activation+1.358
cont
Tokencont
Feature activation+3.208
iguous
Tokeniguous
Feature activation+3.634
territory
Token territory
Feature activation+1.218
"
Token"
Feature activation+0.000
rather
Token rather
Feature activation+0.000
than
Token than
Feature activation+0.000
"
Token "
Feature activation+2.032
multi
Tokenmulti
Feature activation+2.515
-
Token-
Feature activation+0.000
p
Tokenp
Feature activation+0.000
olar
Tokenolar
Feature activation+2.418
world
Token world
Feature activation+3.129
"
Token"
Feature activation+1.190
in
Token in
Feature activation+0.000
which
Token which
Feature activation+0.507
Latin
Token Latin
Feature activation+0.000
American
Token American
Feature activation+0.000
as
Token as
Feature activation+0.000
we
Token we
Feature activation+0.000
speak
Token speak
Feature activation+0.000
,"
Token,"
Feature activation+0.000
said
Token said
Feature activation+2.377
David
Token David
Feature activation+3.173
Bl
Token Bl
Feature activation+2.318
anch
Tokenanch
Feature activation+2.071
ard
Tokenard
Feature activation+3.447
,
Token,
Feature activation+0.827
who
Token who
Feature activation+0.356

INTERVAL 2.464 - 2.874
CONTAINS 0.001%

his
Token his
Feature activation+0.000
guilt
Token guilt
Feature activation+0.000
,
Token,
Feature activation+0.090
but
Token but
Feature activation+0.405
fought
Token fought
Feature activation+2.538
to
Token to
Feature activation+2.737
have
Token have
Feature activation+2.398
his
Token his
Feature activation+0.000
sentence
Token sentence
Feature activation+0.926
transferred
Token transferred
Feature activation+2.254
so
Token so
Feature activation+0.391
Democratic
Token Democratic
Feature activation+0.000
National
Token National
Feature activation+0.000
Convention
Token Convention
Feature activation+0.000
were
Token were
Feature activation+0.424
"
Token "
Feature activation+1.001
blown
Tokenblown
Feature activation+2.495
hugely
Token hugely
Feature activation+0.840
out
Token out
Feature activation+1.220
of
Token of
Feature activation+0.329
proportion
Token proportion
Feature activation+0.098
,"
Token,"
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
"
Token"
Feature activation+0.265
This
TokenThis
Feature activation+0.109
arrest
Token arrest
Feature activation+0.674
is
Token is
Feature activation+2.221
significant
Token significant
Feature activation+2.732
,"
Token,"
Feature activation+0.000
said
Token said
Feature activation+0.000
Sgt
Token Sgt
Feature activation+0.000
.
Token.
Feature activation+0.000
Todd
Token Todd
Feature activation+0.000
admitted
Token admitted
Feature activation+0.000
his
Token his
Feature activation+0.000
guilt
Token guilt
Feature activation+0.000
,
Token,
Feature activation+0.090
but
Token but
Feature activation+0.405
fought
Token fought
Feature activation+2.538
to
Token to
Feature activation+2.737
have
Token have
Feature activation+2.398
his
Token his
Feature activation+0.000
sentence
Token sentence
Feature activation+0.926
transferred
Token transferred
Feature activation+2.254
sets
Token sets
Feature activation+0.173
out
Token out
Feature activation+0.830
an
Token an
Feature activation+0.270
alternative
Token alternative
Feature activation+0.000
menu
Token menu
Feature activation+2.366
of
Token of
Feature activation+2.622
options
Token options
Feature activation+0.375
which
Token which
Feature activation+0.236
he
Token he
Feature activation+0.000
thinks
Token thinks
Feature activation+0.000
might
Token might
Feature activation+0.100

INTERVAL 2.053 - 2.464
CONTAINS 0.002%

Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
but
Tokenbut
Feature activation+0.130
declined
Token declined
Feature activation+0.000
to
Token to
Feature activation+0.000
tour
Token tour
Feature activation+2.428
Bethlehem
Token Bethlehem
Feature activation+0.199
[
Token [
Feature activation+0.000
AFP
TokenAFP
Feature activation+0.000
]
Token]
Feature activation+0.000
"
Token "
Feature activation+0.886
is
Token is
Feature activation+0.648
a
Token a
Feature activation+1.164
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
fund
Tokenfund
Feature activation+1.764
amental
Tokenamental
Feature activation+2.257
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
right
Token right
Feature activation+0.242
,
Token,
Feature activation+0.353
deeply
Token deeply
Feature activation+0.012
,
Token,
Feature activation+0.046
the
Token the
Feature activation+0.128
securities
Token securities
Feature activation+0.000
regulator
Token regulator
Feature activation+0.000
offers
Token offers
Feature activation+0.653
rewards
Token rewards
Feature activation+2.432
of
Token of
Feature activation+2.752
up
Token up
Feature activation+2.780
to
Token to
Feature activation+2.924
$
Token $
Feature activation+3.227
5
Token5
Feature activation+2.368
Canadian
Token Canadian
Feature activation+0.000
government
Token government
Feature activation+0.000
developed
Token developed
Feature activation+0.193
a
Token a
Feature activation+0.235
policy
Token policy
Feature activation+0.819
of
Token of
Feature activation+2.168
"
Token "
Feature activation+1.971
aggressive
Tokenaggressive
Feature activation+2.028
ass
Token ass
Feature activation+0.638
imilation
Tokenimilation
Feature activation+0.234
"
Token"
Feature activation+0.631
overrun
Token overrun
Feature activation+0.000
s
Tokens
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Lockheed
Token Lockheed
Feature activation+0.402
Martin
Token Martin
Feature activation+2.137
F
Token F
Feature activation+0.000
-
Token-
Feature activation+0.000
35
Token35
Feature activation+0.000
,
Token,
Feature activation+0.000
I
Token I
Feature activation+0.000

INTERVAL 1.642 - 2.053
CONTAINS 0.003%

I
TokenI
Feature activation+0.000
's
Token's
Feature activation+0.000
definition
Token definition
Feature activation+0.000
of
Token of
Feature activation+0.000
open
Token open
Feature activation+0.382
source
Token source
Feature activation+1.868
software
Token software
Feature activation+1.683
is
Token is
Feature activation+0.960
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
free
Tokenfree
Feature activation+2.052
home
Token home
Feature activation+0.000
is
Token is
Feature activation+0.648
a
Token a
Feature activation+1.164
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
fund
Tokenfund
Feature activation+1.764
amental
Tokenamental
Feature activation+2.257
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
right
Token right
Feature activation+0.242
,
Token,
Feature activation+0.353
.
Token.
Feature activation+0.018
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.015
"
Token"
Feature activation+0.903
That
TokenThat
Feature activation+1.478
's
Token's
Feature activation+1.797
going
Token going
Feature activation+0.000
to
Token to
Feature activation+1.219
be
Token be
Feature activation+0.990
much
Token much
Feature activation+0.063
more
Token more
Feature activation+0.154
tourists
Token tourists
Feature activation+0.000
'
Token'
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
That
TokenThat
Feature activation+1.528
time
Token time
Feature activation+1.793
they
Token they
Feature activation+0.952
said
Token said
Feature activation+0.000
their
Token their
Feature activation+0.000
slogans
Token slogans
Feature activation+0.000
'
Token '
Feature activation+0.464
best
Token best
Feature activation+0.000
analogy
Token analogy
Feature activation+0.000
to
Token to
Feature activation+0.043
understand
Token understand
Feature activation+0.151
the
Token the
Feature activation+0.538
difference
Token difference
Feature activation+1.874
between
Token between
Feature activation+0.000
public
Token public
Feature activation+0.000
and
Token and
Feature activation+0.000
permission
Token permission
Feature activation+0.000
ed
Tokened
Feature activation+0.000

INTERVAL 1.232 - 1.642
CONTAINS 0.005%

a
Token a
Feature activation+0.371
recent
Token recent
Feature activation+0.000
study
Token study
Feature activation+0.000
.
Token.
Feature activation+0.683
This
Token This
Feature activation+0.969
means
Token means
Feature activation+1.356
a
Token a
Feature activation+0.705
black
Token black
Feature activation+0.185
person
Token person
Feature activation+0.372
was
Token was
Feature activation+0.944
killed
Token killed
Feature activation+1.021
peace
Token peace
Feature activation+0.000
process
Token process
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.510
"
Token"
Feature activation+1.597
He
TokenHe
Feature activation+0.000
really
Token really
Feature activation+0.000
wants
Token wants
Feature activation+0.000
to
Token to
Feature activation+0.000
leave
Token leave
Feature activation+0.000
the
Token the
Feature activation+0.000
colon
Token colon
Feature activation+0.000
consists
Token consists
Feature activation+0.000
of
Token of
Feature activation+0.000
five
Token five
Feature activation+0.053
sections
Token sections
Feature activation+1.370
:
Token:
Feature activation+0.000
the
Token the
Feature activation+0.000
c
Token c
Feature activation+0.000
ec
Tokenec
Feature activation+0.000
um
Tokenum
Feature activation+0.000
snapshot
Token snapshot
Feature activation+0.196
for
Token for
Feature activation+0.000
2009
Token 2009
Feature activation+0.000
of
Token of
Feature activation+2.008
the
Token the
Feature activation+1.358
dominance
Token dominance
Feature activation+1.532
of
Token of
Feature activation+0.704
STEM
Token STEM
Feature activation+0.000
degrees
Token degrees
Feature activation+0.000
within
Token within
Feature activation+0.000
the
Token the
Feature activation+0.000
,
Token,
Feature activation+0.000
The
Token The
Feature activation+0.000
Ill
Token Ill
Feature activation+0.097
og
Tokenog
Feature activation+0.000
ic
Tokenic
Feature activation+1.193
of
Token of
Feature activation+1.535
Kas
Token Kas
Feature activation+0.194
sel
Tokensel
Feature activation+0.000
.
Token.
Feature activation+0.000
Photograph
Token Photograph
Feature activation+0.000
:
Token:
Feature activation+0.000

INTERVAL 0.821 - 1.232
CONTAINS 0.020%

we
Token we
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
re
Tokenre
Feature activation+0.000
beginning
Token beginning
Feature activation+0.566
to
Token to
Feature activation+1.122
do
Token do
Feature activation+0.463
just
Token just
Feature activation+0.000
that
Token that
Feature activation+0.000
,
Token,
Feature activation+0.000
with
Token with
Feature activation+0.000
be
Token be
Feature activation+0.000
a
Token a
Feature activation+0.000
"
Token "
Feature activation+1.358
cont
Tokencont
Feature activation+3.208
iguous
Tokeniguous
Feature activation+3.634
territory
Token territory
Feature activation+1.218
"
Token"
Feature activation+0.000
rather
Token rather
Feature activation+0.000
than
Token than
Feature activation+0.000
a
Token a
Feature activation+0.000
patch
Token patch
Feature activation+0.045
blog
Token blog
Feature activation+0.000
post
Token post
Feature activation+0.000
.
Token.
Feature activation+0.187
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
"
Token"
Feature activation+1.023
By
TokenBy
Feature activation+0.915
reducing
Token reducing
Feature activation+0.000
our
Token our
Feature activation+0.000
cost
Token cost
Feature activation+0.000
structure
Token structure
Feature activation+0.000
':
Token':
Feature activation+0.000
Health
Token Health
Feature activation+0.000
Minister
Token Minister
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Health
TokenHealth
Feature activation+0.922
and
Token and
Feature activation+0.000
A
Token A
Feature activation+0.000
ged
Tokenged
Feature activation+0.000
Care
Token Care
Feature activation+0.000
Minister
Token Minister
Feature activation+0.000
Frank
Token Frank
Feature activation+0.000
l
Tokenl
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
7
Token7
Feature activation+0.000
.
Token.
Feature activation+1.097
All
Token All
Feature activation+0.000
changes
Token changes
Feature activation+0.000
,
Token,
Feature activation+0.000
even
Token even
Feature activation+0.000
the
Token the
Feature activation+0.000

INTERVAL 0.411 - 0.821
CONTAINS 0.061%

on
Token on
Feature activation+0.000
Earth
Token Earth
Feature activation+0.000
during
Token during
Feature activation+0.000
her
Token her
Feature activation+0.149
lifetime
Token lifetime
Feature activation+0.000
.
Token.
Feature activation+0.627
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.114
R
TokenR
Feature activation+0.000
ih
Tokenih
Feature activation+0.000
anna
Tokenanna
Feature activation+0.000
,
Token,
Feature activation+0.000
and
Token and
Feature activation+0.000
the
Token the
Feature activation+0.000
NSA
Token NSA
Feature activation+0.000
,
Token,
Feature activation+0.543
thinks
Token thinks
Feature activation+0.667
the
Token the
Feature activation+0.123
US
Token US
Feature activation+0.000
government
Token government
Feature activation+0.000
should
Token should
Feature activation+0.000
stop
Token stop
Feature activation+0.073
Ċ
TokenĊ
Feature activation+0.000
If
TokenIf
Feature activation+0.000
you
Token you
Feature activation+0.127
want
Token want
Feature activation+0.000
to
Token to
Feature activation+0.000
learn
Token learn
Feature activation+0.477
Arabic
Token Arabic
Feature activation+0.000
,
Token,
Feature activation+0.482
this
Token this
Feature activation+0.145
is
Token is
Feature activation+0.000
the
Token the
Feature activation+0.000
in
Token in
Feature activation+0.000
the
Token the
Feature activation+0.000
last
Token last
Feature activation+0.000
year
Token year
Feature activation+0.000
was
Token was
Feature activation+0.463
we
Token we
Feature activation+0.438
had
Token had
Feature activation+0.691
more
Token more
Feature activation+0.590
men
Token men
Feature activation+0.000
getting
Token getting
Feature activation+0.000
on
Token on
Feature activation+0.000
care
Tokencare
Feature activation+0.000
in
Token in
Feature activation+0.000
Old
Token Old
Feature activation+0.000
Crow
Token Crow
Feature activation+0.000
,
Token,
Feature activation+0.000
had
Token had
Feature activation+0.615
been
Token been
Feature activation+0.678
causing
Token causing
Feature activation+0.213
problems
Token problems
Feature activation+0.242
for
Token for
Feature activation+0.690
many
Token many
Feature activation+0.241

INTERVAL 0.000 - 0.411
CONTAINS 99.907%

.
Token.
Feature activation+0.000
Life
Token Life
Feature activation+0.000
was
Token was
Feature activation+0.000
largely
Token largely
Feature activation+0.000
physically
Token physically
Feature activation+0.000
challenging
Token challenging
Feature activation+0.000
,
Token,
Feature activation+0.000
oft
Token oft
Feature activation+0.000
entimes
Tokenentimes
Feature activation+0.000
painful
Token painful
Feature activation+0.000
,
Token,
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
In
TokenIn
Feature activation+0.000
a
Token a
Feature activation+0.000
statement
Token statement
Feature activation+0.000
released
Token released
Feature activation+0.000
Wednesday
Token Wednesday
Feature activation+0.000
,
Token,
Feature activation+0.000
Caroline
Token Caroline
Feature activation+0.000
Fer
Token Fer
Feature activation+0.000
be
Token be
Feature activation+0.000
seen
Token seen
Feature activation+0.000
on
Token on
Feature activation+0.000
the
Token the
Feature activation+0.000
paper
Token paper
Feature activation+0.000
held
Token held
Feature activation+0.000
in
Token in
Feature activation+0.000
Mar
Token Mar
Feature activation+0.000
at
Tokenat
Feature activation+0.000
's
Token's
Feature activation+0.000
left
Token left
Feature activation+0.000
as
Token as
Feature activation+0.000
"
Token "
Feature activation+0.000
one
Tokenone
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
most
Token most
Feature activation+0.000
active
Token active
Feature activation+0.000
"
Token"
Feature activation+0.000
Islamic
Token Islamic
Feature activation+0.000
State
Token State
Feature activation+0.000
execution
Token execution
Feature activation+0.000
but
Token but
Feature activation+0.000
broke
Token broke
Feature activation+0.000
both
Token both
Feature activation+0.000
legs
Token legs
Feature activation+0.000
in
Token in
Feature activation+0.000
a
Token a
Feature activation+0.000
training
Token training
Feature activation+0.000
accident
Token accident
Feature activation+0.000
and
Token and
Feature activation+0.000
was
Token was
Feature activation+0.000
discharged
Token discharged
Feature activation+0.000

BOTTOM ACTIVATIONS
MIN = 0.000

atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
Development
Token Development
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
<|endoftext|>
Token<|endoftext|>
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000

Top feature 4 in H7.1: (feature 673)

TOP ACTIVATIONS
MAX = 2.306

any
Token any
Feature activation+0.655
band
Token band
Feature activation+1.744
in
Token in
Feature activation+0.436
the
Token the
Feature activation+0.318
district
Token district
Feature activation+0.000
that
Token that
Feature activation+2.306
had
Token had
Feature activation+1.972
applied
Token applied
Feature activation+0.361
to
Token to
Feature activation+0.349
participate
Token participate
Feature activation+0.879
in
Token in
Feature activation+0.240
's
Token's
Feature activation+1.024
Overse
Token Overse
Feature activation+0.182
as
Tokenas
Feature activation+0.019
Territories
Token Territories
Feature activation+1.107
'
Token'
Feature activation+1.437
hidden
Token hidden
Feature activation+2.080
role
Token role
Feature activation+0.232
in
Token in
Feature activation+0.000
developing
Token developing
Feature activation+0.000
countries
Token countries
Feature activation+0.000
.
Token.
Feature activation+0.000
band
Token band
Feature activation+1.744
in
Token in
Feature activation+0.436
the
Token the
Feature activation+0.318
district
Token district
Feature activation+0.000
that
Token that
Feature activation+2.306
had
Token had
Feature activation+1.972
applied
Token applied
Feature activation+0.361
to
Token to
Feature activation+0.349
participate
Token participate
Feature activation+0.879
in
Token in
Feature activation+0.240
President
Token President
Feature activation+0.000
:
Token:
Feature activation+0.000
"
Token "
Feature activation+0.000
He
TokenHe
Feature activation+0.953
won
Token won
Feature activation+0.000
't
Token't
Feature activation+1.570
be
Token be
Feature activation+1.839
involved
Token involved
Feature activation+1.160
in
Token in
Feature activation+0.192
the
Token the
Feature activation+0.168
game
Token game
Feature activation+0.251
.
Token.
Feature activation+0.922
was
Token was
Feature activation+0.089
not
Token not
Feature activation+0.000
aware
Token aware
Feature activation+0.037
of
Token of
Feature activation+0.821
any
Token any
Feature activation+0.655
band
Token band
Feature activation+1.744
in
Token in
Feature activation+0.436
the
Token the
Feature activation+0.318
district
Token district
Feature activation+0.000
that
Token that
Feature activation+2.306
had
Token had
Feature activation+1.972
is
Token is
Feature activation+0.000
opposed
Token opposed
Feature activation+0.000
to
Token to
Feature activation+0.000
the
Token the
Feature activation+0.000
return
Token return
Feature activation+0.299
of
Token of
Feature activation+1.605
former
Token former
Feature activation+0.000
strategist
Token strategist
Feature activation+0.000
Frank
Token Frank
Feature activation+0.000
Flan
Token Flan
Feature activation+0.000
nery
Tokennery
Feature activation+0.381
narrower
Token narrower
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Whether
TokenWhether
Feature activation+0.000
Trump
Token Trump
Feature activation+1.601
himself
Token himself
Feature activation+0.640
knew
Token knew
Feature activation+0.307
anything
Token anything
Feature activation+0.000
about
Token about
Feature activation+0.000
the
Token the
Feature activation+0.000
Reuters
TokenReuters
Feature activation+0.000
:
Token:
Feature activation+0.000
"
Token "
Feature activation+0.000
He
TokenHe
Feature activation+0.953
won
Token won
Feature activation+0.000
't
Token't
Feature activation+1.570
be
Token be
Feature activation+1.839
involved
Token involved
Feature activation+1.160
in
Token in
Feature activation+0.192
the
Token the
Feature activation+0.168
game
Token game
Feature activation+0.251
Ļ
TokenĻ
Feature activation+0.000
t
Tokent
Feature activation+0.000
you
Token you
Feature activation+0.000
just
Token just
Feature activation+0.000
want
Token want
Feature activation+0.535
to
Token to
Feature activation+1.499
grab
Token grab
Feature activation+0.414
a
Token a
Feature activation+0.000
few
Token few
Feature activation+0.000
ropes
Token ropes
Feature activation+0.000
and
Token and
Feature activation+0.000
of
Token of
Feature activation+1.063
what
Token what
Feature activation+0.000
will
Token will
Feature activation+0.000
happen
Token happen
Feature activation+0.000
if
Token if
Feature activation+0.000
Denmark
Token Denmark
Feature activation+1.488
joins
Token joins
Feature activation+0.154
the
Token the
Feature activation+0.223
American
Token American
Feature activation+0.000
-
Token-
Feature activation+0.000
controlled
Tokencontrolled
Feature activation+0.000
to
Token to
Feature activation+0.000
run
Token run
Feature activation+0.000
around
Token around
Feature activation+0.000
.
Token.
Feature activation+0.000
I
Token I
Feature activation+0.032
took
Token took
Feature activation+1.466
Z
Token Z
Feature activation+0.000
uz
Tokenuz
Feature activation+0.000
u
Tokenu
Feature activation+0.000
down
Token down
Feature activation+0.000
for
Token for
Feature activation+0.000
date
Token date
Feature activation+0.000
on
Token on
Feature activation+0.000
which
Token which
Feature activation+0.000
Vital
Token Vital
Feature activation+0.000
iy
Tokeniy
Feature activation+0.440
should
Token should
Feature activation+1.442
come
Token come
Feature activation+1.179
to
Token to
Feature activation+0.374
the
Token the
Feature activation+0.000
recruitment
Token recruitment
Feature activation+0.000
office
Token office
Feature activation+0.000
UK
Token UK
Feature activation+1.059
's
Token's
Feature activation+1.024
Overse
Token Overse
Feature activation+0.182
as
Tokenas
Feature activation+0.019
Territories
Token Territories
Feature activation+1.107
'
Token'
Feature activation+1.437
hidden
Token hidden
Feature activation+2.080
role
Token role
Feature activation+0.232
in
Token in
Feature activation+0.000
developing
Token developing
Feature activation+0.000
countries
Token countries
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
S
TokenS
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
war
Tokenwar
Feature activation+0.191
will
Token will
Feature activation+1.436
resign
Token resign
Feature activation+0.096
as
Token as
Feature activation+0.000
Punjab
Token Punjab
Feature activation+0.000
governor
Token governor
Feature activation+0.000
and
Token and
Feature activation+0.020
had
Token had
Feature activation+0.000
started
Token started
Feature activation+0.000
moves
Token moves
Feature activation+0.000
to
Token to
Feature activation+0.000
have
Token have
Feature activation+0.000
him
Token him
Feature activation+1.433
extrad
Token extrad
Feature activation+0.000
ited
Tokenited
Feature activation+0.000
to
Token to
Feature activation+0.000
Kiev
Token Kiev
Feature activation+0.000
where
Token where
Feature activation+0.000
,
Token,
Feature activation+0.000
and
Token and
Feature activation+0.000
I
Token I
Feature activation+0.000
think
Token think
Feature activation+0.000
she
Token she
Feature activation+0.278
should
Token should
Feature activation+1.432
apologize
Token apologize
Feature activation+0.000
to
Token to
Feature activation+0.000
the
Token the
Feature activation+0.000
court
Token court
Feature activation+0.000
.
Token.
Feature activation+0.000
which
Token which
Feature activation+0.000
we
Token we
Feature activation+0.161
could
Token could
Feature activation+0.068
completely
Token completely
Feature activation+0.000
depend
Token depend
Feature activation+1.107
on
Token on
Feature activation+1.417
others
Token others
Feature activation+0.771
are
Token are
Feature activation+0.000
on
Token on
Feature activation+0.000
the
Token the
Feature activation+0.000
way
Token way
Feature activation+0.000
to
Token to
Feature activation+0.000
know
Token know
Feature activation+0.000
the
Token the
Feature activation+0.000
extent
Token extent
Feature activation+0.000
of
Token of
Feature activation+0.013
his
Token his
Feature activation+1.366
participation
Token participation
Feature activation+0.427
,
Token,
Feature activation+0.000
don
Token don
Feature activation+0.000
't
Token't
Feature activation+0.000
read
Token read
Feature activation+0.000
,
Token,
Feature activation+0.000
Apple
Token Apple
Feature activation+0.000
claims
Token claims
Feature activation+0.000
it
Token it
Feature activation+0.734
does
Token does
Feature activation+0.734
not
Token not
Feature activation+1.306
provide
Token provide
Feature activation+0.000
"
Token "
Feature activation+0.000
any
Tokenany
Feature activation+0.000
government
Token government
Feature activation+0.000
agency
Token agency
Feature activation+0.000
-
Token-
Feature activation+0.000
1990
Token1990
Feature activation+0.000
s
Tokens
Feature activation+0.000
,
Token,
Feature activation+0.000
R
Token R
Feature activation+0.179
itchie
Tokenitchie
Feature activation+1.301
was
Token was
Feature activation+0.336
transferred
Token transferred
Feature activation+0.065
to
Token to
Feature activation+0.000
Lu
Token Lu
Feature activation+0.000
cent
Tokencent
Feature activation+0.000

Top DFA by src position
MAX = 2.011

public
Token public
Feature activation+0.000
Top resid features:
school
Token school
Feature activation-0.001
Top resid features:
marching
Token marching
Feature activation-0.001
Top resid features:
band
Token band
Feature activation+0.007
Top resid features:
has
Token has
Feature activation+0.042
Top resid features:
participated
Token participated
Feature activation+1.488
Top resid features:
in
Token in
Feature activation+0.626
Top resid features:
the
Token the
Feature activation+0.164
Top resid features:
past
Token past
Feature activation+0.009
Top resid features:
five
Token five
Feature activation+0.005
Top resid features:
inaugural
Token inaugural
Feature activation+0.004
Top resid features:
Overse
Token Overse
Feature activation+0.006
Top resid features:
as
Tokenas
Feature activation+0.002
Top resid features:
Territories
Token Territories
Feature activation+0.010
Top resid features:
'
Token'
Feature activation+0.001
Top resid features:
hidden
Token hidden
Feature activation+0.007
Top resid features:
role
Token role
Feature activation+1.736
Top resid features:
in
Token in
Feature activation+0.647
Top resid features:
developing
Token developing
Feature activation+0.021
Top resid features:
countries
Token countries
Feature activation+0.007
Top resid features:
Ċ
TokenĊ
Feature activation-0.016
Top resid features:
Ċ
TokenĊ
Feature activation+0.001
Top resid features:
public
Token public
Feature activation+0.000
Top resid features:
school
Token school
Feature activation-0.001
Top resid features:
marching
Token marching
Feature activation+0.001
Top resid features:
band
Token band
Feature activation+0.029
Top resid features:
has
Token has
Feature activation+0.008
Top resid features:
participated
Token participated
Feature activation+1.024
Top resid features:
in
Token in
Feature activation+0.577
Top resid features:
the
Token the
Feature activation+0.156
Top resid features:
past
Token past
Feature activation+0.012
Top resid features:
five
Token five
Feature activation+0.006
Top resid features:
inaugural
Token inaugural
Feature activation+0.004
Top resid features:
ridge
Tokenridge
Feature activation-0.002
Top resid features:
will
Token will
Feature activation-0.002
Top resid features:
not
Token not
Feature activation-0.002
Top resid features:
play
Token play
Feature activation+0.300
Top resid features:
a
Token a
Feature activation+0.027
Top resid features:
part
Token part
Feature activation+0.752
Top resid features:
in
Token in
Feature activation+0.441
Top resid features:
proceedings
Token proceedings
Feature activation+0.028
Top resid features:
as
Token as
Feature activation+0.721
Top resid features:
he
Token he
Feature activation+0.008
Top resid features:
continues
Token continues
Feature activation+0.001
Top resid features:
public
Token public
Feature activation+0.000
Top resid features:
school
Token school
Feature activation-0.003
Top resid features:
marching
Token marching
Feature activation-0.001
Top resid features:
band
Token band
Feature activation+0.019
Top resid features:
has
Token has
Feature activation+0.016
Top resid features:
participated
Token participated
Feature activation+0.872
Top resid features:
in
Token in
Feature activation+0.750
Top resid features:
the
Token the
Feature activation+0.100
Top resid features:
past
Token past
Feature activation+0.015
Top resid features:
five
Token five
Feature activation+0.004
Top resid features:
inaugural
Token inaugural
Feature activation+0.004
Top resid features:
former
Token former
Feature activation+0.028
Top resid features:
strategist
Token strategist
Feature activation-0.007
Top resid features:
Frank
Token Frank
Feature activation+0.001
Top resid features:
Flan
Token Flan
Feature activation+0.001
Top resid features:
nery
Tokennery
Feature activation+0.024
Top resid features:
to
Token to
Feature activation+2.011
Top resid features:
the
Token the
Feature activation+0.028
Top resid features:
party
Token party
Feature activation+0.002
Top resid features:
fold
Token fold
Feature activation+0.019
Top resid features:
.
Token.
Feature activation-0.020
Top resid features:
Ċ
TokenĊ
Feature activation-0.001
Top resid features:
makes
Token makes
Feature activation-0.001
Top resid features:
the
Token the
Feature activation+0.000
Top resid features:
leap
Token leap
Feature activation+0.001
Top resid features:
to
Token to
Feature activation-0.000
Top resid features:
Trump
Token Trump
Feature activation-0.014
Top resid features:
involvement
Token involvement
Feature activation+1.033
Top resid features:
in
Token in
Feature activation+0.798
Top resid features:
the
Token the
Feature activation+0.128
Top resid features:
actual
Token actual
Feature activation-0.001
Top resid features:
Russia
Token Russia
Feature activation+0.003
Top resid features:
-
Token-
Feature activation+0.005
Top resid features:
play
Token play
Feature activation+0.132
Top resid features:
a
Token a
Feature activation+0.032
Top resid features:
part
Token part
Feature activation+0.451
Top resid features:
in
Token in
Feature activation+0.439
Top resid features:
proceedings
Token proceedings
Feature activation+0.024
Top resid features:
as
Token as
Feature activation+0.947
Top resid features:
he
Token he
Feature activation+0.005
Top resid features:
continues
Token continues
Feature activation-0.003
Top resid features:
to
Token to
Feature activation-0.001
Top resid features:
recover
Token recover
Feature activation-0.011
Top resid features:
from
Token from
Feature activation-0.006
Top resid features:
want
Token want
Feature activation-0.027
Top resid features:
to
Token to
Feature activation+0.004
Top resid features:
be
Token be
Feature activation+0.047
Top resid features:
a
Token a
Feature activation+0.195
Top resid features:
part
Token part
Feature activation+0.394
Top resid features:
of
Token of
Feature activation+1.323
Top resid features:
that
Token that
Feature activation+0.087
Top resid features:
.
Token.
Feature activation+0.001
Top resid features:
Really
Token Really
Feature activation-0.005
Top resid features:
,
Token,
Feature activation-0.003
Top resid features:
looking
Token looking
Feature activation+0.001
Top resid features:
a
Token a
Feature activation+0.001
Top resid features:
potential
Token potential
Feature activation+0.003
Top resid features:
nuclear
Token nuclear
Feature activation-0.003
Top resid features:
attack
Token attack
Feature activation-0.006
Top resid features:
by
Token by
Feature activation+0.001
Top resid features:
joining
Token joining
Feature activation+1.697
Top resid features:
Nato
Token Nato
Feature activation+0.029
Top resid features:
âĢ
TokenâĢ
Feature activation+0.002
Top resid features:
Ļ
TokenĻ
Feature activation+0.000
Top resid features:
s
Tokens
Feature activation+0.002
Top resid features:
missile
Token missile
Feature activation-0.006
Top resid features:
a
Token a
Feature activation-0.002
Top resid features:
chance
Token chance
Feature activation+0.010
Top resid features:
to
Token to
Feature activation+0.009
Top resid features:
take
Token take
Feature activation+0.005
Top resid features:
part
Token part
Feature activation+0.420
Top resid features:
in
Token in
Feature activation+1.256
Top resid features:
some
Token some
Feature activation+0.040
Top resid features:
of
Token of
Feature activation+0.058
Top resid features:
the
Token the
Feature activation+0.058
Top resid features:
festivities
Token festivities
Feature activation+0.014
Top resid features:
downtown
Token downtown
Feature activation+0.002
Top resid features:
that
Token that
Feature activation-0.003
Top resid features:
Vital
Token Vital
Feature activation-0.001
Top resid features:
iy
Tokeniy
Feature activation-0.000
Top resid features:
must
Token must
Feature activation-0.003
Top resid features:
be
Token be
Feature activation-0.019
Top resid features:
included
Token included
Feature activation+0.516
Top resid features:
on
Token on
Feature activation+0.499
Top resid features:
the
Token the
Feature activation+0.114
Top resid features:
âĢ
Token âĢ
Feature activation+0.004
Top resid features:
ľ
Tokenľ
Feature activation+0.000
Top resid features:
single
Tokensingle
Feature activation+0.002
Top resid features:
Overse
Token Overse
Feature activation+0.000
Top resid features:
as
Tokenas
Feature activation+0.001
Top resid features:
Territories
Token Territories
Feature activation+0.020
Top resid features:
'
Token'
Feature activation-0.004
Top resid features:
hidden
Token hidden
Feature activation-0.267
Top resid features:
role
Token role
Feature activation+1.367
Top resid features:
in
Token in
Feature activation+0.674
Top resid features:
developing
Token developing
Feature activation+0.028
Top resid features:
countries
Token countries
Feature activation+0.006
Top resid features:
Ċ
TokenĊ
Feature activation-0.006
Top resid features:
Ċ
TokenĊ
Feature activation+0.003
Top resid features:
to
Token to
Feature activation-0.016
Top resid features:
play
Token play
Feature activation+0.164
Top resid features:
a
Token a
Feature activation+0.047
Top resid features:
more
Token more
Feature activation+0.011
Top resid features:
active
Token active
Feature activation+0.108
Top resid features:
role
Token role
Feature activation+1.170
Top resid features:
in
Token in
Feature activation+0.367
Top resid features:
Pakistani
Token Pakistani
Feature activation+0.004
Top resid features:
politics
Token politics
Feature activation+0.011
Top resid features:
from
Token from
Feature activation+0.014
Top resid features:
the
Token the
Feature activation+0.013
Top resid features:
he
Token he
Feature activation-0.002
Top resid features:
ruled
Token ruled
Feature activation-0.022
Top resid features:
out
Token out
Feature activation-0.025
Top resid features:
taking
Token taking
Feature activation+0.022
Top resid features:
part
Token part
Feature activation+0.597
Top resid features:
in
Token in
Feature activation+1.103
Top resid features:
elections
Token elections
Feature activation+0.117
Top resid features:
planned
Token planned
Feature activation+0.031
Top resid features:
for
Token for
Feature activation+0.010
Top resid features:
25
Token 25
Feature activation+0.002
Top resid features:
May
Token May
Feature activation+0.008
Top resid features:
States
Token States
Feature activation+0.000
Top resid features:
Supreme
Token Supreme
Feature activation+0.002
Top resid features:
Court
Token Court
Feature activation+0.003
Top resid features:
judge
Token judge
Feature activation+0.004
Top resid features:
gets
Token gets
Feature activation+0.167
Top resid features:
involved
Token involved
Feature activation+1.409
Top resid features:
in
Token in
Feature activation+0.162
Top resid features:
a
Token a
Feature activation+0.012
Top resid features:
political
Token political
Feature activation+0.009
Top resid features:
campaign
Token campaign
Feature activation+0.010
Top resid features:
,
Token,
Feature activation+0.004
Top resid features:
no
Token no
Feature activation-0.001
Top resid features:
longer
Token longer
Feature activation+0.007
Top resid features:
rely
Token rely
Feature activation+0.003
Top resid features:
on
Token on
Feature activation+0.003
Top resid features:
America
Token America
Feature activation-0.008
Top resid features:
in
Token in
Feature activation+1.639
Top resid features:
allied
Token allied
Feature activation+0.115
Top resid features:
foreign
Token foreign
Feature activation+0.003
Top resid features:
diplomacy
Token diplomacy
Feature activation+0.046
Top resid features:
.
Token.
Feature activation+0.001
Top resid features:
âĢ
Token âĢ
Feature activation+0.000
Top resid features:
Joker
Token Joker
Feature activation-0.008
Top resid features:
âĢ
TokenâĢ
Feature activation+0.008
Top resid features:
Ļ
TokenĻ
Feature activation-0.002
Top resid features:
s
Tokens
Feature activation-0.011
Top resid features:
involvement
Token involvement
Feature activation+0.630
Top resid features:
in
Token in
Feature activation+1.093
Top resid features:
Suicide
Token Suicide
Feature activation-0.014
Top resid features:
Squad
Token Squad
Feature activation+0.016
Top resid features:
,
Token,
Feature activation+0.036
Top resid features:
so
Token so
Feature activation+0.010
Top resid features:
if
Token if
Feature activation+0.004
Top resid features:
it
Token it
Feature activation+0.002
Top resid features:
has
Token has
Feature activation-0.003
Top resid features:
not
Token not
Feature activation-0.000
Top resid features:
been
Token been
Feature activation+0.009
Top resid features:
involved
Token involved
Feature activation+0.632
Top resid features:
in
Token in
Feature activation+0.719
Top resid features:
the
Token the
Feature activation+0.386
Top resid features:
NSA
Token NSA
Feature activation-0.001
Top resid features:
's
Token's
Feature activation+0.006
Top resid features:
PR
Token PR
Feature activation+0.004
Top resid features:
ISM
TokenISM
Feature activation+0.004
Top resid features:
Ċ
TokenĊ
Feature activation+0.007
Top resid features:
R
TokenR
Feature activation-0.001
Top resid features:
itchie
Tokenitchie
Feature activation-0.011
Top resid features:
was
Token was
Feature activation+0.031
Top resid features:
also
Token also
Feature activation+0.054
Top resid features:
involved
Token involved
Feature activation+1.198
Top resid features:
with
Token with
Feature activation+0.286
Top resid features:
the
Token the
Feature activation+0.048
Top resid features:
development
Token development
Feature activation-0.009
Top resid features:
of
Token of
Feature activation-0.000
Top resid features:
the
Token the
Feature activation-0.003
Top resid features:

Decoder Weights Distribution

Head 0: 0.03

Head 1: 0.50

Head 2: 0.02

Head 3: 0.07

Head 4: 0.02

Head 5: 0.06

Head 6: 0.07

Head 7: 0.05

Head 8: 0.05

Head 9: 0.05

Head 10: 0.05

Head 11: 0.02

Positive logits

cameo3.68

participation3.67

Volunte3.56

participate3.52

participating3.49

volunteering3.43

participates3.41

participated3.32

Joined3.27

involvement3.14

represented3.12

presence3.07

Participation3.01

contributions3.00

role2.97

membership2.92

partake2.88

contribution2.84

volunteered2.83

particip2.83

Negative logits

essim-2.70

panic-2.60

decre-2.46

Doomsday-2.42

tightening-2.33

forecasting-2.32

Fib-2.29

Prices-2.28

rx-2.24

estimating-2.24

omsday-2.22

Problems-2.21

spiral-2.16

formulas-2.12

dict-2.09

inflation-2.07

dams-2.07

Despair-2.05

miscar-2.05

despair-2.04

INTERVAL 2.076 - 2.306
CONTAINS 0.000%

any
Token any
Feature activation+0.655
band
Token band
Feature activation+1.744
in
Token in
Feature activation+0.436
the
Token the
Feature activation+0.318
district
Token district
Feature activation+0.000
that
Token that
Feature activation+2.306
had
Token had
Feature activation+1.972
applied
Token applied
Feature activation+0.361
to
Token to
Feature activation+0.349
participate
Token participate
Feature activation+0.879
in
Token in
Feature activation+0.240
's
Token's
Feature activation+1.024
Overse
Token Overse
Feature activation+0.182
as
Tokenas
Feature activation+0.019
Territories
Token Territories
Feature activation+1.107
'
Token'
Feature activation+1.437
hidden
Token hidden
Feature activation+2.080
role
Token role
Feature activation+0.232
in
Token in
Feature activation+0.000
developing
Token developing
Feature activation+0.000
countries
Token countries
Feature activation+0.000
.
Token.
Feature activation+0.000

INTERVAL 1.845 - 2.076
CONTAINS 0.000%

band
Token band
Feature activation+1.744
in
Token in
Feature activation+0.436
the
Token the
Feature activation+0.318
district
Token district
Feature activation+0.000
that
Token that
Feature activation+2.306
had
Token had
Feature activation+1.972
applied
Token applied
Feature activation+0.361
to
Token to
Feature activation+0.349
participate
Token participate
Feature activation+0.879
in
Token in
Feature activation+0.240
President
Token President
Feature activation+0.000

INTERVAL 1.614 - 1.845
CONTAINS 0.000%

was
Token was
Feature activation+0.089
not
Token not
Feature activation+0.000
aware
Token aware
Feature activation+0.037
of
Token of
Feature activation+0.821
any
Token any
Feature activation+0.655
band
Token band
Feature activation+1.744
in
Token in
Feature activation+0.436
the
Token the
Feature activation+0.318
district
Token district
Feature activation+0.000
that
Token that
Feature activation+2.306
had
Token had
Feature activation+1.972
:
Token:
Feature activation+0.000
"
Token "
Feature activation+0.000
He
TokenHe
Feature activation+0.953
won
Token won
Feature activation+0.000
't
Token't
Feature activation+1.570
be
Token be
Feature activation+1.839
involved
Token involved
Feature activation+1.160
in
Token in
Feature activation+0.192
the
Token the
Feature activation+0.168
game
Token game
Feature activation+0.251
.
Token.
Feature activation+0.922

INTERVAL 1.384 - 1.614
CONTAINS 0.001%

âĢ
TokenâĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
S
TokenS
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
war
Tokenwar
Feature activation+0.191
will
Token will
Feature activation+1.436
resign
Token resign
Feature activation+0.096
as
Token as
Feature activation+0.000
Punjab
Token Punjab
Feature activation+0.000
governor
Token governor
Feature activation+0.000
and
Token and
Feature activation+0.020
UK
Token UK
Feature activation+1.059
's
Token's
Feature activation+1.024
Overse
Token Overse
Feature activation+0.182
as
Tokenas
Feature activation+0.019
Territories
Token Territories
Feature activation+1.107
'
Token'
Feature activation+1.437
hidden
Token hidden
Feature activation+2.080
role
Token role
Feature activation+0.232
in
Token in
Feature activation+0.000
developing
Token developing
Feature activation+0.000
countries
Token countries
Feature activation+0.000
Reuters
TokenReuters
Feature activation+0.000
:
Token:
Feature activation+0.000
"
Token "
Feature activation+0.000
He
TokenHe
Feature activation+0.953
won
Token won
Feature activation+0.000
't
Token't
Feature activation+1.570
be
Token be
Feature activation+1.839
involved
Token involved
Feature activation+1.160
in
Token in
Feature activation+0.192
the
Token the
Feature activation+0.168
game
Token game
Feature activation+0.251
which
Token which
Feature activation+0.000
we
Token we
Feature activation+0.161
could
Token could
Feature activation+0.068
completely
Token completely
Feature activation+0.000
depend
Token depend
Feature activation+1.107
on
Token on
Feature activation+1.417
others
Token others
Feature activation+0.771
are
Token are
Feature activation+0.000
on
Token on
Feature activation+0.000
the
Token the
Feature activation+0.000
way
Token way
Feature activation+0.000
narrower
Token narrower
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Whether
TokenWhether
Feature activation+0.000
Trump
Token Trump
Feature activation+1.601
himself
Token himself
Feature activation+0.640
knew
Token knew
Feature activation+0.307
anything
Token anything
Feature activation+0.000
about
Token about
Feature activation+0.000
the
Token the
Feature activation+0.000

INTERVAL 1.153 - 1.384
CONTAINS 0.001%

,
Token,
Feature activation+0.000
confirmed
Token confirmed
Feature activation+0.000
that
Token that
Feature activation+0.111
the
Token the
Feature activation+0.000
State
Token State
Feature activation+0.000
Department
Token Department
Feature activation+1.217
,
Token,
Feature activation+0.187
in
Token in
Feature activation+0.000
consultation
Token consultation
Feature activation+0.000
with
Token with
Feature activation+0.000
the
Token the
Feature activation+0.000
they
Token they
Feature activation+0.000
wrote
Token wrote
Feature activation+0.000
.
Token.
Feature activation+0.000
Britain
Token Britain
Feature activation+0.587
and
Token and
Feature activation+0.255
France
Token France
Feature activation+1.236
are
Token are
Feature activation+0.733
the
Token the
Feature activation+0.097
EU
Token EU
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
to
Token to
Feature activation+0.486
return
Token return
Feature activation+0.545
to
Token to
Feature activation+0.000
Madison
Token Madison
Feature activation+0.000
to
Token to
Feature activation+0.784
continue
Token continue
Feature activation+1.238
protesting
Token protesting
Feature activation+0.000
this
Token this
Feature activation+0.000
coming
Token coming
Feature activation+0.000
weekend
Token weekend
Feature activation+0.000
.
Token.
Feature activation+0.000
his
Token his
Feature activation+0.000
offense
Token offense
Feature activation+0.000
,
Token,
Feature activation+0.000
I
Token I
Feature activation+0.109
would
Token would
Feature activation+0.000
love
Token love
Feature activation+1.220
to
Token to
Feature activation+1.099
play
Token play
Feature activation+0.024
for
Token for
Feature activation+0.000
him
Token him
Feature activation+0.000
and
Token and
Feature activation+0.000
to
Token to
Feature activation+0.000
know
Token know
Feature activation+0.000
the
Token the
Feature activation+0.000
extent
Token extent
Feature activation+0.000
of
Token of
Feature activation+0.013
his
Token his
Feature activation+1.366
participation
Token participation
Feature activation+0.427
,
Token,
Feature activation+0.000
don
Token don
Feature activation+0.000
't
Token't
Feature activation+0.000
read
Token read
Feature activation+0.000

INTERVAL 0.923 - 1.153
CONTAINS 0.005%

appeared
Token appeared
Feature activation+0.000
on
Token on
Feature activation+0.000
,
Token,
Feature activation+0.000
which
Token which
Feature activation+0.000
villain
Token villain
Feature activation+0.000
it
Token it
Feature activation+1.012
belonged
Token belonged
Feature activation+0.115
to
Token to
Feature activation+0.000
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.000
size
Token size
Feature activation+0.000
City
Token City
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
President
TokenPresident
Feature activation+0.000
Trump
Token Trump
Feature activation+0.726
arrived
Token arrived
Feature activation+1.141
in
Token in
Feature activation+0.000
China
Token China
Feature activation+0.000
on
Token on
Feature activation+0.145
the
Token the
Feature activation+0.000
third
Token third
Feature activation+0.000
to
Token to
Feature activation+0.084
collect
Token collect
Feature activation+0.000
the
Token the
Feature activation+0.000
tax
Token tax
Feature activation+0.000
if
Token if
Feature activation+0.000
they
Token they
Feature activation+1.055
sell
Token sell
Feature activation+0.000
more
Token more
Feature activation+0.000
than
Token than
Feature activation+0.000
$
Token $
Feature activation+0.000
100
Token100
Feature activation+0.000
directly
Token directly
Feature activation+0.000
involved
Token involved
Feature activation+0.000
in
Token in
Feature activation+0.000
that
Token that
Feature activation+0.000
but
Token but
Feature activation+0.000
I
Token I
Feature activation+1.110
am
Token am
Feature activation+0.783
aware
Token aware
Feature activation+0.001
his
Token his
Feature activation+0.076
status
Token status
Feature activation+0.000
is
Token is
Feature activation+0.000
to
Token to
Feature activation+0.000
not
Token not
Feature activation+0.000
just
Token just
Feature activation+0.000
highlight
Token highlight
Feature activation+0.000
the
Token the
Feature activation+0.386
contribution
Token contribution
Feature activation+1.099
made
Token made
Feature activation+0.000
by
Token by
Feature activation+0.221
different
Token different
Feature activation+0.000
ethnic
Token ethnic
Feature activation+0.000
communities
Token communities
Feature activation+0.000

INTERVAL 0.692 - 0.923
CONTAINS 0.010%

would
Token would
Feature activation+0.000
likely
Token likely
Feature activation+0.000
prohibit
Token prohibit
Feature activation+0.433
that
Token that
Feature activation+0.469
kind
Token kind
Feature activation+0.369
of
Token of
Feature activation+0.713
decision
Token decision
Feature activation+0.053
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.017
âĢ
TokenâĢ
Feature activation+0.000
Now
TokenNow
Feature activation+0.000
,
Token,
Feature activation+0.000
as
Token as
Feature activation+0.000
to
Token to
Feature activation+0.000
why
Token why
Feature activation+0.000
I
Token I
Feature activation+0.857
am
Token am
Feature activation+0.471
here
Token here
Feature activation+0.503
now
Token now
Feature activation+0.038
.
Token.
Feature activation+0.000
As
Token As
Feature activation+0.000
.
Token.
Feature activation+0.255
The
Token The
Feature activation+0.000
United
Token United
Feature activation+0.000
States
Token States
Feature activation+0.000
has
Token has
Feature activation+0.382
also
Token also
Feature activation+0.715
rejected
Token rejected
Feature activation+0.109
participating
Token participating
Feature activation+0.748
in
Token in
Feature activation+0.000
the
Token the
Feature activation+0.000
court
Token court
Feature activation+0.000
would
Token would
Feature activation+0.016
be
Token be
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
better
Tokenbetter
Feature activation+0.000
off
Token off
Feature activation+0.880
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
outside
Token outside
Feature activation+0.120
of
Token of
Feature activation+0.154
the
Token the
Feature activation+0.108
0
Token 0
Feature activation+0.000
;
Token;
Feature activation+0.000
if
Token if
Feature activation+0.000
the
Token the
Feature activation+0.000
card
Token card
Feature activation+0.640
is
Token is
Feature activation+0.824
present
Token present
Feature activation+0.239
then
Token then
Feature activation+0.000
it
Token it
Feature activation+0.000
represents
Token represents
Feature activation+0.000
a
Token a
Feature activation+0.000

INTERVAL 0.461 - 0.692
CONTAINS 0.025%

detailing
Token detailing
Feature activation+0.000
how
Token how
Feature activation+0.000
POS
Token POS
Feature activation+0.000
RE
Token RE
Feature activation+0.000
P
TokenP
Feature activation+0.271
was
Token was
Feature activation+0.468
instrumental
Token instrumental
Feature activation+0.191
in
Token in
Feature activation+0.000
a
Token a
Feature activation+0.000
suicide
Token suicide
Feature activation+0.000
intervention
Token intervention
Feature activation+0.000
letter
Token letter
Feature activation+0.000
,
Token,
Feature activation+0.458
lawmakers
Token lawmakers
Feature activation+0.000
want
Token want
Feature activation+0.000
congressional
Token congressional
Feature activation+0.000
staff
Token staff
Feature activation+0.521
to
Token to
Feature activation+0.137
interview
Token interview
Feature activation+0.000
four
Token four
Feature activation+0.118
top
Token top
Feature activation+0.108
EPA
Token EPA
Feature activation+0.000
One
Token One
Feature activation+0.000
,
Token,
Feature activation+0.000
places
Token places
Feature activation+0.000
where
Token where
Feature activation+0.000
he
Token he
Feature activation+0.144
seems
Token seems
Feature activation+0.614
to
Token to
Feature activation+0.249
feel
Token feel
Feature activation+0.000
comfortable
Token comfortable
Feature activation+0.022
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
,
Token,
Feature activation+0.000
we
Token we
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
re
Tokenre
Feature activation+0.000
embracing
Token embracing
Feature activation+0.477
the
Token the
Feature activation+0.000
digital
Token digital
Feature activation+0.000
revolution
Token revolution
Feature activation+0.000
,
Token,
Feature activation+0.000
exploring
Token exploring
Feature activation+0.000
any
Token any
Feature activation+0.139
as
Token as
Feature activation+0.071
p
Tokenp
Feature activation+0.000
ersion
Tokenersion
Feature activation+0.523
should
Token should
Feature activation+0.299
have
Token have
Feature activation+0.512
been
Token been
Feature activation+0.160
cast
Token cast
Feature activation+0.000
on
Token on
Feature activation+0.097
her
Token her
Feature activation+0.547
gender
Token gender
Feature activation+0.039

INTERVAL 0.231 - 0.461
CONTAINS 0.084%

for
Token for
Feature activation+0.000
Trump
Token Trump
Feature activation+0.651
to
Token to
Feature activation+0.650
put
Token put
Feature activation+0.183
distance
Token distance
Feature activation+0.000
between
Token between
Feature activation+0.281
himself
Token himself
Feature activation+0.391
and
Token and
Feature activation+0.417
his
Token his
Feature activation+0.000
own
Token own
Feature activation+0.000
son
Token son
Feature activation+0.000
vote
Token vote
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Will
TokenWill
Feature activation+0.000
Scotland
Token Scotland
Feature activation+0.234
still
Token still
Feature activation+0.000
use
Token use
Feature activation+0.000
sterling
Token sterling
Feature activation+0.000
?
Token?
Feature activation+0.000
And
Token And
Feature activation+0.000
but
Token but
Feature activation+0.000
they
Token they
Feature activation+0.098
may
Token may
Feature activation+0.166
not
Token not
Feature activation+0.062
be
Token be
Feature activation+0.028
out
Token out
Feature activation+0.438
of
Token of
Feature activation+0.151
harm
Token harm
Feature activation+0.000
's
Token's
Feature activation+0.000
way
Token way
Feature activation+0.000
yet
Token yet
Feature activation+0.000
the
Token the
Feature activation+0.000
first
Token first
Feature activation+0.000
U
Token U
Feature activation+0.000
.
Token.
Feature activation+0.000
S
TokenS
Feature activation+0.000
.
Token.
Feature activation+0.413
President
Token President
Feature activation+0.904
to
Token to
Feature activation+0.417
be
Token be
Feature activation+0.878
given
Token given
Feature activation+1.008
a
Token a
Feature activation+0.653
highly
Token highly
Feature activation+0.000
questionable
Token questionable
Feature activation+0.000
offensive
Token offensive
Feature activation+0.000
line
Token line
Feature activation+0.000
.
Token.
Feature activation+0.000
Smith
Token Smith
Feature activation+0.245
joined
Token joined
Feature activation+0.000
the
Token the
Feature activation+0.000
Minnesota
Token Minnesota
Feature activation+0.000
Vikings
Token Vikings
Feature activation+0.000
in
Token in
Feature activation+0.000

INTERVAL 0.000 - 0.231
CONTAINS 99.873%

although
Token although
Feature activation+0.000
the
Token the
Feature activation+0.000
FBI
Token FBI
Feature activation+0.000
wouldn
Token wouldn
Feature activation+0.000
't
Token't
Feature activation+0.000
confirm
Token confirm
Feature activation+0.000
for
Token for
Feature activation+0.000
the
Token the
Feature activation+0.000
paper
Token paper
Feature activation+0.000
that
Token that
Feature activation+0.000
he
Token he
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Original
TokenOriginal
Feature activation+0.000
estimates
Token estimates
Feature activation+0.000
put
Token put
Feature activation+0.000
the
Token the
Feature activation+0.000
amount
Token amount
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
supposedly
Token supposedly
Feature activation+0.000
harmless
Token harmless
Feature activation+0.000
coal
Token coal
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
South
TokenSouth
Feature activation+0.000
Asians
Token Asians
Feature activation+0.000
dominate
Token dominate
Feature activation+0.000
the
Token the
Feature activation+0.000
taxi
Token taxi
Feature activation+0.000
industry
Token industry
Feature activation+0.000
and
Token and
Feature activation+0.000
the
Token the
Feature activation+0.000
Uber
Token Uber
Feature activation+0.000
-
Token -
Feature activation+0.000
$
Token $
Feature activation+0.000
1
Token1
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Sunday
TokenSunday
Feature activation+0.000
,
Token,
Feature activation+0.000
11
Token 11
Feature activation+0.000
am
Tokenam
Feature activation+0.000
-
Token-
Feature activation+0.000
3
Token3
Feature activation+0.000
better
Token better
Feature activation+0.000
understanding
Token understanding
Feature activation+0.000
of
Token of
Feature activation+0.000
it
Token it
Feature activation+0.000
from
Token from
Feature activation+0.000
the
Token the
Feature activation+0.000
user
Token user
Feature activation+0.000
-
Token-
Feature activation+0.000
side
Tokenside
Feature activation+0.000
and
Token and
Feature activation+0.000
b
Token b
Feature activation+0.000

BOTTOM ACTIVATIONS
MIN = 0.000

atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
Development
Token Development
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
<|endoftext|>
Token<|endoftext|>
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000

Top feature 5 in H7.1: (feature 14810)

TOP ACTIVATIONS
MAX = 1.862

given
Token given
Feature activation+0.000
a
Token a
Feature activation+0.000
brief
Token brief
Feature activation+0.000
update
Token update
Feature activation+0.000
on
Token on
Feature activation+0.000
his
Token his
Feature activation+1.862
upcoming
Token upcoming
Feature activation+1.370
games
Token games
Feature activation+0.129
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.037
og
Tokenog
Feature activation+0.000
ers
Tokeners
Feature activation+0.000
.
Token.
Feature activation+0.000
T
Token T
Feature activation+0.000
urd
Tokenurd
Feature activation+0.000
hat
Tokenhat
Feature activation+1.855
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.588
kid
Token kid
Feature activation+0.000
had
Token had
Feature activation+0.000
University
Token University
Feature activation+0.000
,
Token,
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
said
Token said
Feature activation+0.000
Am
Token Am
Feature activation+1.699
alg
Tokenalg
Feature activation+1.134
am
Tokenam
Feature activation+1.110
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.684
big
Token big
Feature activation+0.000
blow
Token blow
Feature activation+0.000
,
Token,
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
Klopp
Token Klopp
Feature activation+1.543
said
Token said
Feature activation+0.080
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
action
Token action
Feature activation+0.656
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
plan
Token plan
Feature activation+1.442
they
Token they
Feature activation+0.068
decided
Token decided
Feature activation+1.077
on
Token on
Feature activation+0.338
followed
Token followed
Feature activation+0.000
the
Token the
Feature activation+0.534
fast
Token fast
Feature activation+0.000
as
Token as
Feature activation+0.000
well
Token well
Feature activation+0.000
,
Token,
Feature activation+0.000
so
Token so
Feature activation+0.105
they
Token they
Feature activation+1.432
should
Token should
Feature activation+0.000
get
Token get
Feature activation+0.000
a
Token a
Feature activation+0.000
good
Token good
Feature activation+0.000
turnout
Token turnout
Feature activation+0.000
,
Token,
Feature activation+0.000
is
Token is
Feature activation+0.000
he
Token he
Feature activation+0.345
going
Token going
Feature activation+0.468
to
Token to
Feature activation+0.841
table
Token table
Feature activation+1.400
a
Token a
Feature activation+1.306
new
Token new
Feature activation+1.019
plan
Token plan
Feature activation+1.084
that
Token that
Feature activation+0.944
will
Token will
Feature activation+0.453
a
Token a
Feature activation+0.000
brief
Token brief
Feature activation+0.000
update
Token update
Feature activation+0.000
on
Token on
Feature activation+0.000
his
Token his
Feature activation+1.862
upcoming
Token upcoming
Feature activation+1.370
games
Token games
Feature activation+0.129
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.037
During
TokenDuring
Feature activation+0.062
and
Token and
Feature activation+0.000
was
Token was
Feature activation+0.000
unreliable
Token unreliable
Feature activation+0.000
,
Token,
Feature activation+0.000
so
Token so
Feature activation+0.197
she
Token she
Feature activation+1.349
searched
Token searched
Feature activation+0.008
for
Token for
Feature activation+0.253
another
Token another
Feature activation+0.437
way
Token way
Feature activation+0.265
of
Token of
Feature activation+0.005
is
Token is
Feature activation+0.000
he
Token he
Feature activation+0.345
going
Token going
Feature activation+0.468
to
Token to
Feature activation+0.841
table
Token table
Feature activation+1.400
a
Token a
Feature activation+1.306
new
Token new
Feature activation+1.019
plan
Token plan
Feature activation+1.084
that
Token that
Feature activation+0.944
will
Token will
Feature activation+0.453
actually
Token actually
Feature activation+0.000
the
Token the
Feature activation+0.102
temptation
Token temptation
Feature activation+0.000
offered
Token offered
Feature activation+0.000
by
Token by
Feature activation+0.000
this
Token this
Feature activation+0.466
faction
Token faction
Feature activation+1.265
with
Token with
Feature activation+0.000
great
Token great
Feature activation+0.000
effort
Token effort
Feature activation+0.000
,
Token,
Feature activation+0.000
though
Token though
Feature activation+0.000
ed
Tokened
Feature activation+0.000
video
Token video
Feature activation+0.000
,
Token,
Feature activation+0.051
the
Token the
Feature activation+0.019
Vermont
Token Vermont
Feature activation+0.000
senator
Token senator
Feature activation+1.260
neither
Token neither
Feature activation+0.000
conceded
Token conceded
Feature activation+0.000
defeat
Token defeat
Feature activation+0.000
nor
Token nor
Feature activation+0.000
endorsed
Token endorsed
Feature activation+0.000
said
Token said
Feature activation+0.000
,
Token,
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
so
Tokenso
Feature activation+0.000
I
Token I
Feature activation+1.247
think
Token think
Feature activation+0.000
Jesus
Token Jesus
Feature activation+0.000
would
Token would
Feature activation+0.000
not
Token not
Feature activation+0.000
run
Token run
Feature activation+0.035
governments
Token governments
Feature activation+0.000
have
Token have
Feature activation+0.037
been
Token been
Feature activation+0.000
reluctant
Token reluctant
Feature activation+0.059
to
Token to
Feature activation+0.238
take
Token take
Feature activation+1.229
measures
Token measures
Feature activation+0.240
to
Token to
Feature activation+0.000
curb
Token curb
Feature activation+0.000
rising
Token rising
Feature activation+0.000
car
Token car
Feature activation+0.000
move
Token move
Feature activation+0.401
might
Token might
Feature activation+0.354
be
Token be
Feature activation+0.147
.
Token.
Feature activation+0.000
Will
Token Will
Feature activation+0.663
he
Token he
Feature activation+1.219
focus
Token focus
Feature activation+0.000
on
Token on
Feature activation+0.000
Ukraine
Token Ukraine
Feature activation+0.000
again
Token again
Feature activation+0.000
,
Token,
Feature activation+0.000
Thursday
Token Thursday
Feature activation+0.000
she
Token she
Feature activation+0.000
had
Token had
Feature activation+0.000
no
Token no
Feature activation+0.000
knowledge
Token knowledge
Feature activation+0.000
of
Token of
Feature activation+1.195
her
Token her
Feature activation+0.000
son
Token son
Feature activation+0.000
's
Token's
Feature activation+0.169
involvement
Token involvement
Feature activation+0.113
in
Token in
Feature activation+0.042
to
Token to
Feature activation+0.280
go
Token go
Feature activation+0.216
,
Token,
Feature activation+0.000
and
Token and
Feature activation+0.000
Sam
Token Sam
Feature activation+0.000
goes
Token goes
Feature activation+1.193
outrage
Token outrage
Feature activation+0.000
ously
Tokenously
Feature activation+1.074
far
Token far
Feature activation+0.975
,
Token,
Feature activation+0.058
and
Token and
Feature activation+0.053
announced
Token announced
Feature activation+0.534
at
Token at
Feature activation+0.000
the
Token the
Feature activation+0.000
show
Token show
Feature activation+0.000
that
Token that
Feature activation+0.596
it
Token it
Feature activation+1.182
would
Token would
Feature activation+0.000
would
Token would
Feature activation+0.000
with
Token with
Feature activation+0.000
partners
Token partners
Feature activation+0.000
through
Token through
Feature activation+0.000
,
Token,
Feature activation+0.000
as
Token as
Feature activation+0.000
leading
Token leading
Feature activation+0.000
Muslim
Token Muslim
Feature activation+0.000
civic
Token civic
Feature activation+0.719
groups
Token groups
Feature activation+1.178
involved
Token involved
Feature activation+0.000
in
Token in
Feature activation+0.059
a
Token a
Feature activation+0.000
major
Token major
Feature activation+0.000
festival
Token festival
Feature activation+0.000
News
Token News
Feature activation+0.000
Why
Token Why
Feature activation+0.000
is
Token is
Feature activation+0.000
the
Token the
Feature activation+0.000
government
Token government
Feature activation+0.289
launching
Token launching
Feature activation+1.173
a
Token a
Feature activation+1.053
campaign
Token campaign
Feature activation+0.195
to
Token to
Feature activation+0.130
end
Token end
Feature activation+0.000
violence
Token violence
Feature activation+0.000

Top DFA by src position
MAX = 2.180

busy
Token busy
Feature activation-0.001
Top resid features:
guy
Token guy
Feature activation+0.002
Top resid features:
.
Token.
Feature activation-0.002
Top resid features:
His
Token His
Feature activation+0.002
Top resid features:
current
Token current
Feature activation+0.058
Top resid features:
projects
Token projects
Feature activation+1.887
Top resid features:
include
Token include
Feature activation+0.106
Top resid features:
Kingdom
Token Kingdom
Feature activation+0.003
Top resid features:
Hearts
Token Hearts
Feature activation+0.019
Top resid features:
III
Token III
Feature activation+0.003
Top resid features:
,
Token,
Feature activation+0.002
Top resid features:
urd
Tokenurd
Feature activation+0.001
Top resid features:
hat
Tokenhat
Feature activation-0.000
Top resid features:
could
Token could
Feature activation+0.013
Top resid features:
come
Token come
Feature activation+0.009
Top resid features:
up
Token up
Feature activation+0.009
Top resid features:
with
Token with
Feature activation+2.180
Top resid features:
.
Token.
Feature activation+0.013
Top resid features:
Including
Token Including
Feature activation+0.002
Top resid features:
âĢ¦
TokenâĢ¦
Feature activation-0.003
Top resid features:
(
Token(
Feature activation-0.001
Top resid features:
d
Tokend
Feature activation-0.001
Top resid features:
Philadelphia
Token Philadelphia
Feature activation+0.000
Top resid features:
.
Token.
Feature activation+0.001
Top resid features:
Am
Token Am
Feature activation+0.001
Top resid features:
alg
Tokenalg
Feature activation-0.085
Top resid features:
am
Tokenam
Feature activation-0.007
Top resid features:
hopes
Token hopes
Feature activation+1.915
Top resid features:
to
Token to
Feature activation+0.047
Top resid features:
build
Token build
Feature activation+0.018
Top resid features:
community
Token community
Feature activation-0.000
Top resid features:
around
Token around
Feature activation-0.000
Top resid features:
comics
Token comics
Feature activation-0.000
Top resid features:
trip
Token trip
Feature activation+0.000
Top resid features:
to
Token to
Feature activation-0.000
Top resid features:
Ireland
Token Ireland
Feature activation-0.000
Top resid features:
but
Token but
Feature activation-0.017
Top resid features:
Klopp
Token Klopp
Feature activation+0.011
Top resid features:
hopes
Token hopes
Feature activation+1.533
Top resid features:
to
Token to
Feature activation+0.234
Top resid features:
have
Token have
Feature activation+0.095
Top resid features:
Philippe
Token Philippe
Feature activation+0.001
Top resid features:
Cout
Token Cout
Feature activation-0.001
Top resid features:
inho
Tokeninho
Feature activation-0.000
Top resid features:
to
Token to
Feature activation-0.001
Top resid features:
discuss
Token discuss
Feature activation+0.009
Top resid features:
the
Token the
Feature activation+0.007
Top resid features:
plan
Token plan
Feature activation+0.025
Top resid features:
of
Token of
Feature activation+0.055
Top resid features:
action
Token action
Feature activation+1.739
Top resid features:
.
Token.
Feature activation-0.007
Top resid features:
Ċ
TokenĊ
Feature activation-0.006
Top resid features:
Ċ
TokenĊ
Feature activation-0.004
Top resid features:
The
TokenThe
Feature activation-0.007
Top resid features:
plan
Token plan
Feature activation-0.031
Top resid features:
weekend
Token weekend
Feature activation-0.005
Top resid features:
.
Token.
Feature activation+0.003
Top resid features:
The
Token The
Feature activation-0.000
Top resid features:
two
Token two
Feature activation+0.006
Top resid features:
initially
Token initially
Feature activation+0.027
Top resid features:
planned
Token planned
Feature activation+1.675
Top resid features:
performances
Token performances
Feature activation+0.013
Top resid features:
sold
Token sold
Feature activation-0.031
Top resid features:
out
Token out
Feature activation-0.023
Top resid features:
so
Token so
Feature activation-0.015
Top resid features:
quickly
Token quickly
Feature activation-0.005
Top resid features:
going
Token going
Feature activation+0.009
Top resid features:
to
Token to
Feature activation+0.010
Top resid features:
table
Token table
Feature activation+0.016
Top resid features:
a
Token a
Feature activation+0.252
Top resid features:
new
Token new
Feature activation+0.256
Top resid features:
plan
Token plan
Feature activation+0.695
Top resid features:
that
Token that
Feature activation+0.428
Top resid features:
will
Token will
Feature activation+0.122
Top resid features:
actually
Token actually
Feature activation+0.026
Top resid features:
create
Token create
Feature activation+0.004
Top resid features:
jobs
Token jobs
Feature activation-0.002
Top resid features:
busy
Token busy
Feature activation-0.001
Top resid features:
guy
Token guy
Feature activation+0.001
Top resid features:
.
Token.
Feature activation-0.000
Top resid features:
His
Token His
Feature activation+0.001
Top resid features:
current
Token current
Feature activation+0.006
Top resid features:
projects
Token projects
Feature activation+1.423
Top resid features:
include
Token include
Feature activation+0.104
Top resid features:
Kingdom
Token Kingdom
Feature activation+0.006
Top resid features:
Hearts
Token Hearts
Feature activation+0.005
Top resid features:
III
Token III
Feature activation+0.003
Top resid features:
,
Token,
Feature activation+0.004
Top resid features:
dimensions
Token dimensions
Feature activation+0.001
Top resid features:
.
Token.
Feature activation+0.001
Top resid features:
Ċ
TokenĊ
Feature activation+0.001
Top resid features:
Ċ
TokenĊ
Feature activation+0.001
Top resid features:
Her
TokenHer
Feature activation+0.001
Top resid features:
goal
Token goal
Feature activation+0.699
Top resid features:
today
Token today
Feature activation+0.088
Top resid features:
was
Token was
Feature activation+0.172
Top resid features:
to
Token to
Feature activation+0.208
Top resid features:
create
Token create
Feature activation+0.341
Top resid features:
a
Token a
Feature activation+0.093
Top resid features:
going
Token going
Feature activation+0.004
Top resid features:
to
Token to
Feature activation+0.008
Top resid features:
table
Token table
Feature activation+0.006
Top resid features:
a
Token a
Feature activation+0.004
Top resid features:
new
Token new
Feature activation+0.097
Top resid features:
plan
Token plan
Feature activation+0.668
Top resid features:
that
Token that
Feature activation+0.640
Top resid features:
will
Token will
Feature activation+0.238
Top resid features:
actually
Token actually
Feature activation+0.034
Top resid features:
create
Token create
Feature activation+0.027
Top resid features:
jobs
Token jobs
Feature activation-0.002
Top resid features:
,
Token,
Feature activation-0.005
Top resid features:
a
Token a
Feature activation-0.002
Top resid features:
third
Token third
Feature activation+0.000
Top resid features:
faction
Token faction
Feature activation-0.009
Top resid features:
wishes
Token wishes
Feature activation+0.560
Top resid features:
to
Token to
Feature activation+0.641
Top resid features:
stage
Token stage
Feature activation+0.250
Top resid features:
a
Token a
Feature activation+0.081
Top resid features:
military
Token military
Feature activation+0.015
Top resid features:
coup
Token coup
Feature activation+0.032
Top resid features:
upon
Token upon
Feature activation+0.005
Top resid features:
68
Token68
Feature activation-0.000
Top resid features:
404
Token404
Feature activation-0.000
Top resid features:
"
Token"
Feature activation-0.001
Top resid features:
true
Token true
Feature activation-0.000
Top resid features:
Sanders
Token Sanders
Feature activation+0.009
Top resid features:
vows
Token vows
Feature activation+1.184
Top resid features:
to
Token to
Feature activation+0.300
Top resid features:
work
Token work
Feature activation+0.085
Top resid features:
with
Token with
Feature activation+0.022
Top resid features:
Clinton
Token Clinton
Feature activation-0.003
Top resid features:
to
Token to
Feature activation+0.014
Top resid features:
s
Tokens
Feature activation-0.001
Top resid features:
Terry
Token Terry
Feature activation-0.000
Top resid features:
Moran
Token Moran
Feature activation-0.003
Top resid features:
that
Token that
Feature activation+0.014
Top resid features:
his
Token his
Feature activation+0.012
Top resid features:
plan
Token plan
Feature activation+0.829
Top resid features:
to
Token to
Feature activation+0.424
Top resid features:
burn
Token burn
Feature activation+0.098
Top resid features:
Qur
Token Qur
Feature activation-0.002
Top resid features:
ans
Tokenans
Feature activation-0.020
Top resid features:
was
Token was
Feature activation+0.179
Top resid features:
taking
Token taking
Feature activation+0.001
Top resid features:
far
Token far
Feature activation+0.033
Top resid features:
more
Token more
Feature activation+0.001
Top resid features:
politically
Token politically
Feature activation+0.009
Top resid features:
challenging
Token challenging
Feature activation+0.074
Top resid features:
steps
Token steps
Feature activation+1.348
Top resid features:
to
Token to
Feature activation+0.041
Top resid features:
tackle
Token tackle
Feature activation+0.019
Top resid features:
emissions
Token emissions
Feature activation-0.006
Top resid features:
from
Token from
Feature activation+0.000
Top resid features:
transport
Token transport
Feature activation+0.001
Top resid features:
to
Token to
Feature activation-0.001
Top resid features:
wonder
Token wonder
Feature activation+0.008
Top resid features:
what
Token what
Feature activation+0.005
Top resid features:
his
Token his
Feature activation+0.011
Top resid features:
next
Token next
Feature activation+0.452
Top resid features:
move
Token move
Feature activation+1.015
Top resid features:
might
Token might
Feature activation+0.005
Top resid features:
be
Token be
Feature activation-0.018
Top resid features:
.
Token.
Feature activation+0.015
Top resid features:
Will
Token Will
Feature activation-0.004
Top resid features:
he
Token he
Feature activation+0.017
Top resid features:
didn
Token didn
Feature activation-0.001
Top resid features:
't
Token't
Feature activation-0.001
Top resid features:
know
Token know
Feature activation-0.001
Top resid features:
about
Token about
Feature activation+0.004
Top resid features:
the
Token the
Feature activation+0.006
Top resid features:
plans
Token plans
Feature activation+1.478
Top resid features:
for
Token for
Feature activation+0.118
Top resid features:
the
Token the
Feature activation+0.024
Top resid features:
San
Token San
Feature activation+0.000
Top resid features:
Bernardino
Token Bernardino
Feature activation+0.001
Top resid features:
attack
Token attack
Feature activation+0.008
Top resid features:
see
Token see
Feature activation-0.000
Top resid features:
Sam
Token Sam
Feature activation+0.000
Top resid features:
going
Token going
Feature activation+0.001
Top resid features:
to
Token to
Feature activation+0.158
Top resid features:
crazy
Token crazy
Feature activation+0.038
Top resid features:
lengths
Token lengths
Feature activation+0.969
Top resid features:
to
Token to
Feature activation+0.054
Top resid features:
find
Token find
Feature activation+0.051
Top resid features:
and
Token and
Feature activation-0.001
Top resid features:
save
Token save
Feature activation+0.007
Top resid features:
demon
Token demon
Feature activation+0.001
Top resid features:
rival
Token rival
Feature activation+0.005
Top resid features:
Google
Token Google
Feature activation+0.002
Top resid features:
has
Token has
Feature activation+0.002
Top resid features:
announced
Token announced
Feature activation+0.019
Top resid features:
its
Token its
Feature activation+0.007
Top resid features:
ambitions
Token ambitions
Feature activation+0.876
Top resid features:
to
Token to
Feature activation+0.475
Top resid features:
bring
Token bring
Feature activation+0.103
Top resid features:
Android
Token Android
Feature activation+0.004
Top resid features:
to
Token to
Feature activation-0.010
Top resid features:
the
Token the
Feature activation-0.004
Top resid features:
<|endoftext|>
Token<|endoftext|>
Feature activation+0.123
Top resid features:
civic
Token civic
Feature activation+0.000
Top resid features:
leaders
Token leaders
Feature activation+0.006
Top resid features:
are
Token are
Feature activation-0.001
Top resid features:
hoping
Token hoping
Feature activation+0.647
Top resid features:
to
Token to
Feature activation+0.635
Top resid features:
draw
Token draw
Feature activation+0.131
Top resid features:
on
Token on
Feature activation+0.091
Top resid features:
the
Token the
Feature activation+0.007
Top resid features:
little
Token little
Feature activation+0.003
Top resid features:
the
Token the
Feature activation-0.000
Top resid features:
focus
Token focus
Feature activation+0.002
Top resid features:
of
Token of
Feature activation-0.000
Top resid features:
this
Token this
Feature activation+0.000
Top resid features:
latest
Token latest
Feature activation+0.003
Top resid features:
strategy
Token strategy
Feature activation+0.658
Top resid features:
because
Token because
Feature activation-0.002
Top resid features:
80
Token 80
Feature activation-0.000
Top resid features:
%
Token%
Feature activation-0.005
Top resid features:
of
Token of
Feature activation-0.001
Top resid features:
domestic
Token domestic
Feature activation-0.000
Top resid features:

Decoder Weights Distribution

Head 0: 0.03

Head 1: 0.48

Head 2: 0.03

Head 3: 0.05

Head 4: 0.02

Head 5: 0.10

Head 6: 0.09

Head 7: 0.03

Head 8: 0.04

Head 9: 0.05

Head 10: 0.05

Head 11: 0.02

Positive logits

plans4.66

proposes4.11

plan4.04

Plans4.01

planned3.96

proposals3.90

propose3.81

proposal3.78

intends3.77

planning3.72

ambitious3.68

envision3.59

ambitions3.58

proposed3.49

proposing3.49

brainstorm3.47

intend3.42

plan3.40

aims3.40

ambition3.39

Negative logits

ejected-2.40

+/--2.39

erved-2.33

served-2.18

ewitness-2.16

Citation-2.15

notations-2.12

rity-2.12

disappeared-2.05

expired-2.03

vation-2.02

fou-2.01

gregation-2.01

Runtime-1.95

emort-1.94

tainted-1.93

AVG-1.91

Baird-1.90

ochet-1.90

inguished-1.89

INTERVAL 1.676 - 1.862
CONTAINS 0.000%

University
Token University
Feature activation+0.000
,
Token,
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
said
Token said
Feature activation+0.000
Am
Token Am
Feature activation+1.699
alg
Tokenalg
Feature activation+1.134
am
Tokenam
Feature activation+1.110
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.684
given
Token given
Feature activation+0.000
a
Token a
Feature activation+0.000
brief
Token brief
Feature activation+0.000
update
Token update
Feature activation+0.000
on
Token on
Feature activation+0.000
his
Token his
Feature activation+1.862
upcoming
Token upcoming
Feature activation+1.370
games
Token games
Feature activation+0.129
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.037
og
Tokenog
Feature activation+0.000
ers
Tokeners
Feature activation+0.000
.
Token.
Feature activation+0.000
T
Token T
Feature activation+0.000
urd
Tokenurd
Feature activation+0.000
hat
Tokenhat
Feature activation+1.855
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.588
kid
Token kid
Feature activation+0.000
had
Token had
Feature activation+0.000

INTERVAL 1.490 - 1.676
CONTAINS 0.000%

big
Token big
Feature activation+0.000
blow
Token blow
Feature activation+0.000
,
Token,
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
Klopp
Token Klopp
Feature activation+1.543
said
Token said
Feature activation+0.080
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000

INTERVAL 1.303 - 1.490
CONTAINS 0.001%

and
Token and
Feature activation+0.000
was
Token was
Feature activation+0.000
unreliable
Token unreliable
Feature activation+0.000
,
Token,
Feature activation+0.000
so
Token so
Feature activation+0.197
she
Token she
Feature activation+1.349
searched
Token searched
Feature activation+0.008
for
Token for
Feature activation+0.253
another
Token another
Feature activation+0.437
way
Token way
Feature activation+0.265
of
Token of
Feature activation+0.005
fast
Token fast
Feature activation+0.000
as
Token as
Feature activation+0.000
well
Token well
Feature activation+0.000
,
Token,
Feature activation+0.000
so
Token so
Feature activation+0.105
they
Token they
Feature activation+1.432
should
Token should
Feature activation+0.000
get
Token get
Feature activation+0.000
a
Token a
Feature activation+0.000
good
Token good
Feature activation+0.000
turnout
Token turnout
Feature activation+0.000
action
Token action
Feature activation+0.656
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
plan
Token plan
Feature activation+1.442
they
Token they
Feature activation+0.068
decided
Token decided
Feature activation+1.077
on
Token on
Feature activation+0.338
followed
Token followed
Feature activation+0.000
the
Token the
Feature activation+0.534
is
Token is
Feature activation+0.000
he
Token he
Feature activation+0.345
going
Token going
Feature activation+0.468
to
Token to
Feature activation+0.841
table
Token table
Feature activation+1.400
a
Token a
Feature activation+1.306
new
Token new
Feature activation+1.019
plan
Token plan
Feature activation+1.084
that
Token that
Feature activation+0.944
will
Token will
Feature activation+0.453
actually
Token actually
Feature activation+0.000
,
Token,
Feature activation+0.000
is
Token is
Feature activation+0.000
he
Token he
Feature activation+0.345
going
Token going
Feature activation+0.468
to
Token to
Feature activation+0.841
table
Token table
Feature activation+1.400
a
Token a
Feature activation+1.306
new
Token new
Feature activation+1.019
plan
Token plan
Feature activation+1.084
that
Token that
Feature activation+0.944
will
Token will
Feature activation+0.453

INTERVAL 1.117 - 1.303
CONTAINS 0.002%

Thursday
Token Thursday
Feature activation+0.000
she
Token she
Feature activation+0.000
had
Token had
Feature activation+0.000
no
Token no
Feature activation+0.000
knowledge
Token knowledge
Feature activation+0.000
of
Token of
Feature activation+1.195
her
Token her
Feature activation+0.000
son
Token son
Feature activation+0.000
's
Token's
Feature activation+0.169
involvement
Token involvement
Feature activation+0.113
in
Token in
Feature activation+0.042
said
Token said
Feature activation+0.000
,
Token,
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
so
Tokenso
Feature activation+0.000
I
Token I
Feature activation+1.247
think
Token think
Feature activation+0.000
Jesus
Token Jesus
Feature activation+0.000
would
Token would
Feature activation+0.000
not
Token not
Feature activation+0.000
run
Token run
Feature activation+0.035
Android
Token Android
Feature activation+0.000
to
Token to
Feature activation+0.000
the
Token the
Feature activation+0.000
vehicle
Token vehicle
Feature activation+0.000
.
Token.
Feature activation+0.000
Google
Token Google
Feature activation+1.160
announced
Token announced
Feature activation+0.534
at
Token at
Feature activation+0.000
the
Token the
Feature activation+0.000
show
Token show
Feature activation+0.000
that
Token that
Feature activation+0.596
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
China
TokenChina
Feature activation+0.000
did
Token did
Feature activation+0.386
not
Token not
Feature activation+0.018
set
Token set
Feature activation+1.132
a
Token a
Feature activation+0.507
specific
Token specific
Feature activation+0.214
target
Token target
Feature activation+0.089
,
Token,
Feature activation+0.000
but
Token but
Feature activation+0.058
News
Token News
Feature activation+0.000
Why
Token Why
Feature activation+0.000
is
Token is
Feature activation+0.000
the
Token the
Feature activation+0.000
government
Token government
Feature activation+0.289
launching
Token launching
Feature activation+1.173
a
Token a
Feature activation+1.053
campaign
Token campaign
Feature activation+0.195
to
Token to
Feature activation+0.130
end
Token end
Feature activation+0.000
violence
Token violence
Feature activation+0.000

INTERVAL 0.931 - 1.117
CONTAINS 0.003%

and
Token and
Feature activation+0.000
Sam
Token Sam
Feature activation+0.000
goes
Token goes
Feature activation+1.193
outrage
Token outrage
Feature activation+0.000
ously
Tokenously
Feature activation+1.074
far
Token far
Feature activation+0.975
,
Token,
Feature activation+0.058
and
Token and
Feature activation+0.053
we
Token we
Feature activation+0.000
see
Token see
Feature activation+0.000
the
Token the
Feature activation+0.000
turn
Token turn
Feature activation+0.000
left
Token left
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.046
He
TokenHe
Feature activation+0.998
had
Token had
Feature activation+0.708
what
Token what
Feature activation+0.000
he
Token he
Feature activation+0.176
called
Token called
Feature activation+0.000
a
Token a
Feature activation+0.342
Why
Token Why
Feature activation+0.000
is
Token is
Feature activation+0.000
the
Token the
Feature activation+0.000
government
Token government
Feature activation+0.289
launching
Token launching
Feature activation+1.173
a
Token a
Feature activation+1.053
campaign
Token campaign
Feature activation+0.195
to
Token to
Feature activation+0.130
end
Token end
Feature activation+0.000
violence
Token violence
Feature activation+0.000
against
Token against
Feature activation+0.000
looked
Token looked
Feature activation+0.000
down
Token down
Feature activation+0.000
upon
Token upon
Feature activation+0.000
the
Token the
Feature activation+0.092
Energy
Token Energy
Feature activation+0.825
Master
Token Master
Feature activation+1.004
Plan
Token Plan
Feature activation+0.000
decreasing
Token decreasing
Feature activation+0.000
from
Token from
Feature activation+0.000
30
Token 30
Feature activation+0.000
percent
Token percent
Feature activation+0.000
because
Token because
Feature activation+0.000
of
Token of
Feature activation+0.000
marijuana
Token marijuana
Feature activation+0.000
.
Token.
Feature activation+0.000
The
Token The
Feature activation+0.000
organizations
Token organizations
Feature activation+1.094
said
Token said
Feature activation+0.539
previous
Token previous
Feature activation+0.000
offenders
Token offenders
Feature activation+0.000
often
Token often
Feature activation+0.000
have
Token have
Feature activation+0.000

INTERVAL 0.745 - 0.931
CONTAINS 0.006%

people
Token people
Feature activation+0.000
have
Token have
Feature activation+0.000
never
Token never
Feature activation+0.000
seen
Token seen
Feature activation+0.000
.
Token.
Feature activation+0.000
I
Token I
Feature activation+0.767
don
Token don
Feature activation+0.144
't
Token't
Feature activation+0.266
want
Token want
Feature activation+0.687
it
Token it
Feature activation+0.000
to
Token to
Feature activation+0.000
Miami
Token Miami
Feature activation+0.000
.
Token.
Feature activation+0.069
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.075
My
TokenMy
Feature activation+0.877
plan
Token plan
Feature activation+0.800
,
Token,
Feature activation+0.000
like
Token like
Feature activation+0.365
his
Token his
Feature activation+0.337
,
Token,
Feature activation+0.000
really
Token really
Feature activation+0.164
2
Token2
Feature activation+0.000
B
TokenB
Feature activation+0.000
,
Token,
Feature activation+0.000
so
Token so
Feature activation+0.000
we
Token we
Feature activation+0.271
decided
Token decided
Feature activation+0.786
that
Token that
Feature activation+0.600
we
Token we
Feature activation+0.477
needed
Token needed
Feature activation+0.377
to
Token to
Feature activation+0.211
do
Token do
Feature activation+0.150
@
Token @
Feature activation+0.000
realDonaldTrump
TokenrealDonaldTrump
Feature activation+0.174
&
Token &
Feature activation+0.000
Senate
Token Senate
Feature activation+0.000
leadership
Token leadership
Feature activation+0.000
about
Token about
Feature activation+0.860
this
Token this
Feature activation+0.896
and
Token and
Feature activation+0.007
agree
Token agree
Feature activation+0.315
.
Token.
Feature activation+0.338
Let
Token Let
Feature activation+0.484
around
Token around
Feature activation+0.000
burning
Token burning
Feature activation+0.000
books
Token books
Feature activation+0.000
,
Token,
Feature activation+0.000
but
Token but
Feature activation+0.000
I
Token I
Feature activation+0.862
think
Token think
Feature activation+0.000
he
Token he
Feature activation+0.000
would
Token would
Feature activation+0.000
burn
Token burn
Feature activation+0.000
this
Token this
Feature activation+0.000

INTERVAL 0.559 - 0.745
CONTAINS 0.019%

in
Token in
Feature activation+0.000
a
Token a
Feature activation+0.000
terrorist
Token terrorist
Feature activation+0.000
attack
Token attack
Feature activation+0.000
.
Token.
Feature activation+0.000
He
Token He
Feature activation+0.580
calls
Token calls
Feature activation+0.000
it
Token it
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
extreme
Tokenextreme
Feature activation+0.000
a
Token a
Feature activation+0.064
digital
Token digital
Feature activation+0.000
art
Token art
Feature activation+0.000
perspective
Token perspective
Feature activation+0.000
,
Token,
Feature activation+0.000
I
Token I
Feature activation+0.595
'm
Token'm
Feature activation+0.517
trying
Token trying
Feature activation+0.766
to
Token to
Feature activation+0.635
capture
Token capture
Feature activation+0.000
something
Token something
Feature activation+0.000
release
Token release
Feature activation+0.000
announcing
Token announcing
Feature activation+0.070
a
Token a
Feature activation+0.000
bill
Token bill
Feature activation+0.000
the
Token the
Feature activation+0.000
NRA
Token NRA
Feature activation+0.705
says
Token says
Feature activation+0.000
could
Token could
Feature activation+0.000
make
Token make
Feature activation+0.000
Wisconsin
Token Wisconsin
Feature activation+0.000
the
Token the
Feature activation+0.000
,
Token,
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
Do
TokenDo
Feature activation+0.000
you
Token you
Feature activation+0.000
think
Token think
Feature activation+0.608
all
Token all
Feature activation+0.000
federal
Token federal
Feature activation+0.000
funding
Token funding
Feature activation+0.000
for
Token for
Feature activation+0.000
Planned
Token Planned
Feature activation+0.000
left
Token left
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.046
He
TokenHe
Feature activation+0.998
had
Token had
Feature activation+0.708
what
Token what
Feature activation+0.000
he
Token he
Feature activation+0.176
called
Token called
Feature activation+0.000
a
Token a
Feature activation+0.342
"
Token "
Feature activation+0.277

INTERVAL 0.372 - 0.559
CONTAINS 0.044%

and
Token and
Feature activation+0.000
said
Token said
Feature activation+0.000
,
Token,
Feature activation+0.000
'
Token '
Feature activation+0.000
Do
TokenDo
Feature activation+0.262
you
Token you
Feature activation+0.492
wanna
Token wanna
Feature activation+0.238
make
Token make
Feature activation+0.329
a
Token a
Feature activation+0.194
Black
Token Black
Feature activation+0.000
Widow
Token Widow
Feature activation+0.000
DRM
Token DRM
Feature activation+0.000
from
Token from
Feature activation+0.000
beginning
Token beginning
Feature activation+0.000
[...]
Token [...]
Feature activation+0.000
but
Token but
Feature activation+0.000
we
Token we
Feature activation+0.449
couldn
Token couldn
Feature activation+0.000
't
Token't
Feature activation+0.000
find
Token find
Feature activation+0.000
a
Token a
Feature activation+0.000
way
Token way
Feature activation+0.000
-
Token-
Feature activation+0.000
like
Tokenlike
Feature activation+0.000
gar
Token gar
Feature activation+0.000
b
Tokenb
Feature activation+0.000
he
Token he
Feature activation+0.454
has
Token has
Feature activation+0.431
her
Token her
Feature activation+0.000
sw
Token sw
Feature activation+0.000
athed
Tokenathed
Feature activation+0.000
in
Token in
Feature activation+0.000
.
Token.
Feature activation+0.049
Ċ
TokenĊ
Feature activation+0.000
"
Token"
Feature activation+0.016
I
TokenI
Feature activation+0.436
don
Token don
Feature activation+0.103
't
Token't
Feature activation+0.229
have
Token have
Feature activation+0.388
any
Token any
Feature activation+0.304
precon
Token precon
Feature activation+0.000
ceived
Tokenceived
Feature activation+0.000
notion
Token notion
Feature activation+0.073
on
Token on
Feature activation+0.228
to
Token to
Feature activation+0.000
come
Token come
Feature activation+0.000
up
Token up
Feature activation+0.000
with
Token with
Feature activation+0.000
ideas
Token ideas
Feature activation+0.364
for
Token for
Feature activation+0.412
his
Token his
Feature activation+0.395
group
Token group
Feature activation+0.239
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.050

INTERVAL 0.186 - 0.372
CONTAINS 0.123%

.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
"
Token"
Feature activation+0.000
We
TokenWe
Feature activation+0.000
're
Token're
Feature activation+0.189
currently
Token currently
Feature activation+0.000
working
Token working
Feature activation+0.000
on
Token on
Feature activation+0.000
an
Token an
Feature activation+0.000
update
Token update
Feature activation+0.000
person
Token person
Feature activation+0.000
is
Token is
Feature activation+0.000
asking
Token asking
Feature activation+0.000
its
Token its
Feature activation+0.000
operatives
Token operatives
Feature activation+0.000
to
Token to
Feature activation+0.199
carry
Token carry
Feature activation+0.000
out
Token out
Feature activation+0.000
lone
Token lone
Feature activation+0.000
wolf
Token wolf
Feature activation+0.000
attacks
Token attacks
Feature activation+0.000
more
Token more
Feature activation+0.000
inventive
Token inventive
Feature activation+0.000
and
Token and
Feature activation+0.000
aggressive
Token aggressive
Feature activation+0.000
in
Token in
Feature activation+0.000
its
Token its
Feature activation+0.270
strategy
Token strategy
Feature activation+0.000
."
Token."
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
But
TokenBut
Feature activation+0.000
to
Token to
Feature activation+0.000
the
Token the
Feature activation+0.000
island
Token island
Feature activation+0.000
to
Token to
Feature activation+0.000
film
Token film
Feature activation+0.180
the
Token the
Feature activation+0.225
ads
Token ads
Feature activation+0.189
,
Token,
Feature activation+0.000
in
Token in
Feature activation+0.000
which
Token which
Feature activation+0.000
she
Token she
Feature activation+0.000
but
Token but
Feature activation+0.000
the
Token the
Feature activation+0.000
mother
Token mother
Feature activation+0.000
contends
Token contends
Feature activation+0.000
that
Token that
Feature activation+0.000
he
Token he
Feature activation+0.215
is
Token is
Feature activation+0.000
merely
Token merely
Feature activation+0.000
bending
Token bending
Feature activation+0.000
to
Token to
Feature activation+0.000
his
Token his
Feature activation+0.000

INTERVAL 0.000 - 0.186
CONTAINS 99.803%

living
Token living
Feature activation+0.000
here
Token here
Feature activation+0.000
the
Token the
Feature activation+0.000
past
Token past
Feature activation+0.000
couple
Token couple
Feature activation+0.000
of
Token of
Feature activation+0.000
months
Token months
Feature activation+0.000
,
Token,
Feature activation+0.000
finishing
Token finishing
Feature activation+0.000
the
Token the
Feature activation+0.000
record
Token record
Feature activation+0.000
said
Token said
Feature activation+0.000
Fox
Token Fox
Feature activation+0.000
cited
Token cited
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
con
Tokencon
Feature activation+0.000
cern
Tokencern
Feature activation+0.000
s
Tokens
Feature activation+0.000
about
Token about
Feature activation+0.000
some
Token some
Feature activation+0.000
elements
Token elements
Feature activation+0.000
family
Token family
Feature activation+0.000
,
Token,
Feature activation+0.000
who
Token who
Feature activation+0.000
have
Token have
Feature activation+0.000
been
Token been
Feature activation+0.000
extremely
Token extremely
Feature activation+0.000
supportive
Token supportive
Feature activation+0.000
and
Token and
Feature activation+0.000
have
Token have
Feature activation+0.000
made
Token made
Feature activation+0.000
significant
Token significant
Feature activation+0.000
the
Token the
Feature activation+0.000
future
Token future
Feature activation+0.000
of
Token of
Feature activation+0.000
3
Token 3
Feature activation+0.000
D
TokenD
Feature activation+0.000
printing
Token printing
Feature activation+0.000
and
Token and
Feature activation+0.000
3
Token 3
Feature activation+0.000
D
TokenD
Feature activation+0.000
modeling
Token modeling
Feature activation+0.000
with
Token with
Feature activation+0.000
in
Token in
Feature activation+0.000
the
Token the
Feature activation+0.000
pilot
Token pilot
Feature activation+0.000
,
Token,
Feature activation+0.000
Clarke
Token Clarke
Feature activation+0.000
mentions
Token mentions
Feature activation+0.000
that
Token that
Feature activation+0.000
her
Token her
Feature activation+0.000
18
Token 18
Feature activation+0.000
th
Tokenth
Feature activation+0.000
birthday
Token birthday
Feature activation+0.000

BOTTOM ACTIVATIONS
MIN = 0.000

atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
Development
Token Development
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
<|endoftext|>
Token<|endoftext|>
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000

Top feature 6 in H7.1: (feature 6438)

TOP ACTIVATIONS
MAX = 4.042

to
Token to
Feature activation+0.000
the
Token the
Feature activation+0.000
ruling
Token ruling
Feature activation+0.000
.
Token.
Feature activation+0.000
Haz
Token Haz
Feature activation+3.207
le
Tokenle
Feature activation+4.042
then
Token then
Feature activation+0.828
found
Token found
Feature activation+0.209
himself
Token himself
Feature activation+0.000
back
Token back
Feature activation+0.000
in
Token in
Feature activation+0.000
fell
Token fell
Feature activation+0.000
off
Token off
Feature activation+0.000
and
Token and
Feature activation+0.000
the
Token the
Feature activation+0.000
pit
Token pit
Feature activation+1.707
bull
Token bull
Feature activation+4.010
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.416
jaws
Token jaws
Feature activation+0.000
locked
Token locked
Feature activation+0.000
Moment
Token Moment
Feature activation+0.000
grew
Token grew
Feature activation+0.000
out
Token out
Feature activation+0.000
of
Token of
Feature activation+0.000
B
Token B
Feature activation+2.522
ala
Tokenala
Feature activation+3.903
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.855
s
Tokens
Feature activation+0.286
interest
Token interest
Feature activation+0.000
in
Token in
Feature activation+0.074
it
Token it
Feature activation+0.000
."
Token."
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.051
X
TokenX
Feature activation+0.934
inos
Tokeninos
Feature activation+3.683
,
Token,
Feature activation+1.603
who
Token who
Feature activation+0.524
unsuccessfully
Token unsuccessfully
Feature activation+0.000
sued
Token sued
Feature activation+0.000
to
Token to
Feature activation+0.000
During
TokenDuring
Feature activation+0.000
the
Token the
Feature activation+0.000
panel
Token panel
Feature activation+0.000
session
Token session
Feature activation+0.000
,
Token,
Feature activation+0.359
Armstrong
Token Armstrong
Feature activation+3.483
told
Token told
Feature activation+0.555
delegates
Token delegates
Feature activation+0.000
the
Token the
Feature activation+0.000
move
Token move
Feature activation+0.000
was
Token was
Feature activation+0.000
according
Token according
Feature activation+0.000
to
Token to
Feature activation+0.000
the
Token the
Feature activation+0.000
ruling
Token ruling
Feature activation+0.000
.
Token.
Feature activation+0.000
Haz
Token Haz
Feature activation+3.207
le
Tokenle
Feature activation+4.042
then
Token then
Feature activation+0.828
found
Token found
Feature activation+0.209
himself
Token himself
Feature activation+0.000
back
Token back
Feature activation+0.000
Wednesday
Token Wednesday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.140
S
TokenS
Feature activation+0.000
olis
Tokenolis
Feature activation+3.095
was
Token was
Feature activation+1.216
notified
Token notified
Feature activation+0.000
of
Token of
Feature activation+0.000
this
Token this
Feature activation+0.000
donation
Token donation
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.300
Apparently
TokenApparently
Feature activation+0.016
this
Token this
Feature activation+0.000
wildlife
Token wildlife
Feature activation+0.000
producer
Token producer
Feature activation+3.057
and
Token and
Feature activation+1.046
director
Token director
Feature activation+0.703
for
Token for
Feature activation+0.000
the
Token the
Feature activation+0.000
popular
Token popular
Feature activation+0.000
vehicle
Token vehicle
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
Trooper
Token Trooper
Feature activation+3.048
diseng
Token diseng
Feature activation+0.000
aged
Tokenaged
Feature activation+0.000
contact
Token contact
Feature activation+0.000
with
Token with
Feature activation+0.000
SC
Token SC
Feature activation+1.346
followed
Token followed
Feature activation+0.000
the
Token the
Feature activation+0.000
Trooper
Token Trooper
Feature activation+1.534
.
Token.
Feature activation+0.000
SC
Token SC
Feature activation+1.580
OTT
TokenOTT
Feature activation+2.983
continued
Token continued
Feature activation+0.155
to
Token to
Feature activation+0.000
advance
Token advance
Feature activation+0.000
towards
Token towards
Feature activation+0.000
the
Token the
Feature activation+0.000
the
Token the
Feature activation+0.000
patrol
Token patrol
Feature activation+0.000
vehicle
Token vehicle
Feature activation+0.000
.
Token.
Feature activation+0.000
SC
Token SC
Feature activation+1.781
OTT
TokenOTT
Feature activation+2.978
dis
Token dis
Feature activation+0.000
mounted
Tokenmounted
Feature activation+0.000
from
Token from
Feature activation+0.000
the
Token the
Feature activation+0.000
suspect
Token suspect
Feature activation+0.000
to
Token to
Feature activation+0.000
fulfill
Token fulfill
Feature activation+0.000
his
Token his
Feature activation+0.000
potential
Token potential
Feature activation+0.000
.
Token.
Feature activation+0.000
Apple
Token Apple
Feature activation+2.960
(
Token (
Feature activation+0.000
whose
Tokenwhose
Feature activation+0.000
given
Token given
Feature activation+0.000
surname
Token surname
Feature activation+0.000
was
Token was
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
pit
Token pit
Feature activation+1.236
bull
Token bull
Feature activation+2.923
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.161
s
Tokens
Feature activation+0.119
owner
Token owner
Feature activation+0.000
had
Token had
Feature activation+0.000
Capitol
Token Capitol
Feature activation+0.000
Bever
Token Bever
Feature activation+0.000
age
Tokenage
Feature activation+0.000
Sales
Token Sales
Feature activation+0.000
.
Token.
Feature activation+0.000
Hans
Token Hans
Feature activation+2.904
and
Token and
Feature activation+0.256
the
Token the
Feature activation+0.000
Morris
Token Morris
Feature activation+0.219
seys
Tokenseys
Feature activation+0.000
welcomed
Token welcomed
Feature activation+0.000
.
Token.
Feature activation+0.171
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.070
P
TokenP
Feature activation+0.221
ant
Tokenant
Feature activation+1.470
ani
Tokenani
Feature activation+2.901
would
Token would
Feature activation+0.628
have
Token have
Feature activation+0.153
been
Token been
Feature activation+0.313
44
Token 44
Feature activation+0.000
on
Token on
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
RIP
TokenRIP
Feature activation+0.000
J
Token J
Feature activation+0.000
.
Token.
Feature activation+0.000
D
Token D
Feature activation+0.000
illa
Tokenilla
Feature activation+2.871
/
Token /
Feature activation+0.000
Jay
Token Jay
Feature activation+0.095
Dee
Token Dee
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
ikk
Tokenikk
Feature activation+0.000
ö
Tokenö
Feature activation+0.000
nen
Tokennen
Feature activation+0.000
for
Token for
Feature activation+0.000
Mald
Token Mald
Feature activation+1.258
on
Tokenon
Feature activation+2.806
ado
Tokenado
Feature activation+2.557
is
Token is
Feature activation+0.000
far
Token far
Feature activation+0.000
from
Token from
Feature activation+0.000
the
Token the
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.333
S
TokenS
Feature activation+0.007
ell
Tokenell
Feature activation+0.000
ers
Tokeners
Feature activation+0.592
,
Token,
Feature activation+1.618
who
Token who
Feature activation+2.791
was
Token was
Feature activation+1.002
impressed
Token impressed
Feature activation+0.000
that
Token that
Feature activation+0.018
Mayor
Token Mayor
Feature activation+0.040
Ethan
Token Ethan
Feature activation+0.000
I
TokenI
Feature activation+0.000
know
Token know
Feature activation+0.000
[
Token [
Feature activation+0.000
W
TokenW
Feature activation+0.947
ol
Tokenol
Feature activation+2.141
ford
Tokenford
Feature activation+2.791
]
Token]
Feature activation+2.355
was
Token was
Feature activation+1.189
really
Token really
Feature activation+0.698
bad
Token bad
Feature activation+0.273
off
Token off
Feature activation+0.104
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Create
TokenCreate
Feature activation+0.000
a
Token a
Feature activation+0.000
portrait
Token portrait
Feature activation+0.000
of
Token of
Feature activation+2.772
Mac
Token Mac
Feature activation+1.432
and
Token and
Feature activation+0.440
he
Token he
Feature activation+0.286
will
Token will
Feature activation+0.185
choose
Token choose
Feature activation+0.000

Top DFA by src position
MAX = 4.325

staff
Token staff
Feature activation+0.002
Top resid features:
reported
Token reported
Feature activation+0.032
Top resid features:
Haz
Token Haz
Feature activation+0.001
Top resid features:
le
Tokenle
Feature activation+0.000
Top resid features:
to
Token to
Feature activation+0.090
Top resid features:
his
Token his
Feature activation+3.687
Top resid features:
parole
Token parole
Feature activation+0.020
Top resid features:
officer
Token officer
Feature activation+0.036
Top resid features:
,
Token,
Feature activation+0.020
Top resid features:
saying
Token saying
Feature activation+0.049
Top resid features:
that
Token that
Feature activation+0.021
Top resid features:
and
Token and
Feature activation-0.000
Top resid features:
was
Token was
Feature activation+0.002
Top resid features:
str
Token str
Feature activation+0.006
Top resid features:
aining
Tokenaining
Feature activation-0.006
Top resid features:
at
Token at
Feature activation+0.012
Top resid features:
his
Token his
Feature activation+4.325
Top resid features:
leash
Token leash
Feature activation+0.013
Top resid features:
,
Token,
Feature activation+0.001
Top resid features:
according
Token according
Feature activation+0.003
Top resid features:
to
Token to
Feature activation+0.001
Top resid features:
Johnson
Token Johnson
Feature activation+0.000
Top resid features:
Ub
Token Ub
Feature activation+0.001
Top resid features:
iqu
Tokeniqu
Feature activation+0.000
Top resid features:
itous
Tokenitous
Feature activation+0.000
Top resid features:
Computing
Token Computing
Feature activation+0.002
Top resid features:
.
Token.
Feature activation+0.001
Top resid features:
His
Token His
Feature activation+4.218
Top resid features:
work
Token work
Feature activation+0.015
Top resid features:
originally
Token originally
Feature activation+0.008
Top resid features:
focused
Token focused
Feature activation+0.004
Top resid features:
on
Token on
Feature activation+0.009
Top resid features:
adapting
Token adapting
Feature activation+0.000
Top resid features:
ated
Tokenated
Feature activation+0.000
Top resid features:
Oak
Token Oak
Feature activation+0.000
Top resid features:
Brook
Token Brook
Feature activation+0.000
Top resid features:
community
Token community
Feature activation+0.004
Top resid features:
where
Token where
Feature activation+0.005
Top resid features:
he
Token he
Feature activation+1.535
Top resid features:
lives
Token lives
Feature activation+0.019
Top resid features:
.
Token.
Feature activation+0.018
Top resid features:
"
Token "
Feature activation-0.000
Top resid features:
This
TokenThis
Feature activation+0.000
Top resid features:
is
Token is
Feature activation+0.001
Top resid features:
Will
Token Will
Feature activation+0.002
Top resid features:
Come
Token Come
Feature activation+0.000
Top resid features:
,
Token,
Feature activation+0.003
Top resid features:
in
Token in
Feature activation+0.004
Top resid features:
which
Token which
Feature activation+0.004
Top resid features:
he
Token he
Feature activation+3.316
Top resid features:
participated
Token participated
Feature activation+0.125
Top resid features:
,
Token,
Feature activation+0.005
Top resid features:
Armstrong
Token Armstrong
Feature activation+0.023
Top resid features:
told
Token told
Feature activation+0.076
Top resid features:
Sports
Token Sports
Feature activation+0.000
Top resid features:
staff
Token staff
Feature activation+0.002
Top resid features:
reported
Token reported
Feature activation+0.005
Top resid features:
Haz
Token Haz
Feature activation+0.006
Top resid features:
le
Tokenle
Feature activation-0.078
Top resid features:
to
Token to
Feature activation+0.081
Top resid features:
his
Token his
Feature activation+3.265
Top resid features:
parole
Token parole
Feature activation+0.021
Top resid features:
officer
Token officer
Feature activation-0.001
Top resid features:
,
Token,
Feature activation+0.002
Top resid features:
saying
Token saying
Feature activation+0.002
Top resid features:
that
Token that
Feature activation+0.002
Top resid features:
up
Token up
Feature activation+0.008
Top resid features:
a
Token a
Feature activation+0.004
Top resid features:
bank
Token bank
Feature activation+0.006
Top resid features:
account
Token account
Feature activation+0.005
Top resid features:
where
Token where
Feature activation+0.003
Top resid features:
his
Token his
Feature activation+2.778
Top resid features:
donations
Token donations
Feature activation+0.081
Top resid features:
will
Token will
Feature activation+0.004
Top resid features:
be
Token be
Feature activation+0.000
Top resid features:
sent
Token sent
Feature activation+0.003
Top resid features:
to
Token to
Feature activation+0.002
Top resid features:
innocent
Token innocent
Feature activation+0.014
Top resid features:
drivers
Token drivers
Feature activation+0.009
Top resid features:
being
Token being
Feature activation-0.003
Top resid features:
assaulted
Token assaulted
Feature activation-0.003
Top resid features:
by
Token by
Feature activation+0.003
Top resid features:
his
Token his
Feature activation+3.189
Top resid features:
antics
Token antics
Feature activation+0.280
Top resid features:
in
Token in
Feature activation+0.022
Top resid features:
what
Token what
Feature activation+0.002
Top resid features:
appears
Token appears
Feature activation+0.008
Top resid features:
to
Token to
Feature activation+0.002
Top resid features:
roof
Token roof
Feature activation+0.002
Top resid features:
of
Token of
Feature activation+0.001
Top resid features:
the
Token the
Feature activation+0.001
Top resid features:
car
Token car
Feature activation+0.002
Top resid features:
with
Token with
Feature activation-0.001
Top resid features:
his
Token his
Feature activation+3.384
Top resid features:
legs
Token legs
Feature activation-0.001
Top resid features:
positioned
Token positioned
Feature activation-0.004
Top resid features:
over
Token over
Feature activation-0.000
Top resid features:
the
Token the
Feature activation+0.000
Top resid features:
windshield
Token windshield
Feature activation+0.001
Top resid features:
roof
Token roof
Feature activation+0.000
Top resid features:
of
Token of
Feature activation+0.000
Top resid features:
the
Token the
Feature activation+0.000
Top resid features:
car
Token car
Feature activation+0.001
Top resid features:
with
Token with
Feature activation+0.000
Top resid features:
his
Token his
Feature activation+2.994
Top resid features:
legs
Token legs
Feature activation+0.000
Top resid features:
positioned
Token positioned
Feature activation-0.001
Top resid features:
over
Token over
Feature activation-0.000
Top resid features:
the
Token the
Feature activation+0.000
Top resid features:
windshield
Token windshield
Feature activation+0.000
Top resid features:
roof
Token roof
Feature activation+0.001
Top resid features:
of
Token of
Feature activation+0.000
Top resid features:
the
Token the
Feature activation+0.000
Top resid features:
car
Token car
Feature activation+0.001
Top resid features:
with
Token with
Feature activation+0.000
Top resid features:
his
Token his
Feature activation+2.954
Top resid features:
legs
Token legs
Feature activation+0.001
Top resid features:
positioned
Token positioned
Feature activation-0.001
Top resid features:
over
Token over
Feature activation-0.000
Top resid features:
the
Token the
Feature activation+0.000
Top resid features:
windshield
Token windshield
Feature activation+0.000
Top resid features:
an
Token an
Feature activation+0.004
Top resid features:
iron
Token iron
Feature activation+0.001
Top resid features:
deficiency
Token deficiency
Feature activation+0.005
Top resid features:
that
Token that
Feature activation+0.006
Top resid features:
prevented
Token prevented
Feature activation+0.011
Top resid features:
him
Token him
Feature activation+2.319
Top resid features:
from
Token from
Feature activation+0.039
Top resid features:
giving
Token giving
Feature activation+0.024
Top resid features:
full
Token full
Feature activation+0.002
Top resid features:
effort
Token effort
Feature activation+0.005
Top resid features:
on
Token on
Feature activation+0.002
Top resid features:
and
Token and
Feature activation+0.001
Top resid features:
was
Token was
Feature activation+0.008
Top resid features:
str
Token str
Feature activation+0.002
Top resid features:
aining
Tokenaining
Feature activation-0.003
Top resid features:
at
Token at
Feature activation+0.006
Top resid features:
his
Token his
Feature activation+2.883
Top resid features:
leash
Token leash
Feature activation+0.003
Top resid features:
,
Token,
Feature activation+0.000
Top resid features:
according
Token according
Feature activation+0.002
Top resid features:
to
Token to
Feature activation+0.001
Top resid features:
Johnson
Token Johnson
Feature activation+0.001
Top resid features:
past
Token past
Feature activation-0.000
Top resid features:
is
Token is
Feature activation-0.000
Top resid features:
imposing
Token imposing
Feature activation+0.016
Top resid features:
exterior
Token exterior
Feature activation+0.002
Top resid features:
,
Token,
Feature activation-0.000
Top resid features:
he
Token he
Feature activation+2.561
Top resid features:
instantly
Token instantly
Feature activation+0.118
Top resid features:
made
Token made
Feature activation+0.174
Top resid features:
you
Token you
Feature activation+0.016
Top resid features:
feel
Token feel
Feature activation+0.001
Top resid features:
comfortable
Token comfortable
Feature activation-0.001
Top resid features:
,
Token,
Feature activation+0.004
Top resid features:
especially
Token especially
Feature activation+0.000
Top resid features:
in
Token in
Feature activation+0.001
Top resid features:
Italy
Token Italy
Feature activation+0.001
Top resid features:
,
Token,
Feature activation+0.004
Top resid features:
he
Token he
Feature activation+1.905
Top resid features:
remains
Token remains
Feature activation+0.021
Top resid features:
a
Token a
Feature activation+0.008
Top resid features:
victim
Token victim
Feature activation+0.036
Top resid features:
of
Token of
Feature activation+0.018
Top resid features:
the
Token the
Feature activation+0.004
Top resid features:
as
Token as
Feature activation+0.000
Top resid features:
good
Token good
Feature activation+0.001
Top resid features:
as
Token as
Feature activation+0.003
Top resid features:
people
Token people
Feature activation+0.002
Top resid features:
say
Token say
Feature activation+0.003
Top resid features:
he
Token he
Feature activation+2.774
Top resid features:
is
Token is
Feature activation+0.465
Top resid features:
.
Token.
Feature activation+0.004
Top resid features:
Ċ
TokenĊ
Feature activation+0.007
Top resid features:
Ċ
TokenĊ
Feature activation+0.007
Top resid features:
RIP
TokenRIP
Feature activation-0.005
Top resid features:
personality
Token personality
Feature activation+0.010
Top resid features:
:
Token:
Feature activation+0.002
Top resid features:
in
Token in
Feature activation+0.002
Top resid features:
2005
Token 2005
Feature activation+0.001
Top resid features:
,
Token,
Feature activation+0.001
Top resid features:
he
Token he
Feature activation+3.045
Top resid features:
seriously
Token seriously
Feature activation+0.011
Top resid features:
injured
Token injured
Feature activation+0.007
Top resid features:
a
Token a
Feature activation+0.003
Top resid features:
marsh
Token marsh
Feature activation+0.001
Top resid features:
al
Tokenal
Feature activation+0.003
Top resid features:
council
Token council
Feature activation+0.007
Top resid features:
before
Token before
Feature activation+0.003
Top resid features:
its
Token its
Feature activation-0.003
Top resid features:
vote
Token vote
Feature activation+0.004
Top resid features:
terminating
Token terminating
Feature activation+0.003
Top resid features:
his
Token his
Feature activation+2.282
Top resid features:
position
Token position
Feature activation+0.097
Top resid features:
this
Token this
Feature activation+0.002
Top resid features:
month
Token month
Feature activation+0.003
Top resid features:
.
Token.
Feature activation+0.009
Top resid features:
He
Token He
Feature activation+0.381
Top resid features:
handling
Token handling
Feature activation+0.002
Top resid features:
venom
Token venom
Feature activation+0.003
Top resid features:
ous
Tokenous
Feature activation+0.003
Top resid features:
snakes
Token snakes
Feature activation+0.023
Top resid features:
,
Token,
Feature activation+0.002
Top resid features:
he
Token he
Feature activation+2.392
Top resid features:
did
Token did
Feature activation+0.229
Top resid features:
his
Token his
Feature activation+0.324
Top resid features:
best
Token best
Feature activation+0.009
Top resid features:
to
Token to
Feature activation+0.022
Top resid features:
make
Token make
Feature activation+0.022
Top resid features:
to
Token to
Feature activation+0.001
Top resid features:
submit
Token submit
Feature activation-0.000
Top resid features:
a
Token a
Feature activation+0.000
Top resid features:
portrait
Token portrait
Feature activation+0.015
Top resid features:
of
Token of
Feature activation+0.004
Top resid features:
him
Token him
Feature activation+3.204
Top resid features:
that
Token that
Feature activation+0.003
Top resid features:
might
Token might
Feature activation+0.008
Top resid features:
end
Token end
Feature activation+0.002
Top resid features:
up
Token up
Feature activation+0.002
Top resid features:
on
Token on
Feature activation+0.001
Top resid features:

Decoder Weights Distribution

Head 0: 0.05

Head 1: 0.48

Head 2: 0.04

Head 3: 0.07

Head 4: 0.02

Head 5: 0.09

Head 6: 0.09

Head 7: 0.02

Head 8: 0.03

Head 9: 0.04

Head 10: 0.04

Head 11: 0.03

Positive logits

himself3.60

Jr2.79

His2.69

Jr2.62

Sr2.56

rio2.47

Himself2.38

master2.38

ertodd2.36

personally2.35

His2.35

Mr2.34

beard2.31

bearded2.30

ouf2.29

Roger2.28

Markus2.27

iannopoulos2.26

handsome2.24

ovich2.20

Negative logits

womb-2.36

uter-2.31

clitor-2.30

femin-2.30

feminism-2.25

UNESCO-2.24

maternity-2.21

Femin-2.18

feminists-2.15

panties-2.15

moms-2.15

estrogen-2.14

Parenthood-2.14

bikini-2.13

uterus-2.11

pregnancy-2.08

actresses-2.08

hijab-2.08

Women-2.05

maid-2.03

INTERVAL 3.637 - 4.042
CONTAINS 0.000%

it
Token it
Feature activation+0.000
."
Token."
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.051
X
TokenX
Feature activation+0.934
inos
Tokeninos
Feature activation+3.683
,
Token,
Feature activation+1.603
who
Token who
Feature activation+0.524
unsuccessfully
Token unsuccessfully
Feature activation+0.000
sued
Token sued
Feature activation+0.000
to
Token to
Feature activation+0.000
Moment
Token Moment
Feature activation+0.000
grew
Token grew
Feature activation+0.000
out
Token out
Feature activation+0.000
of
Token of
Feature activation+0.000
B
Token B
Feature activation+2.522
ala
Tokenala
Feature activation+3.903
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.855
s
Tokens
Feature activation+0.286
interest
Token interest
Feature activation+0.000
in
Token in
Feature activation+0.074
to
Token to
Feature activation+0.000
the
Token the
Feature activation+0.000
ruling
Token ruling
Feature activation+0.000
.
Token.
Feature activation+0.000
Haz
Token Haz
Feature activation+3.207
le
Tokenle
Feature activation+4.042
then
Token then
Feature activation+0.828
found
Token found
Feature activation+0.209
himself
Token himself
Feature activation+0.000
back
Token back
Feature activation+0.000
in
Token in
Feature activation+0.000
fell
Token fell
Feature activation+0.000
off
Token off
Feature activation+0.000
and
Token and
Feature activation+0.000
the
Token the
Feature activation+0.000
pit
Token pit
Feature activation+1.707
bull
Token bull
Feature activation+4.010
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.416
jaws
Token jaws
Feature activation+0.000
locked
Token locked
Feature activation+0.000

INTERVAL 3.233 - 3.637
CONTAINS 0.000%

During
TokenDuring
Feature activation+0.000
the
Token the
Feature activation+0.000
panel
Token panel
Feature activation+0.000
session
Token session
Feature activation+0.000
,
Token,
Feature activation+0.359
Armstrong
Token Armstrong
Feature activation+3.483
told
Token told
Feature activation+0.555
delegates
Token delegates
Feature activation+0.000
the
Token the
Feature activation+0.000
move
Token move
Feature activation+0.000
was
Token was
Feature activation+0.000

INTERVAL 2.829 - 3.233
CONTAINS 0.001%

vehicle
Token vehicle
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
Trooper
Token Trooper
Feature activation+3.048
diseng
Token diseng
Feature activation+0.000
aged
Tokenaged
Feature activation+0.000
contact
Token contact
Feature activation+0.000
with
Token with
Feature activation+0.000
SC
Token SC
Feature activation+1.346
.
Token.
Feature activation+0.171
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.070
P
TokenP
Feature activation+0.221
ant
Tokenant
Feature activation+1.470
ani
Tokenani
Feature activation+2.901
would
Token would
Feature activation+0.628
have
Token have
Feature activation+0.153
been
Token been
Feature activation+0.313
44
Token 44
Feature activation+0.000
on
Token on
Feature activation+0.000
Capitol
Token Capitol
Feature activation+0.000
Bever
Token Bever
Feature activation+0.000
age
Tokenage
Feature activation+0.000
Sales
Token Sales
Feature activation+0.000
.
Token.
Feature activation+0.000
Hans
Token Hans
Feature activation+2.904
and
Token and
Feature activation+0.256
the
Token the
Feature activation+0.000
Morris
Token Morris
Feature activation+0.219
seys
Tokenseys
Feature activation+0.000
welcomed
Token welcomed
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
pit
Token pit
Feature activation+1.236
bull
Token bull
Feature activation+2.923
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.161
s
Tokens
Feature activation+0.119
owner
Token owner
Feature activation+0.000
had
Token had
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.300
Apparently
TokenApparently
Feature activation+0.016
this
Token this
Feature activation+0.000
wildlife
Token wildlife
Feature activation+0.000
producer
Token producer
Feature activation+3.057
and
Token and
Feature activation+1.046
director
Token director
Feature activation+0.703
for
Token for
Feature activation+0.000
the
Token the
Feature activation+0.000
popular
Token popular
Feature activation+0.000

INTERVAL 2.425 - 2.829
CONTAINS 0.004%

era
Token era
Feature activation+0.000
before
Token before
Feature activation+0.000
ubiquitous
Token ubiquitous
Feature activation+0.000
computers
Token computers
Feature activation+0.000
,
Token,
Feature activation+0.000
Wild
Token Wild
Feature activation+2.505
stein
Tokenstein
Feature activation+2.395
pores
Token pores
Feature activation+0.000
over
Token over
Feature activation+0.000
re
Token re
Feature activation+0.000
ams
Tokenams
Feature activation+0.000
years
Token years
Feature activation+0.000
old
Token old
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.022
Sullivan
TokenSullivan
Feature activation+2.675
violated
Token violated
Feature activation+0.000
his
Token his
Feature activation+0.000
probation
Token probation
Feature activation+0.000
and
Token and
Feature activation+0.000
was
Token was
Feature activation+0.000
sharing
Token sharing
Feature activation+0.000
economy
Token economy
Feature activation+0.000
,"
Token,"
Feature activation+0.000
said
Token said
Feature activation+0.000
Par
Token Par
Feature activation+1.101
se
Tokense
Feature activation+2.524
yan
Tokenyan
Feature activation+2.275
,
Token,
Feature activation+1.268
who
Token who
Feature activation+1.401
is
Token is
Feature activation+0.027
now
Token now
Feature activation+0.391
."
Token."
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Par
TokenPar
Feature activation+0.860
se
Tokense
Feature activation+2.000
yan
Tokenyan
Feature activation+2.425
said
Token said
Feature activation+0.000
there
Token there
Feature activation+0.000
was
Token was
Feature activation+0.000
a
Token a
Feature activation+0.000
mix
Token mix
Feature activation+0.000
,
Token,
Feature activation+0.000
said
Token said
Feature activation+0.000
Oy
Token Oy
Feature activation+1.564
ek
Tokenek
Feature activation+1.016
an
Tokenan
Feature activation+1.962
mi
Tokenmi
Feature activation+2.666
began
Token began
Feature activation+0.784
working
Token working
Feature activation+0.000
for
Token for
Feature activation+0.000
the
Token the
Feature activation+0.000
facility
Token facility
Feature activation+0.000

INTERVAL 2.021 - 2.425
CONTAINS 0.009%

do
Token do
Feature activation+0.000
that
Token that
Feature activation+0.000
,
Token,
Feature activation+0.000
given
Token given
Feature activation+0.000
Quint
Token Quint
Feature activation+1.826
ana
Tokenana
Feature activation+2.179
is
Token is
Feature activation+0.627
not
Token not
Feature activation+0.226
just
Token just
Feature activation+0.226
a
Token a
Feature activation+0.171
gifted
Token gifted
Feature activation+0.123
and
Token and
Feature activation+0.000
journals
Token journals
Feature activation+0.000
.[
Token.[
Feature activation+0.000
124
Token124
Feature activation+0.000
]
Token]
Feature activation+0.698
Byron
Token Byron
Feature activation+2.171
and
Token and
Feature activation+0.459
other
Token other
Feature activation+0.037
writers
Token writers
Feature activation+0.028
,
Token,
Feature activation+0.000
such
Token such
Feature activation+0.007
be
Token be
Feature activation+0.000
true
Token true
Feature activation+0.000
,
Token,
Feature activation+0.000
Officer
Token Officer
Feature activation+0.933
McGu
Token McGu
Feature activation+1.280
iness
Tokeniness
Feature activation+2.388
is
Token is
Feature activation+0.000
lying
Token lying
Feature activation+0.000
through
Token through
Feature activation+0.000
his
Token his
Feature activation+0.000
teeth
Token teeth
Feature activation+0.000
is
Token is
Feature activation+0.000
ore
Token ore
Feature activation+0.000
.
Token.
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
Lowell
Token Lowell
Feature activation+2.090
also
Token also
Feature activation+0.100
voiced
Token voiced
Feature activation+0.000
doubts
Token doubts
Feature activation+0.000
that
Token that
Feature activation+0.000
technology
Token technology
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Some
TokenSome
Feature activation+0.000
of
Token of
Feature activation+0.000
Tun
Token Tun
Feature activation+1.411
n
Tokenn
Feature activation+2.611
idge
Tokenidge
Feature activation+2.394
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.066
best
Token best
Feature activation+0.000
advice
Token advice
Feature activation+0.000

INTERVAL 1.617 - 2.021
CONTAINS 0.018%

a
Token a
Feature activation+0.000
fly
Token fly
Feature activation+0.000
route
Token route
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
verett
Tokenverett
Feature activation+1.943
is
Token is
Feature activation+0.296
running
Token running
Feature activation+0.000
with
Token with
Feature activation+0.000
him
Token him
Feature activation+0.000
.
Token.
Feature activation+0.000
extra
Token extra
Feature activation+0.000
push
Token push
Feature activation+0.000
.
Token.
Feature activation+0.000
According
Token According
Feature activation+0.000
to
Token to
Feature activation+0.000
TM
Token TM
Feature activation+1.773
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.000
song
Token song
Feature activation+0.000
got
Token got
Feature activation+0.000
passed
Token passed
Feature activation+0.000
will
Token will
Feature activation+0.000
attempt
Token attempt
Feature activation+0.000
to
Token to
Feature activation+0.000
ascertain
Token ascertain
Feature activation+0.000
the
Token the
Feature activation+0.106
boy
Token boy
Feature activation+1.722
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.611
wishes
Token wishes
Feature activation+0.000
.
Token.
Feature activation+0.000
muzzle
Token muzzle
Feature activation+0.252
fell
Token fell
Feature activation+0.000
off
Token off
Feature activation+0.000
and
Token and
Feature activation+0.000
the
Token the
Feature activation+0.000
pit
Token pit
Feature activation+1.707
bull
Token bull
Feature activation+4.010
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.416
jaws
Token jaws
Feature activation+0.000
rated
Tokenrated
Feature activation+0.063
,
Token,
Feature activation+0.077
Sh
Token Sh
Feature activation+1.593
re
Tokenre
Feature activation+1.803
ate
Tokenate
Feature activation+1.662
h
Tokenh
Feature activation+1.886
took
Token took
Feature activation+0.585
matters
Token matters
Feature activation+0.000
into
Token into
Feature activation+0.000
his
Token his
Feature activation+0.000
own
Token own
Feature activation+0.000

INTERVAL 1.212 - 1.617
CONTAINS 0.033%

the
Token the
Feature activation+0.000
penalty
Token penalty
Feature activation+0.000
area
Token area
Feature activation+0.000
.
Token.
Feature activation+0.000
M
Token M
Feature activation+0.698
ert
Tokenert
Feature activation+1.498
es
Tokenes
Feature activation+0.000
acker
Tokenacker
Feature activation+0.000
attacked
Token attacked
Feature activation+0.000
it
Token it
Feature activation+0.000
with
Token with
Feature activation+0.000
court
Token court
Feature activation+0.000
documents
Token documents
Feature activation+0.000
,
Token,
Feature activation+0.000
which
Token which
Feature activation+0.000
Ph
Token Ph
Feature activation+0.764
o
Tokeno
Feature activation+1.391
took
Token took
Feature activation+0.000
from
Token from
Feature activation+0.321
work
Token work
Feature activation+0.000
between
Token between
Feature activation+0.000
2010
Token 2010
Feature activation+0.000
In
TokenIn
Feature activation+0.000
Top
Token Top
Feature activation+0.000
5
Token 5
Feature activation+0.000
s
Tokens
Feature activation+0.000
,
Token,
Feature activation+0.000
Alex
Token Alex
Feature activation+1.461
creates
Token creates
Feature activation+0.000
a
Token a
Feature activation+0.000
typical
Token typical
Feature activation+0.000
top
Token top
Feature activation+0.000
5
Token 5
Feature activation+0.000
in
Token in
Feature activation+0.000
2015
Token 2015
Feature activation+0.000
,
Token,
Feature activation+0.000
Str
Token Str
Feature activation+0.796
ick
Tokenick
Feature activation+1.000
land
Tokenland
Feature activation+1.309
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.065
s
Tokens
Feature activation+0.270
first
Token first
Feature activation+0.000
full
Token full
Feature activation+0.000
funds
Token funds
Feature activation+0.000
.
Token.
Feature activation+0.000
Sh
Token Sh
Feature activation+0.968
k
Tokenk
Feature activation+1.972
rel
Tokenrel
Feature activation+1.523
i
Tokeni
Feature activation+1.544
lied
Token lied
Feature activation+0.000
to
Token to
Feature activation+0.000
obtain
Token obtain
Feature activation+0.000
investors
Token investors
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000

INTERVAL 0.808 - 1.212
CONTAINS 0.079%

.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Yes
TokenYes
Feature activation+0.000
,
Token,
Feature activation+0.000
Harden
Token Harden
Feature activation+0.842
does
Token does
Feature activation+0.130
exagger
Token exagger
Feature activation+0.000
ate
Tokenate
Feature activation+0.000
contact
Token contact
Feature activation+0.000
.
Token.
Feature activation+0.000
live
Token live
Feature activation+0.000
in
Token in
Feature activation+0.000
Mexico
Token Mexico
Feature activation+0.000
.
Token.
Feature activation+0.388
Sol
Token Sol
Feature activation+0.171
is
Tokenis
Feature activation+1.134
had
Token had
Feature activation+0.865
no
Token no
Feature activation+0.040
insurance
Token insurance
Feature activation+0.000
and
Token and
Feature activation+0.115
no
Token no
Feature activation+0.000
My
Token My
Feature activation+0.000
condolences
Token condolences
Feature activation+0.000
go
Token go
Feature activation+0.000
out
Token out
Feature activation+0.000
to
Token to
Feature activation+0.000
Daniel
Token Daniel
Feature activation+0.932
's
Token's
Feature activation+0.057
mother
Token mother
Feature activation+0.000
and
Token and
Feature activation+0.000
sister
Token sister
Feature activation+0.000
.
Token.
Feature activation+0.000
authorities
Token authorities
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ch
TokenCh
Feature activation+0.464
omsky
Tokenomsky
Feature activation+1.075
is
Token is
Feature activation+0.405
a
Token a
Feature activation+0.000
professor
Token professor
Feature activation+0.000
of
Token of
Feature activation+0.000
lingu
Token lingu
Feature activation+0.000
place
Token place
Feature activation+0.000
nearly
Token nearly
Feature activation+0.000
two
Token two
Feature activation+0.000
weeks
Token weeks
Feature activation+0.000
after
Token after
Feature activation+0.045
West
Token West
Feature activation+0.932
was
Token was
Feature activation+0.459
released
Token released
Feature activation+0.000
from
Token from
Feature activation+0.012
the
Token the
Feature activation+0.000
UCLA
Token UCLA
Feature activation+0.000

INTERVAL 0.404 - 0.808
CONTAINS 0.246%

les
Tokenles
Feature activation+0.000
me
Token me
Feature activation+0.000
,
Token,
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
he
Token he
Feature activation+0.794
says
Token says
Feature activation+0.165
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.172
As
TokenAs
Feature activation+0.167
him
Token him
Feature activation+0.000
what
Token what
Feature activation+0.000
it
Token it
Feature activation+0.000
meant
Token meant
Feature activation+0.000
because
Token because
Feature activation+0.000
he
Token he
Feature activation+0.773
had
Token had
Feature activation+0.060
to
Token to
Feature activation+0.000
go
Token go
Feature activation+0.000
to
Token to
Feature activation+0.000
school
Token school
Feature activation+0.000
it
Token it
Feature activation+0.000
being
Token being
Feature activation+0.000
eight
Token eight
Feature activation+0.000
months
Token months
Feature activation+0.000
since
Token since
Feature activation+0.000
O
Token O
Feature activation+0.450
'
Token'
Feature activation+1.343
Neill
TokenNeill
Feature activation+1.563
's
Token's
Feature activation+0.223
death
Token death
Feature activation+0.000
,
Token,
Feature activation+0.000
it
Token it
Feature activation+0.000
?
Token?
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.288
Ċ
TokenĊ
Feature activation+0.408
He
TokenHe
Feature activation+0.499
then
Token then
Feature activation+0.430
described
Token described
Feature activation+0.000
homosexuals
Token homosexuals
Feature activation+0.000
with
Token with
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.106
Arsenal
TokenArsenal
Feature activation+0.000
star
Token star
Feature activation+0.078
:
Token:
Feature activation+0.114
Sans
Token Sans
Feature activation+0.533
om
Tokenom
Feature activation+0.000
,
Token,
Feature activation+0.000
who
Token who
Feature activation+0.720
at
Token at
Feature activation+0.000
his
Token his
Feature activation+0.195

INTERVAL 0.000 - 0.404
CONTAINS 99.609%

Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
As
TokenAs
Feature activation+0.000
semble
Tokensemble
Feature activation+0.000
the
Token the
Feature activation+0.000
tax
Token tax
Feature activation+0.000
rates
Token rates
Feature activation+0.000
and
Token and
Feature activation+0.000
shipping
Token shipping
Feature activation+0.000
scheme
Token scheme
Feature activation+0.000
into
Token into
Feature activation+0.000
in
Tokenin
Feature activation+0.000
support
Token support
Feature activation+0.000
for
Token for
Feature activation+0.000
the
Token the
Feature activation+0.000
Cookies
Token Cookies
Feature activation+0.000
To
Token To
Feature activation+0.000
Head
Token Head
Feature activation+0.000
ers
Tokeners
Feature activation+0.000
X
Token X
Feature activation+0.000
SR
TokenSR
Feature activation+0.000
F
TokenF
Feature activation+0.000
poor
Token poor
Feature activation+0.000
people
Token people
Feature activation+0.000
,
Token,
Feature activation+0.000
minorities
Token minorities
Feature activation+0.000
,
Token,
Feature activation+0.000
and
Token and
Feature activation+0.000
immigrants
Token immigrants
Feature activation+0.000
in
Token in
Feature activation+0.000
a
Token a
Feature activation+0.000
way
Token way
Feature activation+0.000
that
Token that
Feature activation+0.000
"
Token "
Feature activation+0.000
not
Tokennot
Feature activation+0.000
indifferent
Token indifferent
Feature activation+0.000
"
Token"
Feature activation+0.000
to
Token to
Feature activation+0.000
the
Token the
Feature activation+0.000
f
Token f
Feature activation+0.000
ates
Tokenates
Feature activation+0.000
of
Token of
Feature activation+0.000
families
Token families
Feature activation+0.000
fleeing
Token fleeing
Feature activation+0.000
a
Token a
Feature activation+0.000
decent
Token decent
Feature activation+0.000
chance
Token chance
Feature activation+0.000
The
Token The
Feature activation+0.000
D
Token D
Feature activation+0.000
rought
Tokenrought
Feature activation+0.000
ends
Token ends
Feature activation+0.000
.
Token.
Feature activation+0.000
(
Token (
Feature activation+0.000
Yes
TokenYes
Feature activation+0.000
,
Token,
Feature activation+0.000

BOTTOM ACTIVATIONS
MIN = 0.000

atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
Development
Token Development
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
<|endoftext|>
Token<|endoftext|>
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000

Top feature 7 in H7.1: (feature 27279)

TOP ACTIVATIONS
MAX = 5.805

single
Token single
Feature activation+0.000
site
Token site
Feature activation+0.000
whereas
Token whereas
Feature activation+0.000
T
Token T
Feature activation+2.948
ure
Tokenure
Feature activation+5.065
tsky
Tokentsky
Feature activation+5.805
's
Token's
Feature activation+1.875
team
Token team
Feature activation+0.000
<|endoftext|>
Token<|endoftext|>
Feature activation+0.000
Looks
TokenLooks
Feature activation+0.000
like
Token like
Feature activation+0.000
3
Token3
Feature activation+0.000
]
Token]
Feature activation+0.412
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.158
G
TokenG
Feature activation+2.164
ellar
Tokenellar
Feature activation+5.382
was
Token was
Feature activation+2.429
given
Token given
Feature activation+0.000
a
Token a
Feature activation+0.000
partial
Token partial
Feature activation+0.000
scholarship
Token scholarship
Feature activation+0.000
delivering
Token delivering
Feature activation+0.000
."
Token."
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
E
TokenE
Feature activation+0.013
arl
Tokenarl
Feature activation+5.130
would
Token would
Feature activation+2.180
say
Token say
Feature activation+0.978
later
Token later
Feature activation+0.000
that
Token that
Feature activation+0.672
her
Token her
Feature activation+0.549
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Judge
TokenJudge
Feature activation+3.363
N
Token N
Feature activation+3.298
en
Tokenen
Feature activation+3.670
ney
Tokenney
Feature activation+5.085
also
Token also
Feature activation+1.176
hears
Token hears
Feature activation+0.511
"
Token "
Feature activation+0.000
k
Tokenk
Feature activation+0.000
hal
Tokenhal
Feature activation+0.000
a
Token a
Feature activation+0.000
single
Token single
Feature activation+0.000
site
Token site
Feature activation+0.000
whereas
Token whereas
Feature activation+0.000
T
Token T
Feature activation+2.948
ure
Tokenure
Feature activation+5.065
tsky
Tokentsky
Feature activation+5.805
's
Token's
Feature activation+1.875
team
Token team
Feature activation+0.000
<|endoftext|>
Token<|endoftext|>
Feature activation+0.000
Looks
TokenLooks
Feature activation+0.000
a
Token a
Feature activation+0.000
bit
Token bit
Feature activation+0.000
longer
Token longer
Feature activation+0.000
.
Token.
Feature activation+0.000
Wood
Token Wood
Feature activation+4.452
ruff
Tokenruff
Feature activation+4.847
says
Token says
Feature activation+0.210
first
Token first
Feature activation+0.000
they
Token they
Feature activation+0.000
have
Token have
Feature activation+0.000
to
Token to
Feature activation+0.000
tweet
Token tweet
Feature activation+0.000
myself
Token myself
Feature activation+0.000
,
Token,
Feature activation+0.000
but
Token but
Feature activation+0.000
Ne
Token Ne
Feature activation+1.778
Ne
TokenNe
Feature activation+4.809
seems
Token seems
Feature activation+0.812
to
Token to
Feature activation+0.494
be
Token be
Feature activation+0.790
cool
Token cool
Feature activation+0.054
with
Token with
Feature activation+0.182
low
Token low
Feature activation+0.000
profile
Token profile
Feature activation+0.000
."
Token."
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.811
Davis
TokenDavis
Feature activation+4.687
,
Token,
Feature activation+4.073
who
Token who
Feature activation+2.567
won
Token won
Feature activation+0.355
the
Token the
Feature activation+0.081
supporting
Token supporting
Feature activation+0.000
Pl
TokenPl
Feature activation+0.000
o
Tokeno
Feature activation+0.000
etz
Tokenetz
Feature activation+0.000
suggests
Token suggests
Feature activation+0.000
that
Token that
Feature activation+0.000
Lei
Token Lei
Feature activation+4.576
âĢ
TokenâĢ
Feature activation+1.140
Ļ
TokenĻ
Feature activation+2.380
s
Tokens
Feature activation+0.000
research
Token research
Feature activation+0.000
could
Token could
Feature activation+0.000
roph
Tokenroph
Feature activation+0.000
ilia
Tokenilia
Feature activation+0.000
specifically
Token specifically
Feature activation+0.000
,
Token,
Feature activation+0.000
John
Token John
Feature activation+1.991
sson
Tokensson
Feature activation+4.496
argued
Token argued
Feature activation+0.000
that
Token that
Feature activation+0.000
people
Token people
Feature activation+0.000
should
Token should
Feature activation+0.000
be
Token be
Feature activation+0.000
male
Token male
Feature activation+0.000
so
Token so
Feature activation+0.000
I
Token I
Feature activation+0.000
bo
Tokenbo
Feature activation+1.149
ly
Tokenly
Feature activation+1.148
a
Tokena
Feature activation+4.480
can
Token can
Feature activation+0.802
breed
Token breed
Feature activation+0.000
convention
Token convention
Feature activation+0.000
ally
Tokenally
Feature activation+0.000
next
Token next
Feature activation+0.000
wait
Token wait
Feature activation+0.000
a
Token a
Feature activation+0.000
bit
Token bit
Feature activation+0.000
longer
Token longer
Feature activation+0.000
.
Token.
Feature activation+0.000
Wood
Token Wood
Feature activation+4.452
ruff
Tokenruff
Feature activation+4.847
says
Token says
Feature activation+0.210
first
Token first
Feature activation+0.000
they
Token they
Feature activation+0.000
have
Token have
Feature activation+0.000
she
Token she
Feature activation+0.968
was
Token was
Feature activation+0.428
shot
Token shot
Feature activation+0.000
,
Token,
Feature activation+0.000
Dam
Token Dam
Feature activation+3.230
ond
Tokenond
Feature activation+4.379
was
Token was
Feature activation+1.176
standing
Token standing
Feature activation+0.000
,
Token,
Feature activation+0.000
in
Token in
Feature activation+0.000
her
Token her
Feature activation+0.000
third
Token third
Feature activation+0.000
indictment
Token indictment
Feature activation+0.000
count
Token count
Feature activation+0.000
alleges
Token alleges
Feature activation+0.000
You
Token You
Feature activation+1.407
n
Tokenn
Feature activation+4.355
illegally
Token illegally
Feature activation+0.000
tam
Token tam
Feature activation+0.000
pered
Tokenpered
Feature activation+0.000
with
Token with
Feature activation+0.000
witness
Token witness
Feature activation+0.000
."
Token."
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
S
TokenS
Feature activation+0.445
ark
Tokenark
Feature activation+3.389
is
Tokenis
Feature activation+4.321
ian
Tokenian
Feature activation+3.155
leaps
Token leaps
Feature activation+0.045
from
Token from
Feature activation+0.000
his
Token his
Feature activation+0.000
bed
Token bed
Feature activation+0.000
able
Tokenable
Feature activation+0.000
homicide
Token homicide
Feature activation+0.000
,
Token,
Feature activation+0.000
Mas
Token Mas
Feature activation+0.132
ip
Tokenip
Feature activation+3.668
a
Tokena
Feature activation+4.138
found
Token found
Feature activation+0.000
that
Token that
Feature activation+0.000
"
Token "
Feature activation+0.000
a
Tokena
Feature activation+0.000
reasonable
Token reasonable
Feature activation+0.000
.
Token.
Feature activation+0.308
The
Token The
Feature activation+0.000
crash
Token crash
Feature activation+0.015
occurred
Token occurred
Feature activation+0.000
when
Token when
Feature activation+0.335
Williams
Token Williams
Feature activation+4.136
started
Token started
Feature activation+0.000
to
Token to
Feature activation+0.000
legally
Token legally
Feature activation+0.000
proceed
Token proceed
Feature activation+0.000
.
Token.
Feature activation+0.000
video
Token video
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
incident
Token incident
Feature activation+0.000
showed
Token showed
Feature activation+0.049
Shelby
Token Shelby
Feature activation+4.095
and
Token and
Feature activation+0.516
several
Token several
Feature activation+0.000
other
Token other
Feature activation+0.131
officers
Token officers
Feature activation+0.472
surrounding
Token surrounding
Feature activation+0.374
to
Token to
Feature activation+0.000
Tyr
Token Tyr
Feature activation+0.000
one
Tokenone
Feature activation+0.000
Taylor
Token Taylor
Feature activation+0.000
,
Token,
Feature activation+0.000
Ty
Token Ty
Feature activation+4.082
ke
Tokenke
Feature activation+3.388
's
Token's
Feature activation+0.355
responsible
Token responsible
Feature activation+0.000
trainer
Token trainer
Feature activation+0.000
at
Token at
Feature activation+0.000
profile
Token profile
Feature activation+0.000
."
Token."
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.811
Davis
TokenDavis
Feature activation+4.687
,
Token,
Feature activation+4.073
who
Token who
Feature activation+2.567
won
Token won
Feature activation+0.355
the
Token the
Feature activation+0.081
supporting
Token supporting
Feature activation+0.000
actress
Token actress
Feature activation+0.000

Top DFA by src position
MAX = 6.130

as
Token as
Feature activation-0.000
Top resid features:
T
Token T
Feature activation+0.001
Top resid features:
ure
Tokenure
Feature activation+0.000
Top resid features:
tsky
Tokentsky
Feature activation+0.003
Top resid features:
and
Token and
Feature activation+0.022
Top resid features:
her
Token her
Feature activation+6.130
Top resid features:
team
Token team
Feature activation+0.041
Top resid features:
have
Token have
Feature activation+0.006
Top resid features:
,
Token,
Feature activation+0.003
Top resid features:
scientists
Token scientists
Feature activation+0.018
Top resid features:
can
Token can
Feature activation+0.007
Top resid features:
figure
Token figure
Feature activation+0.007
Top resid features:
sk
Token sk
Feature activation+0.001
Top resid features:
ater
Tokenater
Feature activation+0.008
Top resid features:
to
Token to
Feature activation+0.003
Top resid features:
which
Token which
Feature activation-0.003
Top resid features:
she
Token she
Feature activation+5.561
Top resid features:
finished
Token finished
Feature activation+0.006
Top resid features:
in
Token in
Feature activation+0.000
Top resid features:
third
Token third
Feature activation+0.001
Top resid features:
place
Token place
Feature activation+0.001
Top resid features:
at
Token at
Feature activation-0.000
Top resid features:
Pug
Token Pug
Feature activation+0.000
Top resid features:
et
Tokenet
Feature activation-0.000
Top resid features:
Sound
Token Sound
Feature activation+0.003
Top resid features:
.
Token.
Feature activation+0.005
Top resid features:
With
Token With
Feature activation+0.002
Top resid features:
her
Token her
Feature activation+4.597
Top resid features:
vision
Token vision
Feature activation+0.047
Top resid features:
,
Token,
Feature activation+0.002
Top resid features:
her
Token her
Feature activation+0.276
Top resid features:
ten
Token ten
Feature activation+0.004
Top resid features:
acity
Tokenacity
Feature activation+0.009
Top resid features:
Judge
Token Judge
Feature activation+0.018
Top resid features:
N
Token N
Feature activation+0.007
Top resid features:
en
Tokenen
Feature activation+0.002
Top resid features:
ney
Tokenney
Feature activation+0.014
Top resid features:
says
Token says
Feature activation+0.031
Top resid features:
she
Token she
Feature activation+5.047
Top resid features:
ver
Token ver
Feature activation+0.002
Top resid features:
ifies
Tokenifies
Feature activation+0.016
Top resid features:
the
Token the
Feature activation+0.001
Top resid features:
husband
Token husband
Feature activation-0.000
Top resid features:
's
Token's
Feature activation+0.000
Top resid features:
as
Token as
Feature activation+0.000
Top resid features:
T
Token T
Feature activation+0.001
Top resid features:
ure
Tokenure
Feature activation+0.002
Top resid features:
tsky
Tokentsky
Feature activation-0.157
Top resid features:
and
Token and
Feature activation+0.054
Top resid features:
her
Token her
Feature activation+5.562
Top resid features:
team
Token team
Feature activation+0.006
Top resid features:
have
Token have
Feature activation+0.002
Top resid features:
,
Token,
Feature activation+0.000
Top resid features:
scientists
Token scientists
Feature activation+0.003
Top resid features:
can
Token can
Feature activation+0.002
Top resid features:
months
Token months
Feature activation+0.001
Top resid features:
.
Token.
Feature activation+0.000
Top resid features:
Wood
Token Wood
Feature activation-0.000
Top resid features:
ruff
Tokenruff
Feature activation+0.000
Top resid features:
and
Token and
Feature activation+0.024
Top resid features:
her
Token her
Feature activation+5.130
Top resid features:
team
Token team
Feature activation+0.058
Top resid features:
let
Token let
Feature activation+0.011
Top resid features:
those
Token those
Feature activation+0.000
Top resid features:
babies
Token babies
Feature activation-0.001
Top resid features:
make
Token make
Feature activation-0.000
Top resid features:
Ne
TokenNe
Feature activation-0.006
Top resid features:
made
Token made
Feature activation+0.087
Top resid features:
it
Token it
Feature activation+0.006
Top resid features:
clear
Token clear
Feature activation+0.046
Top resid features:
that
Token that
Feature activation+0.021
Top resid features:
she
Token she
Feature activation+4.761
Top resid features:
is
Token is
Feature activation+0.035
Top resid features:
happy
Token happy
Feature activation+0.008
Top resid features:
for
Token for
Feature activation+0.007
Top resid features:
Kim
Token Kim
Feature activation+0.005
Top resid features:
and
Token and
Feature activation+0.012
Top resid features:
Ċ
TokenĊ
Feature activation+0.004
Top resid features:
"
Token"
Feature activation-0.000
Top resid features:
Yes
TokenYes
Feature activation+0.001
Top resid features:
,
Token,
Feature activation+0.001
Top resid features:
Ge
Token Ge
Feature activation+0.007
Top resid features:
ena
Tokenena
Feature activation+4.443
Top resid features:
Davis
Token Davis
Feature activation+0.177
Top resid features:
is
Token is
Feature activation+0.077
Top resid features:
one
Token one
Feature activation+0.045
Top resid features:
of
Token of
Feature activation+0.007
Top resid features:
the
Token the
Feature activation+0.002
Top resid features:
âĢ
TokenâĢ
Feature activation+0.000
Top resid features:
Ŀ
TokenĿ
Feature activation-0.000
Top resid features:
says
Token says
Feature activation+0.011
Top resid features:
Lei
Token Lei
Feature activation+0.032
Top resid features:
.
Token.
Feature activation+0.030
Top resid features:
She
Token She
Feature activation+4.782
Top resid features:
tested
Token tested
Feature activation+0.017
Top resid features:
it
Token it
Feature activation-0.009
Top resid features:
on
Token on
Feature activation-0.000
Top resid features:
a
Token a
Feature activation-0.000
Top resid features:
cake
Token cake
Feature activation+0.001
Top resid features:
being
Token being
Feature activation-0.000
Top resid features:
disgusting
Token disgusting
Feature activation+0.000
Top resid features:
,
Token,
Feature activation+0.000
Top resid features:
âĢ
TokenâĢ
Feature activation+0.000
Top resid features:
Ŀ
TokenĿ
Feature activation+0.000
Top resid features:
she
Token she
Feature activation+4.748
Top resid features:
said
Token said
Feature activation+0.034
Top resid features:
.
Token.
Feature activation+0.005
Top resid features:
Ċ
TokenĊ
Feature activation+0.003
Top resid features:
Ċ
TokenĊ
Feature activation+0.003
Top resid features:
âĢ
TokenâĢ
Feature activation+0.000
Top resid features:
.
Token.
Feature activation-0.003
Top resid features:
In
Token In
Feature activation+0.001
Top resid features:
that
Token that
Feature activation+0.000
Top resid features:
time
Token time
Feature activation-0.001
Top resid features:
,
Token,
Feature activation-0.000
Top resid features:
she
Token she
Feature activation+4.987
Top resid features:
has
Token has
Feature activation+0.011
Top resid features:
never
Token never
Feature activation-0.001
Top resid features:
shared
Token shared
Feature activation+0.014
Top resid features:
water
Token water
Feature activation-0.001
Top resid features:
with
Token with
Feature activation+0.001
Top resid features:
months
Token months
Feature activation-0.000
Top resid features:
.
Token.
Feature activation+0.000
Top resid features:
Wood
Token Wood
Feature activation+0.001
Top resid features:
ruff
Tokenruff
Feature activation-0.158
Top resid features:
and
Token and
Feature activation+0.011
Top resid features:
her
Token her
Feature activation+4.962
Top resid features:
team
Token team
Feature activation+0.011
Top resid features:
let
Token let
Feature activation+0.004
Top resid features:
those
Token those
Feature activation+0.000
Top resid features:
babies
Token babies
Feature activation-0.000
Top resid features:
make
Token make
Feature activation-0.000
Top resid features:
the
Token the
Feature activation+0.002
Top resid features:
abdomen
Token abdomen
Feature activation+0.003
Top resid features:
and
Token and
Feature activation+0.001
Top resid features:
ultimately
Token ultimately
Feature activation-0.002
Top resid features:
killing
Token killing
Feature activation-0.003
Top resid features:
her
Token her
Feature activation+3.839
Top resid features:
.
Token.
Feature activation+0.022
Top resid features:
Ċ
TokenĊ
Feature activation+0.005
Top resid features:
Ċ
TokenĊ
Feature activation+0.006
Top resid features:
When
TokenWhen
Feature activation-0.002
Top resid features:
she
Token she
Feature activation+1.090
Top resid features:
harbor
Token harbor
Feature activation+0.009
Top resid features:
ing
Tokening
Feature activation+0.005
Top resid features:
them
Token them
Feature activation+0.006
Top resid features:
,
Token,
Feature activation+0.008
Top resid features:
alleging
Token alleging
Feature activation+0.009
Top resid features:
she
Token she
Feature activation+4.629
Top resid features:
did
Token did
Feature activation+0.022
Top resid features:
so
Token so
Feature activation+0.007
Top resid features:
âĢ
Token âĢ
Feature activation+0.012
Top resid features:
ľ
Tokenľ
Feature activation+0.000
Top resid features:
for
Tokenfor
Feature activation+0.001
Top resid features:
've
Token've
Feature activation+0.000
Top resid features:
already
Token already
Feature activation-0.000
Top resid features:
built
Token built
Feature activation-0.000
Top resid features:
it
Token it
Feature activation+0.000
Top resid features:
,"
Token,"
Feature activation-0.000
Top resid features:
she
Token she
Feature activation+4.635
Top resid features:
says
Token says
Feature activation+0.037
Top resid features:
.
Token.
Feature activation+0.005
Top resid features:
"
Token "
Feature activation+0.000
Top resid features:
Ren
TokenRen
Feature activation+0.001
Top resid features:
ov
Tokenov
Feature activation+0.002
Top resid features:
to
Token to
Feature activation+0.000
Top resid features:
conduct
Token conduct
Feature activation-0.000
Top resid features:
himself
Token himself
Feature activation-0.001
Top resid features:
reasonably
Token reasonably
Feature activation-0.000
Top resid features:
,"
Token,"
Feature activation-0.002
Top resid features:
she
Token she
Feature activation+4.219
Top resid features:
said
Token said
Feature activation+0.091
Top resid features:
.
Token.
Feature activation+0.015
Top resid features:
"
Token "
Feature activation+0.005
Top resid features:
I
TokenI
Feature activation+0.094
Top resid features:
am
Token am
Feature activation+0.028
Top resid features:
a
Token a
Feature activation+0.000
Top resid features:
left
Token left
Feature activation-0.001
Top resid features:
turn
Token turn
Feature activation-0.005
Top resid features:
,
Token,
Feature activation-0.002
Top resid features:
causing
Token causing
Feature activation-0.003
Top resid features:
her
Token her
Feature activation+4.490
Top resid features:
to
Token to
Feature activation+0.015
Top resid features:
stop
Token stop
Feature activation+0.017
Top resid features:
advancing
Token advancing
Feature activation+0.026
Top resid features:
through
Token through
Feature activation+0.003
Top resid features:
the
Token the
Feature activation+0.000
Top resid features:
by
Token by
Feature activation+0.002
Top resid features:
Officer
Token Officer
Feature activation+0.003
Top resid features:
Betty
Token Betty
Feature activation+1.951
Top resid features:
Shelby
Token Shelby
Feature activation+0.227
Top resid features:
as
Token as
Feature activation+0.009
Top resid features:
she
Token she
Feature activation+2.183
Top resid features:
responded
Token responded
Feature activation+0.078
Top resid features:
to
Token to
Feature activation+0.012
Top resid features:
reports
Token reports
Feature activation+0.008
Top resid features:
of
Token of
Feature activation-0.001
Top resid features:
an
Token an
Feature activation+0.000
Top resid features:
the
Token the
Feature activation+0.001
Top resid features:
shooting
Token shooting
Feature activation+0.001
Top resid features:
of
Token of
Feature activation-0.001
Top resid features:
Ty
Token Ty
Feature activation+0.001
Top resid features:
ke
Tokenke
Feature activation-0.125
Top resid features:
herself
Token herself
Feature activation+4.598
Top resid features:
(
Token (
Feature activation+0.004
Top resid features:
both
Tokenboth
Feature activation+0.001
Top resid features:
of
Token of
Feature activation+0.001
Top resid features:
which
Token which
Feature activation-0.000
Top resid features:
took
Token took
Feature activation+0.000
Top resid features:
Ċ
TokenĊ
Feature activation+0.004
Top resid features:
"
Token"
Feature activation+0.000
Top resid features:
Yes
TokenYes
Feature activation+0.002
Top resid features:
,
Token,
Feature activation+0.001
Top resid features:
Ge
Token Ge
Feature activation+0.010
Top resid features:
ena
Tokenena
Feature activation+2.994
Top resid features:
Davis
Token Davis
Feature activation+0.174
Top resid features:
is
Token is
Feature activation+0.034
Top resid features:
one
Token one
Feature activation+0.014
Top resid features:
of
Token of
Feature activation+0.004
Top resid features:
the
Token the
Feature activation+0.003
Top resid features:

Decoder Weights Distribution

Head 0: 0.03

Head 1: 0.48

Head 2: 0.04

Head 3: 0.08

Head 4: 0.02

Head 5: 0.11

Head 6: 0.08

Head 7: 0.02

Head 8: 0.03

Head 9: 0.04

Head 10: 0.05

Head 11: 0.02

Positive logits

herself5.54

heroine4.09

woman3.82

her3.70

woman3.65

she3.57

actress3.54

Marie3.51

Devi3.48

girl3.46

lady3.42

Actress3.42

She3.37

maid3.36

Louise3.34

blonde3.34

Woman3.33

SHE3.33

female3.30

princess3.29

Negative logits

enhagen-2.43

-1.99

allocations-1.92

atos-1.92

atus-1.90

querque-1.89

ipment-1.89

-1.89

agos-1.88

keys-1.86

shaft-1.85

MPEG-1.82

alion-1.82

resy-1.81

Tact-1.80

espie-1.80

ombat-1.80

routers-1.78

SDL-1.78

Tanks-1.77

INTERVAL 5.224 - 5.805
CONTAINS 0.000%

single
Token single
Feature activation+0.000
site
Token site
Feature activation+0.000
whereas
Token whereas
Feature activation+0.000
T
Token T
Feature activation+2.948
ure
Tokenure
Feature activation+5.065
tsky
Tokentsky
Feature activation+5.805
's
Token's
Feature activation+1.875
team
Token team
Feature activation+0.000
<|endoftext|>
Token<|endoftext|>
Feature activation+0.000
Looks
TokenLooks
Feature activation+0.000
like
Token like
Feature activation+0.000
3
Token3
Feature activation+0.000
]
Token]
Feature activation+0.412
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.158
G
TokenG
Feature activation+2.164
ellar
Tokenellar
Feature activation+5.382
was
Token was
Feature activation+2.429
given
Token given
Feature activation+0.000
a
Token a
Feature activation+0.000
partial
Token partial
Feature activation+0.000
scholarship
Token scholarship
Feature activation+0.000

INTERVAL 4.644 - 5.224
CONTAINS 0.001%

delivering
Token delivering
Feature activation+0.000
."
Token."
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
E
TokenE
Feature activation+0.013
arl
Tokenarl
Feature activation+5.130
would
Token would
Feature activation+2.180
say
Token say
Feature activation+0.978
later
Token later
Feature activation+0.000
that
Token that
Feature activation+0.672
her
Token her
Feature activation+0.549
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Judge
TokenJudge
Feature activation+3.363
N
Token N
Feature activation+3.298
en
Tokenen
Feature activation+3.670
ney
Tokenney
Feature activation+5.085
also
Token also
Feature activation+1.176
hears
Token hears
Feature activation+0.511
"
Token "
Feature activation+0.000
k
Tokenk
Feature activation+0.000
hal
Tokenhal
Feature activation+0.000
a
Token a
Feature activation+0.000
single
Token single
Feature activation+0.000
site
Token site
Feature activation+0.000
whereas
Token whereas
Feature activation+0.000
T
Token T
Feature activation+2.948
ure
Tokenure
Feature activation+5.065
tsky
Tokentsky
Feature activation+5.805
's
Token's
Feature activation+1.875
team
Token team
Feature activation+0.000
<|endoftext|>
Token<|endoftext|>
Feature activation+0.000
Looks
TokenLooks
Feature activation+0.000
a
Token a
Feature activation+0.000
bit
Token bit
Feature activation+0.000
longer
Token longer
Feature activation+0.000
.
Token.
Feature activation+0.000
Wood
Token Wood
Feature activation+4.452
ruff
Tokenruff
Feature activation+4.847
says
Token says
Feature activation+0.210
first
Token first
Feature activation+0.000
they
Token they
Feature activation+0.000
have
Token have
Feature activation+0.000
to
Token to
Feature activation+0.000
tweet
Token tweet
Feature activation+0.000
myself
Token myself
Feature activation+0.000
,
Token,
Feature activation+0.000
but
Token but
Feature activation+0.000
Ne
Token Ne
Feature activation+1.778
Ne
TokenNe
Feature activation+4.809
seems
Token seems
Feature activation+0.812
to
Token to
Feature activation+0.494
be
Token be
Feature activation+0.790
cool
Token cool
Feature activation+0.054
with
Token with
Feature activation+0.182

INTERVAL 4.063 - 4.644
CONTAINS 0.001%

profile
Token profile
Feature activation+0.000
."
Token."
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.811
Davis
TokenDavis
Feature activation+4.687
,
Token,
Feature activation+4.073
who
Token who
Feature activation+2.567
won
Token won
Feature activation+0.355
the
Token the
Feature activation+0.081
supporting
Token supporting
Feature activation+0.000
actress
Token actress
Feature activation+0.000
roph
Tokenroph
Feature activation+0.000
ilia
Tokenilia
Feature activation+0.000
specifically
Token specifically
Feature activation+0.000
,
Token,
Feature activation+0.000
John
Token John
Feature activation+1.991
sson
Tokensson
Feature activation+4.496
argued
Token argued
Feature activation+0.000
that
Token that
Feature activation+0.000
people
Token people
Feature activation+0.000
should
Token should
Feature activation+0.000
be
Token be
Feature activation+0.000
wait
Token wait
Feature activation+0.000
a
Token a
Feature activation+0.000
bit
Token bit
Feature activation+0.000
longer
Token longer
Feature activation+0.000
.
Token.
Feature activation+0.000
Wood
Token Wood
Feature activation+4.452
ruff
Tokenruff
Feature activation+4.847
says
Token says
Feature activation+0.210
first
Token first
Feature activation+0.000
they
Token they
Feature activation+0.000
have
Token have
Feature activation+0.000
."
Token."
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
S
TokenS
Feature activation+0.445
ark
Tokenark
Feature activation+3.389
is
Tokenis
Feature activation+4.321
ian
Tokenian
Feature activation+3.155
leaps
Token leaps
Feature activation+0.045
from
Token from
Feature activation+0.000
his
Token his
Feature activation+0.000
bed
Token bed
Feature activation+0.000
third
Token third
Feature activation+0.000
indictment
Token indictment
Feature activation+0.000
count
Token count
Feature activation+0.000
alleges
Token alleges
Feature activation+0.000
You
Token You
Feature activation+1.407
n
Tokenn
Feature activation+4.355
illegally
Token illegally
Feature activation+0.000
tam
Token tam
Feature activation+0.000
pered
Tokenpered
Feature activation+0.000
with
Token with
Feature activation+0.000
witness
Token witness
Feature activation+0.000

INTERVAL 3.483 - 4.063
CONTAINS 0.002%

they
Token they
Feature activation+0.000
function
Token function
Feature activation+0.000
.
Token.
Feature activation+0.096
Ms
Token Ms
Feature activation+0.000
.
Token.
Feature activation+0.502
Johann
Token Johann
Feature activation+3.589
sen
Tokensen
Feature activation+1.844
worked
Token worked
Feature activation+0.011
for
Token for
Feature activation+0.000
an
Token an
Feature activation+0.000
academic
Token academic
Feature activation+0.000
said
Token said
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Har
TokenHar
Feature activation+0.000
oon
Tokenoon
Feature activation+4.015
told
Token told
Feature activation+0.588
the
Token the
Feature activation+0.000
Austin
Token Austin
Feature activation+0.000
Monitor
Token Monitor
Feature activation+0.000
that
Token that
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
R
TokenR
Feature activation+0.552
if
Tokenif
Feature activation+1.083
qa
Tokenqa
Feature activation+3.880
âĢ
TokenâĢ
Feature activation+1.993
Ļ
TokenĻ
Feature activation+1.683
s
Tokens
Feature activation+1.449
guardian
Token guardian
Feature activation+0.000
ad
Token ad
Feature activation+0.000
,
Token,
Feature activation+0.000
but
Token but
Feature activation+0.534
no
Token no
Feature activation+0.000
.
Token.
Feature activation+0.326
Hel
Token Hel
Feature activation+1.865
a
Tokena
Feature activation+3.939
stro
Token stro
Feature activation+0.000
ve
Tokenve
Feature activation+0.000
to
Token to
Feature activation+0.227
take
Token take
Feature activation+0.673
over
Token over
Feature activation+0.066
she
Token she
Feature activation+0.227
said
Token said
Feature activation+0.145
Oliver
Token Oliver
Feature activation+1.586
-
Token-
Feature activation+0.746
Mach
TokenMach
Feature activation+1.936
ado
Tokenado
Feature activation+3.949
did
Token did
Feature activation+0.643
a
Token a
Feature activation+0.000
good
Token good
Feature activation+0.000
job
Token job
Feature activation+0.000
finishing
Token finishing
Feature activation+0.000

INTERVAL 2.902 - 3.483
CONTAINS 0.004%

.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
P
TokenP
Feature activation+1.209
el
Tokenel
Feature activation+0.533
osi
Tokenosi
Feature activation+2.961
made
Token made
Feature activation+0.297
headlines
Token headlines
Feature activation+0.000
on
Token on
Feature activation+0.000
Tuesday
Token Tuesday
Feature activation+0.000
when
Token when
Feature activation+0.000
walk
Token walk
Feature activation+0.000
.
Token.
Feature activation+0.064
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.336
M
TokenM
Feature activation+0.676
orrow
Tokenorrow
Feature activation+2.994
adopted
Token adopted
Feature activation+0.166
the
Token the
Feature activation+0.565
name
Token name
Feature activation+0.538
Amber
Token Amber
Feature activation+0.000
because
Token because
Feature activation+0.001
from
Token from
Feature activation+0.439
a
Token a
Feature activation+0.185
domestic
Token domestic
Feature activation+0.000
violence
Token violence
Feature activation+0.000
complaint
Token complaint
Feature activation+0.332
by
Token by
Feature activation+3.008
his
Token his
Feature activation+0.000
ex
Token ex
Feature activation+2.011
-
Token-
Feature activation+0.000
girlfriend
Tokengirlfriend
Feature activation+1.049
.
Token.
Feature activation+0.000
.
Token.
Feature activation+0.000
A
TokenA
Feature activation+0.000
magistrate
Token magistrate
Feature activation+0.000
judge
Token judge
Feature activation+0.000
set
Token set
Feature activation+0.000
H
Token H
Feature activation+3.232
aney
Tokenaney
Feature activation+1.960
's
Token's
Feature activation+0.000
bond
Token bond
Feature activation+0.000
at
Token at
Feature activation+0.000
$
Token $
Feature activation+0.000
Parker
Token Parker
Feature activation+0.000
.
Token.
Feature activation+0.112
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+1.460
Park
TokenPark
Feature activation+3.173
er
Tokener
Feature activation+2.999
will
Token will
Feature activation+1.015
star
Token star
Feature activation+0.000
as
Token as
Feature activation+0.583
brilliant
Token brilliant
Feature activation+0.000
aerospace
Token aerospace
Feature activation+0.000

INTERVAL 2.322 - 2.902
CONTAINS 0.007%

âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
Edwards
Token Edwards
Feature activation+2.602
will
Token will
Feature activation+0.396
receive
Token receive
Feature activation+0.000
$
Token $
Feature activation+0.000
120
Token120
Feature activation+0.000
,
Token,
Feature activation+0.000
said
Token said
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Fer
TokenFer
Feature activation+2.183
land
Tokenland
Feature activation+2.448
also
Token also
Feature activation+0.521
insisted
Token insisted
Feature activation+0.119
that
Token that
Feature activation+0.000
the
Token the
Feature activation+0.000
court
Token court
Feature activation+0.000
to
Token to
Feature activation+0.000
look
Token look
Feature activation+0.000
beyond
Token beyond
Feature activation+0.000
what
Token what
Feature activation+0.000
the
Token the
Feature activation+0.000
blogger
Token blogger
Feature activation+2.476
was
Token was
Feature activation+0.226
trying
Token trying
Feature activation+0.000
to
Token to
Feature activation+0.000
say
Token say
Feature activation+0.000
,
Token,
Feature activation+0.000
It
TokenIt
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.000
what
Token what
Feature activation+0.000
Ts
Token Ts
Feature activation+2.783
ed
Tokened
Feature activation+2.234
aka
Tokenaka
Feature activation+1.245
,
Token,
Feature activation+0.840
41
Token 41
Feature activation+0.172
,
Token,
Feature activation+1.186
."
Token."
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.811
Davis
TokenDavis
Feature activation+4.687
,
Token,
Feature activation+4.073
who
Token who
Feature activation+2.567
won
Token won
Feature activation+0.355
the
Token the
Feature activation+0.081
supporting
Token supporting
Feature activation+0.000
actress
Token actress
Feature activation+0.000
Academy
Token Academy
Feature activation+0.000

INTERVAL 1.741 - 2.322
CONTAINS 0.015%

Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.820
Mrs
TokenMrs
Feature activation+0.654
O
Token O
Feature activation+0.364
'
Token'
Feature activation+0.642
Neill
TokenNeill
Feature activation+1.992
also
Token also
Feature activation+0.798
said
Token said
Feature activation+0.650
a
Token a
Feature activation+0.000
Sinn
Token Sinn
Feature activation+0.000
F
Token F
Feature activation+0.000
)
Token)
Feature activation+0.000
remembered
Token remembered
Feature activation+0.172
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.365
Clar
TokenClar
Feature activation+1.962
ke
Tokenke
Feature activation+3.499
didn
Token didn
Feature activation+2.563
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
t
Tokent
Feature activation+0.583
Emily
Token Emily
Feature activation+0.000
St
Token St
Feature activation+0.000
.
Token.
Feature activation+0.000
John
Token John
Feature activation+0.003
Mand
Token Mand
Feature activation+0.059
el
Tokenel
Feature activation+1.895
handles
Token handles
Feature activation+0.000
it
Token it
Feature activation+0.000
with
Token with
Feature activation+0.000
a
Token a
Feature activation+0.000
light
Token light
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Investigators
TokenInvestigators
Feature activation+0.000
said
Token said
Feature activation+0.000
Le
Token Le
Feature activation+1.828
bl
Tokenbl
Feature activation+1.885
anc
Tokenanc
Feature activation+2.845
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.241
s
Tokens
Feature activation+0.666
of
Token of
Feature activation+0.000
equality
Token equality
Feature activation+0.000
to
Token to
Feature activation+0.000
slay
Token slay
Feature activation+0.000
,
Token,
Feature activation+0.012
Sch
Token Sch
Feature activation+2.280
alf
Tokenalf
Feature activation+0.000
ly
Tokenly
Feature activation+3.229
became
Token became
Feature activation+0.988
just
Token just
Feature activation+0.000
another
Token another
Feature activation+0.423

INTERVAL 1.161 - 1.741
CONTAINS 0.038%

.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Suddenly
TokenSuddenly
Feature activation+0.000
,
Token,
Feature activation+0.000
Weiss
Token Weiss
Feature activation+1.564
felt
Token felt
Feature activation+0.000
a
Token a
Feature activation+0.000
pair
Token pair
Feature activation+0.000
of
Token of
Feature activation+0.000
hands
Token hands
Feature activation+0.000
beauty
Token beauty
Feature activation+0.032
queen
Token queen
Feature activation+1.535
-
Token -
Feature activation+0.974
who
Token who
Feature activation+2.277
is
Token is
Feature activation+1.266
also
Token also
Feature activation+1.190
known
Token known
Feature activation+1.172
as
Token as
Feature activation+0.822
Day
Token Day
Feature activation+0.062
ana
Tokenana
Feature activation+0.041
Mend
Token Mend
Feature activation+0.000
Er
TokenEr
Feature activation+1.615
in
Tokenin
Feature activation+1.144
Cor
Token Cor
Feature activation+1.378
win
Tokenwin
Feature activation+1.011
,
Token,
Feature activation+0.736
who
Token who
Feature activation+1.242
has
Token has
Feature activation+0.313
been
Token been
Feature activation+0.000
missing
Token missing
Feature activation+0.000
from
Token from
Feature activation+0.000
the
Token the
Feature activation+0.000
Ċ
TokenĊ
Feature activation+1.285
11
Token11
Feature activation+0.000
The
Token The
Feature activation+0.013
court
Token court
Feature activation+0.000
heard
Token heard
Feature activation+0.000
L
Token L
Feature activation+1.650
avin
Tokenavin
Feature activation+1.105
ia
Tokenia
Feature activation+1.622
has
Token has
Feature activation+0.631
battled
Token battled
Feature activation+0.016
addiction
Token addiction
Feature activation+0.000
heart
Token heart
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
village
Token village
Feature activation+0.000
,"
Token,"
Feature activation+0.000
B
Token B
Feature activation+1.346
enez
Tokenenez
Feature activation+1.423
ra
Tokenra
Feature activation+1.677
says
Token says
Feature activation+0.000
as
Token as
Feature activation+0.000
she
Token she
Feature activation+0.131

INTERVAL 0.580 - 1.161
CONTAINS 0.130%

P
Token P
Feature activation+0.000
iring
Tokeniring
Feature activation+0.000
er
Tokener
Feature activation+0.000
confirmed
Token confirmed
Feature activation+0.000
the
Token the
Feature activation+0.000
woman
Token woman
Feature activation+0.582
was
Token was
Feature activation+0.235
taken
Token taken
Feature activation+0.000
the
Token the
Feature activation+0.000
the
Token the
Feature activation+0.000
hospital
Token hospital
Feature activation+0.000
a
Token a
Feature activation+0.000
medical
Token medical
Feature activation+0.000
problem
Token problem
Feature activation+0.000
that
Token that
Feature activation+0.000
even
Token even
Feature activation+0.000
a
Token a
Feature activation+1.091
perfect
Token perfect
Feature activation+0.000
health
Token health
Feature activation+0.000
-
Token-
Feature activation+0.000
care
Tokencare
Feature activation+0.000
system
Token system
Feature activation+0.000
raise
Token raise
Feature activation+0.000
money
Token money
Feature activation+0.000
to
Token to
Feature activation+0.000
help
Token help
Feature activation+0.000
keep
Token keep
Feature activation+0.000
Aunt
Token Aunt
Feature activation+0.836
May
Token May
Feature activation+2.221
from
Token from
Feature activation+0.075
being
Token being
Feature activation+1.164
ev
Token ev
Feature activation+0.000
icted
Tokenicted
Feature activation+0.000
to
Token to
Feature activation+0.000
study
Token study
Feature activation+0.000
,
Token,
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
Sen
Token Sen
Feature activation+0.892
arat
Tokenarat
Feature activation+1.717
na
Tokenna
Feature activation+1.016
says
Token says
Feature activation+0.030
.
Token.
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
irted
Tokenirted
Feature activation+0.000
with
Token with
Feature activation+0.000
the
Token the
Feature activation+0.000
idea
Token idea
Feature activation+0.000
of
Token of
Feature activation+0.000
Car
Token Car
Feature activation+1.149
ano
Tokenano
Feature activation+1.457
returning
Token returning
Feature activation+0.000
from
Token from
Feature activation+0.000
a
Token a
Feature activation+0.000
five
Token five
Feature activation+0.000

INTERVAL 0.000 - 0.580
CONTAINS 99.802%

,
Token,
Feature activation+0.000
as
Token as
Feature activation+0.000
well
Token well
Feature activation+0.000
as
Token as
Feature activation+0.000
energy
Token energy
Feature activation+0.000
and
Token and
Feature activation+0.000
real
Token real
Feature activation+0.000
estate
Token estate
Feature activation+0.000
clients
Token clients
Feature activation+0.000
,
Token,
Feature activation+0.000
explained
Token explained
Feature activation+0.000
restore
Token restore
Feature activation+0.000
Black
Token Black
Feature activation+0.000
farmers
Token farmers
Feature activation+0.000
to
Token to
Feature activation+0.000
the
Token the
Feature activation+0.000
land
Token land
Feature activation+0.000
they
Token they
Feature activation+0.000
lost
Token lost
Feature activation+0.000
from
Token from
Feature activation+0.000
discrimination
Token discrimination
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Here
TokenHere
Feature activation+0.000
in
Token in
Feature activation+0.000
Barcelona
Token Barcelona
Feature activation+0.000
,
Token,
Feature activation+0.000
you
Token you
Feature activation+0.000
can
Token can
Feature activation+0.000
take
Token take
Feature activation+0.000
your
Token your
Feature activation+0.000
bikes
Token bikes
Feature activation+0.000
on
Token on
Feature activation+0.000
chairman
Token chairman
Feature activation+0.000
Imran
Token Imran
Feature activation+0.000
Khan
Token Khan
Feature activation+0.000
over
Token over
Feature activation+0.000
his
Token his
Feature activation+0.000
allegations
Token allegations
Feature activation+0.000
that
Token that
Feature activation+0.000
he
Token he
Feature activation+0.000
was
Token was
Feature activation+0.000
offered
Token offered
Feature activation+0.000
Rs
Token Rs
Feature activation+0.000
and
Token and
Feature activation+0.218
offered
Token offered
Feature activation+0.242
to
Token to
Feature activation+0.000
not
Token not
Feature activation+0.000
only
Token only
Feature activation+0.172
feature
Token feature
Feature activation+0.206
my
Token my
Feature activation+0.000
music
Token music
Feature activation+0.000
but
Token but
Feature activation+0.000
to
Token to
Feature activation+0.216
also
Token also
Feature activation+0.000

BOTTOM ACTIVATIONS
MIN = 0.000

atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
Development
Token Development
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
<|endoftext|>
Token<|endoftext|>
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000

Top feature 8 in H7.1: (feature 9657)

TOP ACTIVATIONS
MAX = 2.284

.
Token.
Feature activation+0.099
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.280
Instead
TokenInstead
Feature activation+0.000
the
Token the
Feature activation+0.175
gang
Token gang
Feature activation+2.284
continued
Token continued
Feature activation+1.836
them
Token them
Feature activation+0.929
ay
Tokenay
Feature activation+0.000
hem
Tokenhem
Feature activation+0.000
unab
Token unab
Feature activation+0.000
that
Token that
Feature activation+0.597
Pakistan
Token Pakistan
Feature activation+0.000
is
Token is
Feature activation+0.005
helping
Token helping
Feature activation+1.383
terrorists
Token terrorists
Feature activation+1.983
to
Token to
Feature activation+2.206
enter
Token enter
Feature activation+0.135
our
Token our
Feature activation+0.000
territory
Token territory
Feature activation+0.000
.
Token.
Feature activation+0.178
We
Token We
Feature activation+0.000
Economist
Token Economist
Feature activation+0.000
,
Token,
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
the
Tokenthe
Feature activation+0.039
attacker
Token attacker
Feature activation+2.065
simply
Token simply
Feature activation+0.955
read
Token read
Feature activation+0.000
the
Token the
Feature activation+0.000
terms
Token terms
Feature activation+0.000
and
Token and
Feature activation+0.000
on
Token on
Feature activation+0.000
the
Token the
Feature activation+0.000
pig
Token pig
Feature activation+1.453
and
Token and
Feature activation+0.859
there
Token there
Feature activation+0.619
he
Token he
Feature activation+2.031
was
Token was
Feature activation+1.341
,
Token,
Feature activation+0.570
scr
Token scr
Feature activation+0.000
unch
Tokenunch
Feature activation+0.000
ing
Tokening
Feature activation+0.000
information
Token information
Feature activation+0.000
that
Token that
Feature activation+0.597
Pakistan
Token Pakistan
Feature activation+0.000
is
Token is
Feature activation+0.005
helping
Token helping
Feature activation+1.383
terrorists
Token terrorists
Feature activation+1.983
to
Token to
Feature activation+2.206
enter
Token enter
Feature activation+0.135
our
Token our
Feature activation+0.000
territory
Token territory
Feature activation+0.000
.
Token.
Feature activation+0.178
GM
TokenGM
Feature activation+0.000
)
Token)
Feature activation+0.000
crops
Token crops
Feature activation+0.000
,
Token,
Feature activation+0.000
GM
Token GM
Feature activation+0.000
microbes
Token microbes
Feature activation+1.906
have
Token have
Feature activation+1.328
been
Token been
Feature activation+0.861
released
Token released
Feature activation+0.182
for
Token for
Feature activation+0.000
at
Token at
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.280
Instead
TokenInstead
Feature activation+0.000
the
Token the
Feature activation+0.175
gang
Token gang
Feature activation+2.284
continued
Token continued
Feature activation+1.836
them
Token them
Feature activation+0.929
ay
Tokenay
Feature activation+0.000
hem
Tokenhem
Feature activation+0.000
unab
Token unab
Feature activation+0.000
ated
Tokenated
Feature activation+0.223
significant
Tokensignificant
Feature activation+0.176
and
Token and
Feature activation+0.037
ten
Token ten
Feature activation+0.478
acious
Tokenacious
Feature activation+1.530
âĢ
TokenâĢ
Feature activation+0.057
Ŀ
TokenĿ
Feature activation+1.794
May
Token May
Feature activation+0.108
21
Token 21
Feature activation+0.135
attack
Token attack
Feature activation+0.270
was
Token was
Feature activation+0.117
detected
Token detected
Feature activation+0.263
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
He
TokenHe
Feature activation+0.000
said
Token said
Feature activation+0.000
the
Token the
Feature activation+0.794
women
Token women
Feature activation+1.787
initially
Token initially
Feature activation+0.224
charged
Token charged
Feature activation+0.000
very
Token very
Feature activation+0.000
little
Token little
Feature activation+0.000
,
Token,
Feature activation+0.000
Led
Token Led
Feature activation+0.000
ward
Tokenward
Feature activation+0.000
Spencer
Token Spencer
Feature activation+0.019
,
Token,
Feature activation+0.000
accused
Token accused
Feature activation+0.138
of
Token of
Feature activation+1.787
burg
Token burg
Feature activation+0.000
lar
Tokenlar
Feature activation+0.014
izing
Tokenizing
Feature activation+0.000
Texas
Token Texas
Feature activation+0.000
State
Token State
Feature activation+0.000
.
Token.
Feature activation+0.017
In
Token In
Feature activation+0.000
fact
Token fact
Feature activation+0.000
,
Token,
Feature activation+0.377
such
Token such
Feature activation+0.222
worms
Token worms
Feature activation+1.761
do
Token do
Feature activation+1.046
exist
Token exist
Feature activation+0.000
in
Token in
Feature activation+0.000
the
Token the
Feature activation+0.000
deep
Token deep
Feature activation+0.000
of
Token of
Feature activation+0.000
defense
Token defense
Feature activation+0.000
against
Token against
Feature activation+0.662
this
Token this
Feature activation+1.473
and
Token and
Feature activation+0.000
other
Token other
Feature activation+1.725
threats
Token threats
Feature activation+0.000
,
Token,
Feature activation+0.000
and
Token and
Feature activation+0.000
we
Token we
Feature activation+0.000
encourage
Token encourage
Feature activation+0.000
asking
Token asking
Feature activation+0.000
its
Token its
Feature activation+0.000
operatives
Token operatives
Feature activation+0.000
to
Token to
Feature activation+0.511
carry
Token carry
Feature activation+0.659
out
Token out
Feature activation+1.679
lone
Token lone
Feature activation+0.000
wolf
Token wolf
Feature activation+0.441
attacks
Token attacks
Feature activation+0.207
in
Token in
Feature activation+0.000
India
Token India
Feature activation+0.000
in
Token in
Feature activation+0.000
jail
Token jail
Feature activation+0.000
for
Token for
Feature activation+0.000
what
Token what
Feature activation+0.025
he
Token he
Feature activation+0.362
did
Token did
Feature activation+1.679
.
Token.
Feature activation+0.000
Certainly
Token Certainly
Feature activation+0.000
we
Token we
Feature activation+0.000
have
Token have
Feature activation+0.000
to
Token to
Feature activation+0.000
All
TokenAll
Feature activation+0.000
these
Token these
Feature activation+0.619
other
Token other
Feature activation+0.000
guys
Token guys
Feature activation+1.576
that
Token that
Feature activation+1.056
tried
Token tried
Feature activation+1.678
the
Token the
Feature activation+0.050
approach
Token approach
Feature activation+0.311
of
Token of
Feature activation+0.415
trying
Token trying
Feature activation+0.468
to
Token to
Feature activation+0.614
of
Token of
Feature activation+0.000
N
Token N
Feature activation+0.000
abl
Tokenabl
Feature activation+0.000
us
Tokenus
Feature activation+0.000
.
Token.
Feature activation+0.380
They
Token They
Feature activation+1.657
pat
Token pat
Feature activation+0.108
rolled
Tokenrolled
Feature activation+0.000
in
Token in
Feature activation+0.000
the
Token the
Feature activation+0.000
streets
Token streets
Feature activation+0.000
all
Tokenall
Feature activation+0.000
Mer
Token Mer
Feature activation+0.000
rigan
Tokenrigan
Feature activation+0.000
,
Token,
Feature activation+0.000
this
Token this
Feature activation+0.260
group
Token group
Feature activation+1.624
is
Token is
Feature activation+0.968
behind
Token behind
Feature activation+0.816
around
Token around
Feature activation+0.158
nearly
Token nearly
Feature activation+0.000
16
Token 16
Feature activation+0.038
drinking
Token drinking
Feature activation+0.000
the
Token the
Feature activation+0.000
lot
Token lot
Feature activation+0.000
,
Token,
Feature activation+0.000
before
Token before
Feature activation+0.560
starting
Token starting
Feature activation+1.619
an
Token an
Feature activation+0.747
altercation
Token altercation
Feature activation+0.040
with
Token with
Feature activation+0.000
a
Token a
Feature activation+0.000
by
Token by
Feature activation+0.000
suspected
Token suspected
Feature activation+0.000
of
Token of
Feature activation+0.314
taking
Token taking
Feature activation+0.198
part
Token part
Feature activation+0.396
in
Token in
Feature activation+1.230
the
Token the
Feature activation+1.599
violence
Token violence
Feature activation+0.095
and
Token and
Feature activation+0.000
are
Token are
Feature activation+0.000
looking
Token looking
Feature activation+0.000
for
Token for
Feature activation+0.000
could
Token could
Feature activation+0.000
include
Token include
Feature activation+0.000
more
Token more
Feature activation+0.000
examples
Token examples
Feature activation+0.000
of
Token of
Feature activation+0.273
his
Token his
Feature activation+1.596
incredibly
Token incredibly
Feature activation+1.588
abusive
Token abusive
Feature activation+0.738
behavior
Token behavior
Feature activation+0.229
,
Token,
Feature activation+0.000
but
Token but
Feature activation+0.000

Top DFA by src position
MAX = 2.179

gang
Token gang
Feature activation+0.037
Top resid features:
of
Token of
Feature activation+0.370
Top resid features:
naked
Token naked
Feature activation+0.056
Top resid features:
youngsters
Token youngsters
Feature activation+0.065
Top resid features:
ramp
Token ramp
Feature activation+0.220
Top resid features:
aging
Tokenaging
Feature activation+1.137
Top resid features:
through
Token through
Feature activation+0.642
Top resid features:
Cow
Token Cow
Feature activation-0.000
Top resid features:
den
Tokenden
Feature activation-0.000
Top resid features:
be
Tokenbe
Feature activation-0.000
Top resid features:
ath
Tokenath
Feature activation+0.000
Top resid features:
it
Token it
Feature activation-0.000
Top resid features:
was
Token was
Feature activation-0.001
Top resid features:
helping
Token helping
Feature activation+0.001
Top resid features:
terrorists
Token terrorists
Feature activation+0.048
Top resid features:
to
Token to
Feature activation+0.063
Top resid features:
infiltrate
Token infiltrate
Feature activation+1.954
Top resid features:
into
Token into
Feature activation+0.346
Top resid features:
the
Token the
Feature activation+0.155
Top resid features:
country
Token country
Feature activation+0.008
Top resid features:
.
Token.
Feature activation+0.001
Top resid features:
Ċ
TokenĊ
Feature activation+0.000
Top resid features:
,
Token,
Feature activation-0.001
Top resid features:
which
Token which
Feature activation-0.001
Top resid features:
a
Token a
Feature activation-0.002
Top resid features:
malicious
Token malicious
Feature activation+0.178
Top resid features:
attacker
Token attacker
Feature activation+0.625
Top resid features:
was
Token was
Feature activation+0.978
Top resid features:
able
Token able
Feature activation+0.113
Top resid features:
to
Token to
Feature activation+0.280
Top resid features:
exploit
Token exploit
Feature activation+0.156
Top resid features:
.
Token.
Feature activation+0.010
Top resid features:
Ċ
TokenĊ
Feature activation+0.005
Top resid features:
land
Tokenland
Feature activation+0.000
Top resid features:
in
Token in
Feature activation-0.000
Top resid features:
Western
Token Western
Feature activation-0.000
Top resid features:
Australian
Token Australian
Feature activation-0.000
Top resid features:
,
Token,
Feature activation-0.000
Top resid features:
stealing
Token stealing
Feature activation+1.423
Top resid features:
three
Token three
Feature activation+0.020
Top resid features:
six
Token six
Feature activation-0.000
Top resid features:
-
Token-
Feature activation-0.000
Top resid features:
packs
Tokenpacks
Feature activation+0.014
Top resid features:
of
Token of
Feature activation+0.001
Top resid features:
it
Token it
Feature activation-0.001
Top resid features:
was
Token was
Feature activation-0.002
Top resid features:
helping
Token helping
Feature activation+0.000
Top resid features:
terrorists
Token terrorists
Feature activation+0.072
Top resid features:
to
Token to
Feature activation+0.181
Top resid features:
infiltrate
Token infiltrate
Feature activation+1.730
Top resid features:
into
Token into
Feature activation+0.244
Top resid features:
the
Token the
Feature activation+0.094
Top resid features:
country
Token country
Feature activation+0.008
Top resid features:
.
Token.
Feature activation+0.004
Top resid features:
Ċ
TokenĊ
Feature activation+0.001
Top resid features:
Ċ
TokenĊ
Feature activation-0.001
Top resid features:
GM
TokenGM
Feature activation-0.001
Top resid features:
Micro
Token Micro
Feature activation+0.002
Top resid features:
bes
Tokenbes
Feature activation+0.033
Top resid features:
Inv
Token Inv
Feature activation+0.020
Top resid features:
ade
Tokenade
Feature activation+2.179
Top resid features:
North
Token North
Feature activation-0.001
Top resid features:
America
Token America
Feature activation-0.001
Top resid features:
Ċ
TokenĊ
Feature activation+0.000
Top resid features:
Ċ
TokenĊ
Feature activation-0.001
Top resid features:
While
TokenWhile
Feature activation-0.007
Top resid features:
of
Token of
Feature activation+0.122
Top resid features:
naked
Token naked
Feature activation+0.025
Top resid features:
youngsters
Token youngsters
Feature activation+0.022
Top resid features:
ramp
Token ramp
Feature activation+0.286
Top resid features:
aging
Tokenaging
Feature activation+0.409
Top resid features:
through
Token through
Feature activation+0.935
Top resid features:
Cow
Token Cow
Feature activation-0.001
Top resid features:
den
Tokenden
Feature activation-0.000
Top resid features:
be
Tokenbe
Feature activation-0.000
Top resid features:
ath
Tokenath
Feature activation+0.001
Top resid features:
last
Token last
Feature activation+0.002
Top resid features:
ten
Tokenten
Feature activation-0.001
Top resid features:
acious
Tokenacious
Feature activation+0.016
Top resid features:
âĢ
TokenâĢ
Feature activation-0.003
Top resid features:
Ŀ
TokenĿ
Feature activation+0.005
Top resid features:
cyber
Token cyber
Feature activation+0.152
Top resid features:
attack
Tokenattack
Feature activation+1.580
Top resid features:
on
Token on
Feature activation+0.148
Top resid features:
its
Token its
Feature activation+0.065
Top resid features:
network
Token network
Feature activation+0.001
Top resid features:
last
Token last
Feature activation-0.003
Top resid features:
weekend
Token weekend
Feature activation+0.001
Top resid features:
of
Token of
Feature activation-0.002
Top resid features:
Melissa
Token Melissa
Feature activation-0.002
Top resid features:
and
Token and
Feature activation-0.002
Top resid features:
Vanessa
Token Vanessa
Feature activation-0.000
Top resid features:
âĢĵ
Token âĢĵ
Feature activation+0.010
Top resid features:
approached
Token approached
Feature activation+1.859
Top resid features:
public
Token public
Feature activation+0.004
Top resid features:
figures
Token figures
Feature activation+0.015
Top resid features:
promising
Token promising
Feature activation+0.053
Top resid features:
to
Token to
Feature activation+0.014
Top resid features:
help
Token help
Feature activation+0.016
Top resid features:
,
Token,
Feature activation-0.005
Top resid features:
accused
Token accused
Feature activation+0.001
Top resid features:
of
Token of
Feature activation+0.004
Top resid features:
burg
Token burg
Feature activation+0.146
Top resid features:
lar
Tokenlar
Feature activation+0.375
Top resid features:
izing
Tokenizing
Feature activation+1.670
Top resid features:
Texas
Token Texas
Feature activation+0.006
Top resid features:
State
Token State
Feature activation-0.001
Top resid features:
dorm
Token dorm
Feature activation-0.003
Top resid features:
(
Token (
Feature activation+0.014
Top resid features:
San
TokenSan
Feature activation+0.001
Top resid features:
in
Token in
Feature activation+0.208
Top resid features:
the
Token the
Feature activation+0.055
Top resid features:
dark
Token dark
Feature activation+0.014
Top resid features:
,
Token,
Feature activation+0.018
Top resid features:
settling
Token settling
Feature activation+0.361
Top resid features:
onto
Token onto
Feature activation+0.472
Top resid features:
dead
Token dead
Feature activation+0.013
Top resid features:
animals
Token animals
Feature activation+0.004
Top resid features:
and
Token and
Feature activation+0.011
Top resid features:
sending
Token sending
Feature activation+0.378
Top resid features:
out
Token out
Feature activation+0.022
Top resid features:
:
Token:
Feature activation-0.001
Top resid features:
Ċ
TokenĊ
Feature activation-0.005
Top resid features:
Ċ
TokenĊ
Feature activation-0.008
Top resid features:
R
TokenR
Feature activation-0.000
Top resid features:
ansom
Tokenansom
Feature activation+0.083
Top resid features:
ware
Tokenware
Feature activation+0.782
Top resid features:
is
Token is
Feature activation+0.061
Top resid features:
a
Token a
Feature activation+0.031
Top resid features:
type
Token type
Feature activation+0.007
Top resid features:
of
Token of
Feature activation+0.018
Top resid features:
malicious
Token malicious
Feature activation+0.318
Top resid features:
in
Token in
Feature activation-0.005
Top resid features:
Kerala
Token Kerala
Feature activation+0.001
Top resid features:
to
Token to
Feature activation+0.007
Top resid features:
carry
Token carry
Feature activation-0.004
Top resid features:
out
Token out
Feature activation-0.001
Top resid features:
attacks
Token attacks
Feature activation+1.725
Top resid features:
on
Token on
Feature activation+0.270
Top resid features:
Indian
Token Indian
Feature activation-0.001
Top resid features:
festivals
Token festivals
Feature activation-0.006
Top resid features:
like
Token like
Feature activation+0.021
Top resid features:
K
Token K
Feature activation-0.000
Top resid features:
he
Token he
Feature activation+0.000
Top resid features:
launched
Token launched
Feature activation+0.060
Top resid features:
his
Token his
Feature activation+0.006
Top resid features:
bot
Token bot
Feature activation+0.113
Top resid features:
net
Tokennet
Feature activation+0.387
Top resid features:
attack
Token attack
Feature activation+1.043
Top resid features:
(
Token (
Feature activation+0.022
Top resid features:
which
Tokenwhich
Feature activation+0.006
Top resid features:
did
Token did
Feature activation+0.002
Top resid features:
.
Token .
Feature activation-0.006
Top resid features:
00000000
Token00000000
Feature activation-0.002
Top resid features:
of
Token of
Feature activation+0.007
Top resid features:
attempts
Token attempts
Feature activation+0.155
Top resid features:
from
Token from
Feature activation+0.207
Top resid features:
hackers
Token hackers
Feature activation+0.374
Top resid features:
to
Token to
Feature activation+0.401
Top resid features:
crack
Token crack
Feature activation+0.802
Top resid features:
the
Token the
Feature activation+0.072
Top resid features:
ecosystem
Token ecosystem
Feature activation-0.005
Top resid features:
.
Token.
Feature activation-0.001
Top resid features:
Ċ
TokenĊ
Feature activation-0.000
Top resid features:
Ċ
TokenĊ
Feature activation-0.001
Top resid features:
Strip
Token Strip
Feature activation-0.001
Top resid features:
,
Token,
Feature activation-0.003
Top resid features:
during
Token during
Feature activation-0.003
Top resid features:
which
Token which
Feature activation+0.000
Top resid features:
they
Token they
Feature activation+0.017
Top resid features:
abducted
Token abducted
Feature activation+0.932
Top resid features:
19
Token 19
Feature activation+0.005
Top resid features:
Palestinian
Token Palestinian
Feature activation-0.000
Top resid features:
civilians
Token civilians
Feature activation+0.001
Top resid features:
,
Token,
Feature activation+0.002
Top resid features:
including
Token including
Feature activation+0.001
Top resid features:
involvement
Token involvement
Feature activation+0.004
Top resid features:
of
Token of
Feature activation+0.003
Top resid features:
a
Token a
Feature activation+0.015
Top resid features:
professional
Token professional
Feature activation+0.001
Top resid features:
ransomware
Token ransomware
Feature activation+0.244
Top resid features:
group
Token group
Feature activation+0.796
Top resid features:
known
Token known
Feature activation+0.326
Top resid features:
as
Token as
Feature activation+0.247
Top resid features:
Kraken
Token Kraken
Feature activation+0.102
Top resid features:
.
Token.
Feature activation+0.163
Top resid features:
Ċ
TokenĊ
Feature activation+0.005
Top resid features:
pig
Token pig
Feature activation-0.000
Top resid features:
has
Token has
Feature activation-0.001
Top resid features:
gone
Token gone
Feature activation-0.001
Top resid features:
on
Token on
Feature activation+0.001
Top resid features:
the
Token the
Feature activation+0.009
Top resid features:
rampage
Token rampage
Feature activation+0.785
Top resid features:
at
Token at
Feature activation+0.278
Top resid features:
the
Token the
Feature activation+0.007
Top resid features:
De
Token De
Feature activation-0.001
Top resid features:
Grey
TokenGrey
Feature activation+0.001
Top resid features:
River
Token River
Feature activation+0.002
Top resid features:
house
Token house
Feature activation+0.003
Top resid features:
and
Token and
Feature activation+0.010
Top resid features:
went
Token went
Feature activation+0.092
Top resid features:
in
Token in
Feature activation+0.012
Top resid features:
and
Token and
Feature activation+0.005
Top resid features:
dragged
Token dragged
Feature activation+0.706
Top resid features:
me
Token me
Feature activation-0.000
Top resid features:
out
Token out
Feature activation+0.001
Top resid features:
,
Token,
Feature activation+0.022
Top resid features:
threw
Token threw
Feature activation+0.338
Top resid features:
me
Token me
Feature activation-0.002
Top resid features:
January
Token January
Feature activation+0.000
Top resid features:
2014
Token 2014
Feature activation+0.000
Top resid features:
,
Token,
Feature activation+0.000
Top resid features:
he
Token he
Feature activation+0.002
Top resid features:
has
Token has
Feature activation+0.003
Top resid features:
harassed
Token harassed
Feature activation+1.295
Top resid features:
me
Token me
Feature activation+0.002
Top resid features:
,
Token,
Feature activation+0.007
Top resid features:
picked
Token picked
Feature activation+0.233
Top resid features:
fights
Token fights
Feature activation+0.039
Top resid features:
with
Token with
Feature activation+0.017
Top resid features:

Decoder Weights Distribution

Head 0: 0.03

Head 1: 0.48

Head 2: 0.02

Head 3: 0.07

Head 4: 0.02

Head 5: 0.08

Head 6: 0.11

Head 7: 0.03

Head 8: 0.06

Head 9: 0.05

Head 10: 0.03

Head 11: 0.02

Positive logits

extortion3.59

thugs3.56

malicious3.42

looting3.40

infiltr3.38

menacing3.31

harassing3.29

hij3.28

intimidation3.26

mischief3.26

menace3.26

attacks3.24

stalking3.18

tactics3.17

brazen3.16

rampage3.16

hijacked3.16

harass3.13

assaults3.08

thieves3.08

Negative logits

Remastered-2.41

nutrit-2.41

grades-2.33

clarity-2.21

condol-2.18

parity-2.17

graduation-2.16

ometer-2.15

childbirth-2.13

maternity-2.09

+/--2.08

milestones-2.03

consolation-2.02

iatrics-2.01

NPR-1.96

ournals-1.94

vowel-1.94

grain-1.91

deferred-1.90

Essential-1.89

INTERVAL 2.056 - 2.284
CONTAINS 0.000%

.
Token.
Feature activation+0.099
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.280
Instead
TokenInstead
Feature activation+0.000
the
Token the
Feature activation+0.175
gang
Token gang
Feature activation+2.284
continued
Token continued
Feature activation+1.836
them
Token them
Feature activation+0.929
ay
Tokenay
Feature activation+0.000
hem
Tokenhem
Feature activation+0.000
unab
Token unab
Feature activation+0.000
that
Token that
Feature activation+0.597
Pakistan
Token Pakistan
Feature activation+0.000
is
Token is
Feature activation+0.005
helping
Token helping
Feature activation+1.383
terrorists
Token terrorists
Feature activation+1.983
to
Token to
Feature activation+2.206
enter
Token enter
Feature activation+0.135
our
Token our
Feature activation+0.000
territory
Token territory
Feature activation+0.000
.
Token.
Feature activation+0.178
We
Token We
Feature activation+0.000
Economist
Token Economist
Feature activation+0.000
,
Token,
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
the
Tokenthe
Feature activation+0.039
attacker
Token attacker
Feature activation+2.065
simply
Token simply
Feature activation+0.955
read
Token read
Feature activation+0.000
the
Token the
Feature activation+0.000
terms
Token terms
Feature activation+0.000
and
Token and
Feature activation+0.000

INTERVAL 1.827 - 2.056
CONTAINS 0.001%

on
Token on
Feature activation+0.000
the
Token the
Feature activation+0.000
pig
Token pig
Feature activation+1.453
and
Token and
Feature activation+0.859
there
Token there
Feature activation+0.619
he
Token he
Feature activation+2.031
was
Token was
Feature activation+1.341
,
Token,
Feature activation+0.570
scr
Token scr
Feature activation+0.000
unch
Tokenunch
Feature activation+0.000
ing
Tokening
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.280
Instead
TokenInstead
Feature activation+0.000
the
Token the
Feature activation+0.175
gang
Token gang
Feature activation+2.284
continued
Token continued
Feature activation+1.836
them
Token them
Feature activation+0.929
ay
Tokenay
Feature activation+0.000
hem
Tokenhem
Feature activation+0.000
unab
Token unab
Feature activation+0.000
ated
Tokenated
Feature activation+0.223
information
Token information
Feature activation+0.000
that
Token that
Feature activation+0.597
Pakistan
Token Pakistan
Feature activation+0.000
is
Token is
Feature activation+0.005
helping
Token helping
Feature activation+1.383
terrorists
Token terrorists
Feature activation+1.983
to
Token to
Feature activation+2.206
enter
Token enter
Feature activation+0.135
our
Token our
Feature activation+0.000
territory
Token territory
Feature activation+0.000
.
Token.
Feature activation+0.178
GM
TokenGM
Feature activation+0.000
)
Token)
Feature activation+0.000
crops
Token crops
Feature activation+0.000
,
Token,
Feature activation+0.000
GM
Token GM
Feature activation+0.000
microbes
Token microbes
Feature activation+1.906
have
Token have
Feature activation+1.328
been
Token been
Feature activation+0.861
released
Token released
Feature activation+0.182
for
Token for
Feature activation+0.000
at
Token at
Feature activation+0.000

INTERVAL 1.599 - 1.827
CONTAINS 0.001%

all
Tokenall
Feature activation+0.000
Mer
Token Mer
Feature activation+0.000
rigan
Tokenrigan
Feature activation+0.000
,
Token,
Feature activation+0.000
this
Token this
Feature activation+0.260
group
Token group
Feature activation+1.624
is
Token is
Feature activation+0.968
behind
Token behind
Feature activation+0.816
around
Token around
Feature activation+0.158
nearly
Token nearly
Feature activation+0.000
16
Token 16
Feature activation+0.038
Led
Token Led
Feature activation+0.000
ward
Tokenward
Feature activation+0.000
Spencer
Token Spencer
Feature activation+0.019
,
Token,
Feature activation+0.000
accused
Token accused
Feature activation+0.138
of
Token of
Feature activation+1.787
burg
Token burg
Feature activation+0.000
lar
Tokenlar
Feature activation+0.014
izing
Tokenizing
Feature activation+0.000
Texas
Token Texas
Feature activation+0.000
State
Token State
Feature activation+0.000
asking
Token asking
Feature activation+0.000
its
Token its
Feature activation+0.000
operatives
Token operatives
Feature activation+0.000
to
Token to
Feature activation+0.511
carry
Token carry
Feature activation+0.659
out
Token out
Feature activation+1.679
lone
Token lone
Feature activation+0.000
wolf
Token wolf
Feature activation+0.441
attacks
Token attacks
Feature activation+0.207
in
Token in
Feature activation+0.000
India
Token India
Feature activation+0.000
of
Token of
Feature activation+0.000
defense
Token defense
Feature activation+0.000
against
Token against
Feature activation+0.662
this
Token this
Feature activation+1.473
and
Token and
Feature activation+0.000
other
Token other
Feature activation+1.725
threats
Token threats
Feature activation+0.000
,
Token,
Feature activation+0.000
and
Token and
Feature activation+0.000
we
Token we
Feature activation+0.000
encourage
Token encourage
Feature activation+0.000
All
TokenAll
Feature activation+0.000
these
Token these
Feature activation+0.619
other
Token other
Feature activation+0.000
guys
Token guys
Feature activation+1.576
that
Token that
Feature activation+1.056
tried
Token tried
Feature activation+1.678
the
Token the
Feature activation+0.050
approach
Token approach
Feature activation+0.311
of
Token of
Feature activation+0.415
trying
Token trying
Feature activation+0.468
to
Token to
Feature activation+0.614

INTERVAL 1.370 - 1.599
CONTAINS 0.002%

.
Token.
Feature activation+0.000
In
Token In
Feature activation+0.000
the
Token the
Feature activation+0.000
letter
Token letter
Feature activation+0.000
the
Token the
Feature activation+0.000
person
Token person
Feature activation+1.586
called
Token called
Feature activation+0.216
him
Token him
Feature activation+0.000
out
Token out
Feature activation+0.000
for
Token for
Feature activation+0.269
participating
Token participating
Feature activation+0.000
have
Token have
Feature activation+0.000
information
Token information
Feature activation+0.000
that
Token that
Feature activation+0.597
Pakistan
Token Pakistan
Feature activation+0.000
is
Token is
Feature activation+0.005
helping
Token helping
Feature activation+1.383
terrorists
Token terrorists
Feature activation+1.983
to
Token to
Feature activation+2.206
enter
Token enter
Feature activation+0.135
our
Token our
Feature activation+0.000
territory
Token territory
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
Hartford
Token Hartford
Feature activation+0.000
Soldiers
Token Soldiers
Feature activation+1.412
were
Token were
Feature activation+0.728
blamed
Token blamed
Feature activation+0.000
for
Token for
Feature activation+0.082
the
Token the
Feature activation+0.000
vandalism
Token vandalism
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.247
significant
Tokensignificant
Feature activation+0.176
and
Token and
Feature activation+0.037
ten
Token ten
Feature activation+0.478
acious
Tokenacious
Feature activation+1.530
âĢ
TokenâĢ
Feature activation+0.057
Ŀ
TokenĿ
Feature activation+1.794
May
Token May
Feature activation+0.108
21
Token 21
Feature activation+0.135
attack
Token attack
Feature activation+0.270
suspected
Token suspected
Feature activation+0.000
of
Token of
Feature activation+0.314
taking
Token taking
Feature activation+0.198
part
Token part
Feature activation+0.396
in
Token in
Feature activation+1.230
the
Token the
Feature activation+1.599
violence
Token violence
Feature activation+0.095
and
Token and
Feature activation+0.000
are
Token are
Feature activation+0.000
looking
Token looking
Feature activation+0.000
for
Token for
Feature activation+0.000

INTERVAL 1.142 - 1.370
CONTAINS 0.006%

what
Token what
Feature activation+0.000
G
Token G
Feature activation+0.000
j
Tokenj
Feature activation+0.000
oni
Tokenoni
Feature activation+0.000
had
Token had
Feature activation+1.029
done
Token done
Feature activation+1.260
,
Token,
Feature activation+0.000
including
Token including
Feature activation+0.000
her
Token her
Feature activation+0.018
allegation
Token allegation
Feature activation+0.000
that
Token that
Feature activation+0.000
(
Token (
Feature activation+0.000
K
TokenK
Feature activation+0.000
X
TokenX
Feature activation+0.000
AN
TokenAN
Feature activation+0.000
)
Token)
Feature activation+0.000
--
Token --
Feature activation+1.305
Early
Token Early
Feature activation+0.000
Sunday
Token Sunday
Feature activation+0.000
morning
Token morning
Feature activation+0.000
,
Token,
Feature activation+0.123
police
Token police
Feature activation+0.000
.
Token.
Feature activation+0.000
According
Token According
Feature activation+0.000
to
Token to
Feature activation+0.000
Torres
Token Torres
Feature activation+0.000
,
Token,
Feature activation+0.000
Phillips
Token Phillips
Feature activation+1.257
then
Token then
Feature activation+0.409
shot
Token shot
Feature activation+0.000
and
Token and
Feature activation+0.000
killed
Token killed
Feature activation+0.000
himself
Token himself
Feature activation+0.000
list
Token list
Feature activation+0.000
.
Token.
Feature activation+0.078
There
Token There
Feature activation+0.000
is
Token is
Feature activation+0.039
speculation
Token speculation
Feature activation+0.225
that
Token that
Feature activation+1.186
many
Token many
Feature activation+0.139
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
migrants
Token migrants
Feature activation+0.000
are
Token are
Feature activation+0.000
shooting
Token shooting
Feature activation+0.000
,
Token,
Feature activation+0.000
but
Token but
Feature activation+0.000
say
Token say
Feature activation+0.000
Fre
Token Fre
Feature activation+0.025
in
Tokenin
Feature activation+1.277
has
Token has
Feature activation+0.000
a
Token a
Feature activation+0.000
long
Token long
Feature activation+0.000
-
Token-
Feature activation+0.000
standing
Tokenstanding
Feature activation+0.000

INTERVAL 0.914 - 1.142
CONTAINS 0.011%

in
Token in
Feature activation+0.016
mud
Token mud
Feature activation+0.000
,
Token,
Feature activation+0.000
allowing
Token allowing
Feature activation+0.022
Cunningham
Token Cunningham
Feature activation+1.223
to
Token to
Feature activation+1.108
get
Token get
Feature activation+0.130
into
Token into
Feature activation+0.341
the
Token the
Feature activation+0.230
cruiser
Token cruiser
Feature activation+0.000
and
Token and
Feature activation+0.000
defense
Token defense
Feature activation+0.000
industry
Token industry
Feature activation+0.000
suggest
Token suggest
Feature activation+0.000
the
Token the
Feature activation+0.057
group
Token group
Feature activation+1.142
is
Token is
Feature activation+0.962
working
Token working
Feature activation+0.912
for
Token for
Feature activation+0.000
agencies
Token agencies
Feature activation+0.000
of
Token of
Feature activation+0.000
a
Token a
Feature activation+0.000
Taiwan
Token Taiwan
Feature activation+0.000
;
Token;
Feature activation+0.000
recently
Token recently
Feature activation+0.000
the
Token the
Feature activation+0.000
group
Token group
Feature activation+1.297
has
Token has
Feature activation+0.998
focused
Token focused
Feature activation+0.588
on
Token on
Feature activation+0.000
targets
Token targets
Feature activation+0.000
in
Token in
Feature activation+0.000
South
Token South
Feature activation+0.000
on
Token on
Feature activation+0.000
the
Token the
Feature activation+0.000
next
Token next
Feature activation+0.000
day
Token day
Feature activation+0.000
.
Token.
Feature activation+0.010
He
Token He
Feature activation+0.985
had
Token had
Feature activation+0.345
spent
Token spent
Feature activation+0.113
28
Token 28
Feature activation+0.000
days
Token days
Feature activation+0.247
at
Token at
Feature activation+0.008
reused
Token reused
Feature activation+0.000
by
Token by
Feature activation+0.000
subsequent
Token subsequent
Feature activation+0.000
generations
Token generations
Feature activation+0.000
of
Token of
Feature activation+0.319
was
Token was
Feature activation+0.926
ps
Tokenps
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
How
TokenHow
Feature activation+0.000

INTERVAL 0.685 - 0.914
CONTAINS 0.026%

att
Tokenatt
Feature activation+0.000
ucks
Tokenucks
Feature activation+0.000
maintain
Token maintain
Feature activation+0.000
Drug
Token Drug
Feature activation+0.158
Task
Token Task
Feature activation+0.442
Force
Token Force
Feature activation+0.882
members
Token members
Feature activation+1.466
pointed
Token pointed
Feature activation+0.315
their
Token their
Feature activation+0.038
guns
Token guns
Feature activation+0.000
at
Token at
Feature activation+0.000
swift
Token swift
Feature activation+0.000
response
Token response
Feature activation+0.000
prevented
Token prevented
Feature activation+0.000
an
Token an
Feature activation+0.000
even
Token even
Feature activation+0.000
worse
Token worse
Feature activation+0.764
attack
Token attack
Feature activation+0.054
.
Token.
Feature activation+0.000
And
Token And
Feature activation+0.000
I
Token I
Feature activation+0.000
extend
Token extend
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.000
legal
Token legal
Feature activation+0.000
definition
Token definition
Feature activation+0.000
of
Token of
Feature activation+0.563
a
Token a
Feature activation+0.772
pyramid
Token pyramid
Feature activation+0.535
scheme
Token scheme
Feature activation+0.197
fits
Token fits
Feature activation+0.000
with
Token with
Feature activation+0.000
most
Token most
Feature activation+0.000
exists
Token exists
Feature activation+0.000
to
Token to
Feature activation+0.000
serve
Token serve
Feature activation+0.000
her
Token her
Feature activation+0.214
.
Token.
Feature activation+0.000
She
Token She
Feature activation+0.811
uses
Token uses
Feature activation+0.856
ins
Token ins
Feature activation+0.025
ince
Tokenince
Feature activation+0.000
re
Tokenre
Feature activation+0.000
,
Token,
Feature activation+0.029
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.062
Den
TokenDen
Feature activation+0.952
ne
Tokenne
Feature activation+0.586
hy
Tokenhy
Feature activation+0.819
's
Token's
Feature activation+0.717
wife
Token wife
Feature activation+0.000
,
Token,
Feature activation+0.000
Tam
Token Tam
Feature activation+0.000
ara
Tokenara
Feature activation+0.000
Den
Token Den
Feature activation+0.000

INTERVAL 0.457 - 0.685
CONTAINS 0.064%

out
Token out
Feature activation+0.000
to
Token to
Feature activation+0.000
tackle
Token tackle
Feature activation+0.623
the
Token the
Feature activation+1.158
blaze
Token blaze
Feature activation+0.049
.
Token.
Feature activation+0.490
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.303
The
TokenThe
Feature activation+0.431
incident
Token incident
Feature activation+0.313
has
Token has
Feature activation+0.000
dictator
Token dictator
Feature activation+0.118
.
Token.
Feature activation+0.196
Gear
Token Gear
Feature activation+0.000
y
Tokeny
Feature activation+0.760
resists
Token resists
Feature activation+0.000
the
Token the
Feature activation+0.518
temptation
Token temptation
Feature activation+0.028
offered
Token offered
Feature activation+0.000
by
Token by
Feature activation+0.390
this
Token this
Feature activation+0.766
faction
Token faction
Feature activation+1.297
charges
Token charges
Feature activation+0.000
pred
Token pred
Feature activation+0.000
icated
Tokenicated
Feature activation+0.000
upon
Token upon
Feature activation+0.000
a
Token a
Feature activation+0.000
pet
Token pet
Feature activation+0.541
it
Tokenit
Feature activation+0.861
l
Token l
Feature activation+0.213
arc
Tokenarc
Feature activation+0.000
eny
Tokeneny
Feature activation+0.000
allegation
Token allegation
Feature activation+0.000
and
Token and
Feature activation+0.000
upper
Token upper
Feature activation+0.000
arm
Token arm
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
and
Token and
Feature activation+0.605
infl
Token infl
Feature activation+0.000
amed
Tokenamed
Feature activation+0.000
a
Token a
Feature activation+0.000
pre
Token pre
Feature activation+0.000
-
Token-
Feature activation+0.000
October
Token October
Feature activation+0.000
2011
Token 2011
Feature activation+0.000
,
Token,
Feature activation+0.000
Br
Token Br
Feature activation+0.442
isk
Tokenisk
Feature activation+0.315
deals
Tokendeals
Feature activation+0.569
had
Token had
Feature activation+0.000
sold
Token sold
Feature activation+0.000
through
Token through
Feature activation+0.000
its
Token its
Feature activation+0.165
website
Token website
Feature activation+0.000

INTERVAL 0.228 - 0.457
CONTAINS 0.203%

itudes
Tokenitudes
Feature activation+0.000
.
Token.
Feature activation+0.006
But
Token But
Feature activation+0.000
the
Token the
Feature activation+0.000
jam
Token jam
Feature activation+0.179
ming
Tokenming
Feature activation+0.276
will
Token will
Feature activation+0.288
only
Token only
Feature activation+0.163
affect
Token affect
Feature activation+0.000
aircraft
Token aircraft
Feature activation+0.000
above
Token above
Feature activation+0.000
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.213
researchers
Token researchers
Feature activation+0.000
found
Token found
Feature activation+0.649
that
Token that
Feature activation+0.355
the
Token the
Feature activation+0.423
Meter
Token Meter
Feature activation+0.382
pre
Tokenpre
Feature activation+0.004
ter
Tokenter
Feature activation+0.064
code
Token code
Feature activation+0.000
was
Token was
Feature activation+0.063
prior
Token prior
Feature activation+0.058
to
Token to
Feature activation+0.103
a
Token a
Feature activation+0.135
"
Token "
Feature activation+0.208
very
Tokenvery
Feature activation+0.000
strong
Token strong
Feature activation+0.435
sound
Token sound
Feature activation+0.000
"
Token"
Feature activation+0.029
that
Token that
Feature activation+0.000
she
Token she
Feature activation+0.000
took
Token took
Feature activation+0.000
has
Token has
Feature activation+0.066
demonstrated
Token demonstrated
Feature activation+0.287
exactly
Token exactly
Feature activation+0.070
the
Token the
Feature activation+0.373
same
Token same
Feature activation+0.288
by
Token by
Feature activation+0.368
developing
Token developing
Feature activation+0.076
a
Token a
Feature activation+0.497
new
Token new
Feature activation+0.076
hacking
Token hacking
Feature activation+0.058
technique
Token technique
Feature activation+0.096
.
Token.
Feature activation+0.000
Because
Token Because
Feature activation+0.000
those
Token those
Feature activation+0.000
western
Token western
Feature activation+0.182
males
Token males
Feature activation+0.104
losers
Token losers
Feature activation+0.414
are
Token are
Feature activation+0.020
naturally
Token naturally
Feature activation+0.000
racist
Token racist
Feature activation+0.000
,
Token,
Feature activation+0.000
so
Token so
Feature activation+0.000

INTERVAL 0.000 - 0.228
CONTAINS 99.687%

represents
Token represents
Feature activation+0.000
the
Token the
Feature activation+0.000
first
Token first
Feature activation+0.000
time
Token time
Feature activation+0.000
the
Token the
Feature activation+0.000
Argent
Token Argent
Feature activation+0.000
inian
Tokeninian
Feature activation+0.000
pont
Token pont
Feature activation+0.000
iff
Tokeniff
Feature activation+0.000
has
Token has
Feature activation+0.000
met
Token met
Feature activation+0.000
og
Tokenog
Feature activation+0.000
re
Tokenre
Feature activation+0.000
sleeps
Token sleeps
Feature activation+0.000
below
Token below
Feature activation+0.000
.
Token.
Feature activation+0.000
Max
Token Max
Feature activation+0.000
ie
Tokenie
Feature activation+0.000
and
Token and
Feature activation+0.000
other
Token other
Feature activation+0.000
Max
Token Max
Feature activation+0.000
ie
Tokenie
Feature activation+0.000
,
Token,
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
Kal
Token Kal
Feature activation+0.000
an
Tokenan
Feature activation+0.000
ick
Tokenick
Feature activation+0.000
exclaim
Token exclaim
Feature activation+0.000
s
Tokens
Feature activation+0.000
,
Token,
Feature activation+0.000
after
Token after
Feature activation+0.000
driver
Token driver
Feature activation+0.000
national
Token national
Feature activation+0.000
influences
Token influences
Feature activation+0.000
so
Token so
Feature activation+0.000
much
Token much
Feature activation+0.000
that
Token that
Feature activation+0.000
they
Token they
Feature activation+0.074
went
Token went
Feature activation+0.000
a
Token a
Feature activation+0.000
step
Token step
Feature activation+0.000
beyond
Token beyond
Feature activation+0.000
international
Token international
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Four
TokenFour
Feature activation+0.000
months
Token months
Feature activation+0.000
later
Token later
Feature activation+0.000
,
Token,
Feature activation+0.000
Dor
Token Dor
Feature activation+0.000
val
Tokenval
Feature activation+0.000
received
Token received
Feature activation+0.000
a
Token a
Feature activation+0.000
notice
Token notice
Feature activation+0.000
that
Token that
Feature activation+0.000

BOTTOM ACTIVATIONS
MIN = 0.000

atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
Development
Token Development
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
<|endoftext|>
Token<|endoftext|>
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000

Top feature 9 in H7.1: (feature 41689)

TOP ACTIVATIONS
MAX = 1.825

report
Token report
Feature activation+0.000
illustrates
Token illustrates
Feature activation+0.000
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.000
implementation
Token implementation
Feature activation+0.594
of
Token of
Feature activation+1.575
the
Token the
Feature activation+1.216
directive
Token directive
Feature activation+0.544
led
Token led
Feature activation+0.000
to
Token to
Feature activation+0.000
the
Token the
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Democrats
TokenDemocrats
Feature activation+0.000
commonly
Token commonly
Feature activation+0.000
prefer
Token prefer
Feature activation+1.410
a
Token a
Feature activation+1.561
statistical
Token statistical
Feature activation+0.000
sampling
Token sampling
Feature activation+0.000
method
Token method
Feature activation+0.000
meant
Token meant
Feature activation+0.000
to
Token to
Feature activation+0.000
ever
Token ever
Feature activation+0.000
think
Token think
Feature activation+0.000
that
Token that
Feature activation+0.000
this
Token this
Feature activation+0.004
wonderful
Token wonderful
Feature activation+0.218
new
Token new
Feature activation+1.556
program
Token program
Feature activation+0.577
sounded
Token sounded
Feature activation+0.000
like
Token like
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
,
Token,
Feature activation+0.000
who
Token who
Feature activation+0.000
has
Token has
Feature activation+0.000
pushed
Token pushed
Feature activation+0.000
to
Token to
Feature activation+0.000
keep
Token keep
Feature activation+1.542
a
Token a
Feature activation+0.507
DACA
Token DACA
Feature activation+0.000
fix
Token fix
Feature activation+0.188
separate
Token separate
Feature activation+0.000
from
Token from
Feature activation+0.000
emerged
Token emerged
Feature activation+0.068
.
Token.
Feature activation+0.020
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
operation
Token operation
Feature activation+1.460
saw
Token saw
Feature activation+1.031
US
Token US
Feature activation+0.000
agents
Token agents
Feature activation+0.000
lose
Token lose
Feature activation+0.000
track
Token track
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Democrats
TokenDemocrats
Feature activation+0.000
commonly
Token commonly
Feature activation+0.000
prefer
Token prefer
Feature activation+1.410
a
Token a
Feature activation+1.561
statistical
Token statistical
Feature activation+0.000
sampling
Token sampling
Feature activation+0.000
method
Token method
Feature activation+0.000
meant
Token meant
Feature activation+0.000
reviewed
Tokenreviewed
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
"
Token"
Feature activation+0.433
This
TokenThis
Feature activation+1.368
study
Token study
Feature activation+0.444
does
Token does
Feature activation+0.448
not
Token not
Feature activation+0.000
show
Token show
Feature activation+0.000
that
Token that
Feature activation+0.000
following
Token following
Feature activation+0.000
the
Token the
Feature activation+0.000
launch
Token launch
Feature activation+0.093
of
Token of
Feature activation+0.431
AJ
Token AJ
Feature activation+0.390
E
TokenE
Feature activation+1.358
in
Token in
Feature activation+0.000
Canada
Token Canada
Feature activation+0.000
concerning
Token concerning
Feature activation+0.000
its
Token its
Feature activation+0.489
programming
Token programming
Feature activation+0.000
U
Token U
Feature activation+0.000
.
Token.
Feature activation+0.000
S
TokenS
Feature activation+0.000
.
Token.
Feature activation+0.000
exports
Token exports
Feature activation+0.000
under
Token under
Feature activation+1.314
the
Token the
Feature activation+1.215
FTA
Token FTA
Feature activation+0.444
brought
Token brought
Feature activation+0.000
a
Token a
Feature activation+0.000
21
Token 21
Feature activation+0.000
door
Token door
Feature activation+0.294
-
Token-
Feature activation+0.184
to
Tokento
Feature activation+0.000
-
Token-
Feature activation+0.000
door
Tokendoor
Feature activation+0.758
approach
Token approach
Feature activation+1.309
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
A
TokenA
Feature activation+0.000
1999
Token 1999
Feature activation+0.000
local
Token local
Feature activation+0.000
film
Token film
Feature activation+0.000
school
Token school
Feature activation+0.000
.
Token.
Feature activation+0.000
The
Token The
Feature activation+0.000
Rogue
Token Rogue
Feature activation+1.295
Film
Token Film
Feature activation+0.574
School
Token School
Feature activation+0.000
is
Token is
Feature activation+0.000
about
Token about
Feature activation+0.000
a
Token a
Feature activation+0.000
announced
Token announced
Feature activation+0.000
it
Token it
Feature activation+0.000
was
Token was
Feature activation+0.000
delaying
Token delaying
Feature activation+0.000
implementation
Token implementation
Feature activation+0.000
of
Token of
Feature activation+1.276
the
Token the
Feature activation+1.107
rule
Token rule
Feature activation+0.022
.
Token.
Feature activation+0.000
Sch
Token Sch
Feature activation+0.000
len
Tokenlen
Feature activation+0.000
size
Token size
Feature activation+0.000
and
Token and
Feature activation+0.000
impact
Token impact
Feature activation+0.000
of
Token of
Feature activation+0.154
the
Token the
Feature activation+0.165
CIA
Token CIA
Feature activation+1.217
's
Token's
Feature activation+1.167
Syria
Token Syria
Feature activation+0.000
weapons
Token weapons
Feature activation+0.000
program
Token program
Feature activation+0.000
,
Token,
Feature activation+0.000
illustrates
Token illustrates
Feature activation+0.000
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.000
implementation
Token implementation
Feature activation+0.594
of
Token of
Feature activation+1.575
the
Token the
Feature activation+1.216
directive
Token directive
Feature activation+0.544
led
Token led
Feature activation+0.000
to
Token to
Feature activation+0.000
the
Token the
Feature activation+0.000
ordering
Token ordering
Feature activation+0.000
.
Token.
Feature activation+0.000
S
TokenS
Feature activation+0.000
.
Token.
Feature activation+0.000
exports
Token exports
Feature activation+0.000
under
Token under
Feature activation+1.314
the
Token the
Feature activation+1.215
FTA
Token FTA
Feature activation+0.444
brought
Token brought
Feature activation+0.000
a
Token a
Feature activation+0.000
21
Token 21
Feature activation+0.000
percent
Token percent
Feature activation+0.000
are
Token are
Feature activation+0.000
signing
Token signing
Feature activation+0.000
up
Token up
Feature activation+0.000
to
Token to
Feature activation+0.000
our
Token our
Feature activation+0.172
programme
Token programme
Feature activation+1.182
.
Token.
Feature activation+0.332
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
"
Token"
Feature activation+0.000
Since
TokenSince
Feature activation+0.000
as
Token as
Feature activation+0.000
one
Token one
Feature activation+0.000
officer
Token officer
Feature activation+0.000
used
Token used
Feature activation+0.798
the
Token the
Feature activation+0.976
new
Token new
Feature activation+1.173
knowledge
Token knowledge
Feature activation+0.000
to
Token to
Feature activation+0.000
help
Token help
Feature activation+0.000
protect
Token protect
Feature activation+0.000
a
Token a
Feature activation+0.000
to
Token to
Feature activation+0.000
protect
Token protect
Feature activation+0.000
immigrants
Token immigrants
Feature activation+0.000
under
Token under
Feature activation+0.000
the
Token the
Feature activation+0.059
program
Token program
Feature activation+1.171
,
Token,
Feature activation+0.000
commonly
Token commonly
Feature activation+0.000
known
Token known
Feature activation+0.000
as
Token as
Feature activation+0.000
Dream
Token Dream
Feature activation+0.000
and
Token and
Feature activation+0.000
impact
Token impact
Feature activation+0.000
of
Token of
Feature activation+0.154
the
Token the
Feature activation+0.165
CIA
Token CIA
Feature activation+1.217
's
Token's
Feature activation+1.167
Syria
Token Syria
Feature activation+0.000
weapons
Token weapons
Feature activation+0.000
program
Token program
Feature activation+0.000
,
Token,
Feature activation+0.000
but
Token but
Feature activation+0.000
Union
Token Union
Feature activation+0.000
in
Token in
Feature activation+0.000
Afghanistan
Token Afghanistan
Feature activation+0.000
.
Token.
Feature activation+0.000
That
Token That
Feature activation+0.023
secretive
Token secretive
Feature activation+1.166
years
Token years
Feature activation+0.000
long
Token long
Feature activation+0.000
operation
Token operation
Feature activation+0.129
resulted
Token resulted
Feature activation+0.000
in
Token in
Feature activation+0.000

Top DFA by src position
MAX = 1.990

H
TokenH
Feature activation-0.008
Top resid features:
ann
Tokenann
Feature activation-0.000
Top resid features:
ibal
Tokenibal
Feature activation-0.003
Top resid features:
âĢ
TokenâĢ
Feature activation-0.001
Top resid features:
Ŀ
TokenĿ
Feature activation-0.002
Top resid features:
procedure
Token procedure
Feature activation+1.711
Top resid features:
following
Token following
Feature activation+0.048
Top resid features:
the
Token the
Feature activation+0.002
Top resid features:
capture
Token capture
Feature activation-0.004
Top resid features:
of
Token of
Feature activation+0.002
Top resid features:
Lieutenant
Token Lieutenant
Feature activation-0.004
Top resid features:
-
Token-
Feature activation-0.001
Top resid features:
to
Tokento
Feature activation-0.000
Top resid features:
-
Token-
Feature activation-0.000
Top resid features:
door
Tokendoor
Feature activation+0.021
Top resid features:
"
Token"
Feature activation+0.018
Top resid features:
approach
Token approach
Feature activation+1.268
Top resid features:
,
Token,
Feature activation+0.069
Top resid features:
which
Token which
Feature activation+0.344
Top resid features:
actually
Token actually
Feature activation+0.011
Top resid features:
is
Token is
Feature activation+0.010
Top resid features:
done
Token done
Feature activation+0.114
Top resid features:
editorial
Token editorial
Feature activation-0.005
Top resid features:
board
Token board
Feature activation-0.003
Top resid features:
endorsed
Token endorsed
Feature activation+0.004
Top resid features:
the
Token the
Feature activation+0.006
Top resid features:
new
Token new
Feature activation+0.001
Top resid features:
program
Token program
Feature activation+1.856
Top resid features:
yesterday
Token yesterday
Feature activation-0.003
Top resid features:
,
Token,
Feature activation+0.017
Top resid features:
insisting
Token insisting
Feature activation-0.002
Top resid features:
it
Token it
Feature activation+0.004
Top resid features:
was
Token was
Feature activation+0.008
Top resid features:
rescind
Token rescind
Feature activation+0.000
Top resid features:
the
Token the
Feature activation+0.005
Top resid features:
Obama
Token Obama
Feature activation+0.008
Top resid features:
-
Token-
Feature activation+0.003
Top resid features:
era
Tokenera
Feature activation+0.005
Top resid features:
program
Token program
Feature activation+1.187
Top resid features:
,
Token,
Feature activation+0.159
Top resid features:
giving
Token giving
Feature activation-0.002
Top resid features:
Congress
Token Congress
Feature activation+0.003
Top resid features:
a
Token a
Feature activation-0.001
Top resid features:
six
Token six
Feature activation-0.000
Top resid features:
how
Token how
Feature activation+0.003
Top resid features:
problems
Token problems
Feature activation+0.011
Top resid features:
with
Token with
Feature activation+0.013
Top resid features:
Fast
Token Fast
Feature activation+0.032
Top resid features:
and
Token and
Feature activation+0.004
Top resid features:
Furious
Token Furious
Feature activation+1.144
Top resid features:
emerged
Token emerged
Feature activation+0.015
Top resid features:
.
Token.
Feature activation+0.003
Top resid features:
Ċ
TokenĊ
Feature activation-0.007
Top resid features:
Ċ
TokenĊ
Feature activation-0.008
Top resid features:
The
TokenThe
Feature activation-0.017
Top resid features:
-
Token-
Feature activation-0.000
Top resid features:
to
Tokento
Feature activation-0.000
Top resid features:
-
Token-
Feature activation-0.000
Top resid features:
door
Tokendoor
Feature activation+0.002
Top resid features:
"
Token"
Feature activation+0.005
Top resid features:
approach
Token approach
Feature activation+1.016
Top resid features:
,
Token,
Feature activation+0.200
Top resid features:
which
Token which
Feature activation+0.317
Top resid features:
actually
Token actually
Feature activation+0.010
Top resid features:
is
Token is
Feature activation+0.005
Top resid features:
done
Token done
Feature activation+0.132
Top resid features:
Bever
Token Bever
Feature activation-0.001
Top resid features:
age
Tokenage
Feature activation-0.000
Top resid features:
Association
Token Association
Feature activation+0.004
Top resid features:
dismissed
Token dismissed
Feature activation+0.002
Top resid features:
the
Token the
Feature activation+0.036
Top resid features:
study
Token study
Feature activation+1.599
Top resid features:
when
Token when
Feature activation+0.002
Top resid features:
it
Token it
Feature activation+0.000
Top resid features:
was
Token was
Feature activation-0.001
Top resid features:
first
Token first
Feature activation-0.004
Top resid features:
presented
Token presented
Feature activation+0.005
Top resid features:
for
Token for
Feature activation+0.004
Top resid features:
the
Token the
Feature activation+0.005
Top resid features:
addition
Token addition
Feature activation-0.004
Top resid features:
of
Token of
Feature activation+0.004
Top resid features:
the
Token the
Feature activation+0.005
Top resid features:
service
Token service
Feature activation+1.649
Top resid features:
and
Token and
Feature activation-0.002
Top resid features:
the
Token the
Feature activation+0.002
Top resid features:
many
Token many
Feature activation+0.001
Top resid features:
positive
Token positive
Feature activation+0.000
Top resid features:
comments
Token comments
Feature activation+0.000
Top resid features:
experience
Token experience
Feature activation+0.002
Top resid features:
export
Token export
Feature activation-0.002
Top resid features:
growth
Token growth
Feature activation-0.002
Top resid features:
under
Token under
Feature activation+0.003
Top resid features:
the
Token the
Feature activation-0.004
Top resid features:
deal
Token deal
Feature activation+1.687
Top resid features:
.
Token.
Feature activation+0.002
Top resid features:
The
Token The
Feature activation-0.002
Top resid features:
decline
Token decline
Feature activation-0.016
Top resid features:
in
Token in
Feature activation-0.003
Top resid features:
U
Token U
Feature activation+0.007
Top resid features:
approach
Token approach
Feature activation+0.140
Top resid features:
,
Token,
Feature activation+0.062
Top resid features:
which
Token which
Feature activation+0.319
Top resid features:
actually
Token actually
Feature activation+0.048
Top resid features:
is
Token is
Feature activation+0.045
Top resid features:
done
Token done
Feature activation+0.705
Top resid features:
primarily
Token primarily
Feature activation+0.021
Top resid features:
by
Token by
Feature activation+0.089
Top resid features:
self
Token self
Feature activation-0.004
Top resid features:
-
Token-
Feature activation-0.000
Top resid features:
en
Tokenen
Feature activation-0.000
Top resid features:
advance
Token advance
Feature activation-0.000
Top resid features:
.
Token.
Feature activation-0.003
Top resid features:
The
Token The
Feature activation+0.003
Top resid features:
Rogue
Token Rogue
Feature activation+0.006
Top resid features:
Film
Token Film
Feature activation-0.202
Top resid features:
School
Token School
Feature activation+1.990
Top resid features:
will
Token will
Feature activation+0.048
Top resid features:
not
Token not
Feature activation+0.001
Top resid features:
teach
Token teach
Feature activation+0.025
Top resid features:
anything
Token anything
Feature activation+0.000
Top resid features:
technical
Token technical
Feature activation-0.001
Top resid features:
Republicans
Token Republicans
Feature activation-0.001
Top resid features:
tried
Token tried
Feature activation+0.001
Top resid features:
to
Token to
Feature activation+0.001
Top resid features:
rescind
Token rescind
Feature activation-0.005
Top resid features:
the
Token the
Feature activation+0.008
Top resid features:
rule
Token rule
Feature activation+1.519
Top resid features:
in
Token in
Feature activation+0.036
Top resid features:
May
Token May
Feature activation-0.001
Top resid features:
,
Token,
Feature activation-0.004
Top resid features:
but
Token but
Feature activation+0.001
Top resid features:
that
Token that
Feature activation-0.000
Top resid features:
-
Token -
Feature activation-0.001
Top resid features:
the
Token the
Feature activation+0.000
Top resid features:
1980
Token 1980
Feature activation+0.002
Top resid features:
's
Token's
Feature activation+0.002
Top resid features:
CIA
Token CIA
Feature activation+0.009
Top resid features:
program
Token program
Feature activation+0.820
Top resid features:
to
Token to
Feature activation+0.149
Top resid features:
arm
Token arm
Feature activation+0.194
Top resid features:
Afghan
Token Afghan
Feature activation-0.002
Top resid features:
and
Token and
Feature activation-0.000
Top resid features:
Arab
Token Arab
Feature activation-0.000
Top resid features:
H
TokenH
Feature activation-0.006
Top resid features:
ann
Tokenann
Feature activation+0.000
Top resid features:
ibal
Tokenibal
Feature activation-0.004
Top resid features:
âĢ
TokenâĢ
Feature activation-0.001
Top resid features:
Ŀ
TokenĿ
Feature activation-0.002
Top resid features:
procedure
Token procedure
Feature activation+1.184
Top resid features:
following
Token following
Feature activation+0.027
Top resid features:
the
Token the
Feature activation+0.002
Top resid features:
capture
Token capture
Feature activation-0.006
Top resid features:
of
Token of
Feature activation+0.001
Top resid features:
Lieutenant
Token Lieutenant
Feature activation-0.000
Top resid features:
experience
Token experience
Feature activation+0.000
Top resid features:
export
Token export
Feature activation-0.002
Top resid features:
growth
Token growth
Feature activation-0.000
Top resid features:
under
Token under
Feature activation+0.000
Top resid features:
the
Token the
Feature activation+0.006
Top resid features:
deal
Token deal
Feature activation+1.518
Top resid features:
.
Token.
Feature activation+0.022
Top resid features:
The
Token The
Feature activation-0.001
Top resid features:
decline
Token decline
Feature activation-0.011
Top resid features:
in
Token in
Feature activation-0.010
Top resid features:
U
Token U
Feature activation+0.012
Top resid features:
are
Token are
Feature activation+0.007
Top resid features:
signing
Token signing
Feature activation+0.001
Top resid features:
up
Token up
Feature activation-0.004
Top resid features:
to
Token to
Feature activation-0.003
Top resid features:
our
Token our
Feature activation+0.122
Top resid features:
programme
Token programme
Feature activation+1.389
Top resid features:
.
Token.
Feature activation+0.000
Top resid features:
Ċ
TokenĊ
Feature activation+0.000
Top resid features:
Ċ
TokenĊ
Feature activation+0.000
Top resid features:
"
Token"
Feature activation+0.000
Top resid features:
Since
TokenSince
Feature activation+0.000
Top resid features:
Dur
TokenDur
Feature activation+0.001
Top resid features:
ham
Tokenham
Feature activation+0.003
Top resid features:
Police
Token Police
Feature activation+0.003
Top resid features:
said
Token said
Feature activation-0.006
Top resid features:
the
Token the
Feature activation+0.005
Top resid features:
technique
Token technique
Feature activation+1.010
Top resid features:
has
Token has
Feature activation+0.070
Top resid features:
already
Token already
Feature activation+0.008
Top resid features:
proven
Token proven
Feature activation+0.062
Top resid features:
to
Token to
Feature activation+0.020
Top resid features:
be
Token be
Feature activation+0.044
Top resid features:
to
Token to
Feature activation+0.005
Top resid features:
protect
Token protect
Feature activation+0.017
Top resid features:
immigrants
Token immigrants
Feature activation-0.011
Top resid features:
under
Token under
Feature activation-0.009
Top resid features:
the
Token the
Feature activation+0.014
Top resid features:
program
Token program
Feature activation+1.096
Top resid features:
,
Token,
Feature activation+0.000
Top resid features:
commonly
Token commonly
Feature activation+0.000
Top resid features:
known
Token known
Feature activation+0.000
Top resid features:
as
Token as
Feature activation+0.000
Top resid features:
Dream
Token Dream
Feature activation+0.000
Top resid features:
-
Token -
Feature activation-0.000
Top resid features:
the
Token the
Feature activation-0.000
Top resid features:
1980
Token 1980
Feature activation+0.001
Top resid features:
's
Token's
Feature activation+0.002
Top resid features:
CIA
Token CIA
Feature activation+0.008
Top resid features:
program
Token program
Feature activation+0.661
Top resid features:
to
Token to
Feature activation+0.117
Top resid features:
arm
Token arm
Feature activation+0.158
Top resid features:
Afghan
Token Afghan
Feature activation-0.001
Top resid features:
and
Token and
Feature activation-0.000
Top resid features:
Arab
Token Arab
Feature activation+0.000
Top resid features:
-
Token -
Feature activation+0.000
Top resid features:
the
Token the
Feature activation+0.001
Top resid features:
1980
Token 1980
Feature activation+0.017
Top resid features:
's
Token's
Feature activation+0.018
Top resid features:
CIA
Token CIA
Feature activation+0.038
Top resid features:
program
Token program
Feature activation+1.385
Top resid features:
to
Token to
Feature activation+0.152
Top resid features:
arm
Token arm
Feature activation+0.089
Top resid features:
Afghan
Token Afghan
Feature activation+0.003
Top resid features:
and
Token and
Feature activation-0.000
Top resid features:
Arab
Token Arab
Feature activation-0.000
Top resid features:

Decoder Weights Distribution

Head 0: 0.06

Head 1: 0.48

Head 2: 0.03

Head 3: 0.07

Head 4: 0.03

Head 5: 0.08

Head 6: 0.07

Head 7: 0.04

Head 8: 0.03

Head 9: 0.05

Head 10: 0.05

Head 11: 0.03

Positive logits

Program2.95

program2.88

method2.69

programs2.66

PROGRAM2.63

Programme2.58

sham2.52

Method2.50

Simulation2.49

Prevention2.44

Intervention2.43

cycle2.38

Training2.33

contraceptive2.32

scam2.30

Experiment2.29

Method2.29

Invention2.28

experimented2.27

Solution2.27

Negative logits

cher-2.39

erning-2.38

-2.28

-2.22

urities-2.16

cious-2.16

retweet-2.14

Emin-2.14

terness-2.13

ital-2.12

citiz-2.11

Yon-2.11

bruised-2.10

adject-2.08

-2.06

-2.05

¯-2.05

fret-2.02

condol-2.01

-2.01

INTERVAL 1.643 - 1.825
CONTAINS 0.000%

INTERVAL 1.460 - 1.643
CONTAINS 0.001%

Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Democrats
TokenDemocrats
Feature activation+0.000
commonly
Token commonly
Feature activation+0.000
prefer
Token prefer
Feature activation+1.410
a
Token a
Feature activation+1.561
statistical
Token statistical
Feature activation+0.000
sampling
Token sampling
Feature activation+0.000
method
Token method
Feature activation+0.000
meant
Token meant
Feature activation+0.000
to
Token to
Feature activation+0.000
report
Token report
Feature activation+0.000
illustrates
Token illustrates
Feature activation+0.000
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.000
implementation
Token implementation
Feature activation+0.594
of
Token of
Feature activation+1.575
the
Token the
Feature activation+1.216
directive
Token directive
Feature activation+0.544
led
Token led
Feature activation+0.000
to
Token to
Feature activation+0.000
the
Token the
Feature activation+0.000
ever
Token ever
Feature activation+0.000
think
Token think
Feature activation+0.000
that
Token that
Feature activation+0.000
this
Token this
Feature activation+0.004
wonderful
Token wonderful
Feature activation+0.218
new
Token new
Feature activation+1.556
program
Token program
Feature activation+0.577
sounded
Token sounded
Feature activation+0.000
like
Token like
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
,
Token,
Feature activation+0.000
who
Token who
Feature activation+0.000
has
Token has
Feature activation+0.000
pushed
Token pushed
Feature activation+0.000
to
Token to
Feature activation+0.000
keep
Token keep
Feature activation+1.542
a
Token a
Feature activation+0.507
DACA
Token DACA
Feature activation+0.000
fix
Token fix
Feature activation+0.188
separate
Token separate
Feature activation+0.000
from
Token from
Feature activation+0.000
emerged
Token emerged
Feature activation+0.068
.
Token.
Feature activation+0.020
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
operation
Token operation
Feature activation+1.460
saw
Token saw
Feature activation+1.031
US
Token US
Feature activation+0.000
agents
Token agents
Feature activation+0.000
lose
Token lose
Feature activation+0.000
track
Token track
Feature activation+0.000

INTERVAL 1.278 - 1.460
CONTAINS 0.001%

following
Token following
Feature activation+0.000
the
Token the
Feature activation+0.000
launch
Token launch
Feature activation+0.093
of
Token of
Feature activation+0.431
AJ
Token AJ
Feature activation+0.390
E
TokenE
Feature activation+1.358
in
Token in
Feature activation+0.000
Canada
Token Canada
Feature activation+0.000
concerning
Token concerning
Feature activation+0.000
its
Token its
Feature activation+0.489
programming
Token programming
Feature activation+0.000
door
Token door
Feature activation+0.294
-
Token-
Feature activation+0.184
to
Tokento
Feature activation+0.000
-
Token-
Feature activation+0.000
door
Tokendoor
Feature activation+0.758
approach
Token approach
Feature activation+1.309
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
A
TokenA
Feature activation+0.000
1999
Token 1999
Feature activation+0.000
local
Token local
Feature activation+0.000
film
Token film
Feature activation+0.000
school
Token school
Feature activation+0.000
.
Token.
Feature activation+0.000
The
Token The
Feature activation+0.000
Rogue
Token Rogue
Feature activation+1.295
Film
Token Film
Feature activation+0.574
School
Token School
Feature activation+0.000
is
Token is
Feature activation+0.000
about
Token about
Feature activation+0.000
a
Token a
Feature activation+0.000
reviewed
Tokenreviewed
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
"
Token"
Feature activation+0.433
This
TokenThis
Feature activation+1.368
study
Token study
Feature activation+0.444
does
Token does
Feature activation+0.448
not
Token not
Feature activation+0.000
show
Token show
Feature activation+0.000
that
Token that
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Democrats
TokenDemocrats
Feature activation+0.000
commonly
Token commonly
Feature activation+0.000
prefer
Token prefer
Feature activation+1.410
a
Token a
Feature activation+1.561
statistical
Token statistical
Feature activation+0.000
sampling
Token sampling
Feature activation+0.000
method
Token method
Feature activation+0.000
meant
Token meant
Feature activation+0.000

INTERVAL 1.095 - 1.278
CONTAINS 0.002%

the
Token the
Feature activation+0.452
âĢ
Token âĢ
Feature activation+0.108
ľ
Tokenľ
Feature activation+0.683
H
TokenH
Feature activation+0.000
ann
Tokenann
Feature activation+0.000
ibal
Tokenibal
Feature activation+1.153
Directive
Token Directive
Feature activation+0.009
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.331
,
Token,
Feature activation+0.021
Israeli
Token Israeli
Feature activation+0.000
state
Token state
Feature activation+0.000
-
Token-
Feature activation+0.000
by
Tokenby
Feature activation+0.000
-
Token-
Feature activation+0.000
state
Tokenstate
Feature activation+0.000
format
Token format
Feature activation+1.135
will
Token will
Feature activation+0.000
likely
Token likely
Feature activation+0.000
encounter
Token encounter
Feature activation+0.000
.
Token.
Feature activation+0.000
Also
Token Also
Feature activation+0.000
care
Token care
Feature activation+0.000
industries
Token industries
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Another
TokenAnother
Feature activation+1.115
1
Token 1
Feature activation+0.080
-
Token-
Feature activation+0.000
tr
Tokentr
Feature activation+0.000
illion
Tokenillion
Feature activation+0.000
-
Token-
Feature activation+0.000
it
Token it
Feature activation+0.000
was
Token was
Feature activation+0.000
delaying
Token delaying
Feature activation+0.000
implementation
Token implementation
Feature activation+0.000
of
Token of
Feature activation+1.276
the
Token the
Feature activation+1.107
rule
Token rule
Feature activation+0.022
.
Token.
Feature activation+0.000
Sch
Token Sch
Feature activation+0.000
len
Tokenlen
Feature activation+0.000
ker
Tokenker
Feature activation+0.000
illustrates
Token illustrates
Feature activation+0.000
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.000
implementation
Token implementation
Feature activation+0.594
of
Token of
Feature activation+1.575
the
Token the
Feature activation+1.216
directive
Token directive
Feature activation+0.544
led
Token led
Feature activation+0.000
to
Token to
Feature activation+0.000
the
Token the
Feature activation+0.000
ordering
Token ordering
Feature activation+0.000

INTERVAL 0.913 - 1.095
CONTAINS 0.002%

.
Token.
Feature activation+0.020
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
operation
Token operation
Feature activation+1.460
saw
Token saw
Feature activation+1.031
US
Token US
Feature activation+0.000
agents
Token agents
Feature activation+0.000
lose
Token lose
Feature activation+0.000
track
Token track
Feature activation+0.000
of
Token of
Feature activation+0.000
button
Token button
Feature activation+0.000
on
Token on
Feature activation+0.000
the
Token the
Feature activation+0.000
notification
Token notification
Feature activation+0.000
,
Token,
Feature activation+0.000
Ch
Token Ch
Feature activation+1.010
urnal
Tokenurnal
Feature activation+0.447
ism
Tokenism
Feature activation+0.010
overl
Token overl
Feature activation+0.000
ays
Tokenays
Feature activation+0.000
a
Token a
Feature activation+0.000
âĢ
Token âĢ
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
a
Tokena
Feature activation+0.502
re
Token re
Feature activation+0.195
education
Tokeneducation
Feature activation+0.419
camp
Token camp
Feature activation+1.051
being
Token being
Feature activation+0.174
administered
Token administered
Feature activation+0.268
by
Token by
Feature activation+0.000
an
Token an
Feature activation+0.000
oppressed
Token oppressed
Feature activation+0.000
Why
Token Why
Feature activation+0.000
is
Token is
Feature activation+0.000
the
Token the
Feature activation+0.000
government
Token government
Feature activation+0.049
launching
Token launching
Feature activation+0.618
a
Token a
Feature activation+1.038
campaign
Token campaign
Feature activation+0.000
to
Token to
Feature activation+0.000
end
Token end
Feature activation+0.000
violence
Token violence
Feature activation+0.000
against
Token against
Feature activation+0.000
,
Token,
Feature activation+0.000
who
Token who
Feature activation+0.000
had
Token had
Feature activation+0.000
supported
Token supported
Feature activation+0.000
reform
Token reform
Feature activation+0.338
to
Token to
Feature activation+1.067
that
Token that
Feature activation+0.272
point
Token point
Feature activation+0.000
,
Token,
Feature activation+0.000
voted
Token voted
Feature activation+0.000
no
Token no
Feature activation+0.000

INTERVAL 0.730 - 0.913
CONTAINS 0.004%

Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Under
TokenUnder
Feature activation+0.436
the
Token the
Feature activation+0.205
planned
Token planned
Feature activation+0.739
expansion
Token expansion
Feature activation+0.794
,
Token,
Feature activation+0.071
the
Token the
Feature activation+0.000
agency
Token agency
Feature activation+0.000
would
Token would
Feature activation+0.000
hand
Token hand
Feature activation+0.000
he
Token he
Feature activation+0.000
holds
Token holds
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Jim
TokenJim
Feature activation+0.766
B
Token B
Feature activation+0.000
uss
Tokenuss
Feature activation+0.000
will
Token will
Feature activation+0.000
be
Token be
Feature activation+0.000
gone
Token gone
Feature activation+0.000
ľ
Tokenľ
Feature activation+0.000
It
TokenIt
Feature activation+0.118
was
Token was
Feature activation+0.000
not
Token not
Feature activation+0.000
a
Token a
Feature activation+0.445
Wild
Token Wild
Feature activation+0.746
cat
Tokencat
Feature activation+0.122
play
Token play
Feature activation+0.000
,
Token,
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ŀ
TokenĿ
Feature activation+0.000
to
Token to
Feature activation+0.000
line
Token line
Feature activation+0.000
up
Token up
Feature activation+0.000
all
Token all
Feature activation+0.000
the
Token the
Feature activation+0.000
K
Token K
Feature activation+0.785
-
Token-
Feature activation+0.583
C
TokenC
Feature activation+0.000
ups
Tokenups
Feature activation+0.000
discarded
Token discarded
Feature activation+0.000
in
Token in
Feature activation+0.000
Overall
Token Overall
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
I
TokenI
Feature activation+0.000
read
Token read
Feature activation+0.000
this
Token this
Feature activation+0.732
book
Token book
Feature activation+0.004
the
Token the
Feature activation+0.000
first
Token first
Feature activation+0.000
time
Token time
Feature activation+0.000
at
Token at
Feature activation+0.000

INTERVAL 0.548 - 0.730
CONTAINS 0.012%

and
Token and
Feature activation+0.000
will
Token will
Feature activation+0.000
drop
Token drop
Feature activation+0.000
the
Token the
Feature activation+0.000
conver
Token conver
Feature activation+0.446
ged
Tokenged
Feature activation+0.697
flight
Token flight
Feature activation+0.000
system
Token system
Feature activation+0.080
from
Token from
Feature activation+0.000
final
Token final
Feature activation+0.000
plans
Token plans
Feature activation+0.000
oval
Token oval
Feature activation+0.000
pitches
Token pitches
Feature activation+0.000
.
Token.
Feature activation+0.080
Australian
Token Australian
Feature activation+0.000
Rules
Token Rules
Feature activation+0.000
Football
Token Football
Feature activation+0.646
is
Token is
Feature activation+0.098
not
Token not
Feature activation+0.000
the
Token the
Feature activation+0.000
same
Token same
Feature activation+0.000
as
Token as
Feature activation+0.000
North
Token North
Feature activation+0.000
Island
Token Island
Feature activation+0.000
).
Token).
Feature activation+0.000
It
Token It
Feature activation+0.000
was
Token was
Feature activation+0.000
held
Token held
Feature activation+0.661
at
Token at
Feature activation+0.000
Tak
Token Tak
Feature activation+0.000
ap
Tokenap
Feature activation+0.000
Å«
Tokenū
Feature activation+0.000
w
Tokenw
Feature activation+0.000
looking
Token looking
Feature activation+0.000
for
Token for
Feature activation+0.194
ways
Token ways
Feature activation+0.000
to
Token to
Feature activation+0.000
keep
Token keep
Feature activation+0.210
the
Token the
Feature activation+0.667
practice
Token practice
Feature activation+0.500
alive
Token alive
Feature activation+0.000
here
Token here
Feature activation+0.000
in
Token in
Feature activation+0.000
the
Token the
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.325
still
Token still
Feature activation+0.109
legal
Token legal
Feature activation+0.219
to
Token to
Feature activation+0.618
capture
Token capture
Feature activation+0.000
killer
Token killer
Feature activation+0.000
whales
Token whales
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000

INTERVAL 0.365 - 0.548
CONTAINS 0.028%

fact
Token fact
Feature activation+0.000
that
Token that
Feature activation+0.000
the
Token the
Feature activation+0.000
majority
Token majority
Feature activation+0.000
is
Token is
Feature activation+0.000
against
Token against
Feature activation+0.482
something
Token something
Feature activation+0.111
isn
Token isn
Feature activation+0.000
't
Token't
Feature activation+0.000
much
Token much
Feature activation+0.000
of
Token of
Feature activation+0.000
to
Token to
Feature activation+0.000
Trump
Token Trump
Feature activation+0.000
have
Token have
Feature activation+0.000
called
Token called
Feature activation+0.000
for
Token for
Feature activation+0.238
Single
Token Single
Feature activation+0.503
P
Token P
Feature activation+0.000
ayer
Tokenayer
Feature activation+0.029
and
Token and
Feature activation+0.000
most
Token most
Feature activation+0.000
importantly
Token importantly
Feature activation+0.000
weight
Token weight
Feature activation+0.000
-
Token-
Feature activation+0.000
wise
Tokenwise
Feature activation+0.000
.
Token.
Feature activation+0.000
What
TokenWhat
Feature activation+0.000
Pin
Token Pin
Feature activation+0.367
ion
Tokenion
Feature activation+0.000
have
Token have
Feature activation+0.159
behind
Token behind
Feature activation+0.000
them
Token them
Feature activation+0.000
is
Token is
Feature activation+0.000
,
Token,
Feature activation+0.000
with
Token with
Feature activation+0.000
the
Token the
Feature activation+0.000
advent
Token advent
Feature activation+0.000
of
Token of
Feature activation+0.000
reverse
Token reverse
Feature activation+0.369
vending
Token vending
Feature activation+0.000
machines
Token machines
Feature activation+0.000
making
Token making
Feature activation+0.000
deposit
Token deposit
Feature activation+0.000
refunds
Token refunds
Feature activation+0.000
my
Token my
Feature activation+0.000
community
Token community
Feature activation+0.000
who
Token who
Feature activation+0.000
talk
Token talk
Feature activation+0.000
about
Token about
Feature activation+0.000
it
Token it
Feature activation+0.443
as
Token as
Feature activation+0.025
if
Token if
Feature activation+0.000
it
Token it
Feature activation+0.582
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000

INTERVAL 0.183 - 0.365
CONTAINS 0.081%

Stewart
Token Stewart
Feature activation+0.000
Johnston
Token Johnston
Feature activation+0.000
,
Token,
Feature activation+0.000
the
Token the
Feature activation+0.000
landlord
Token landlord
Feature activation+0.000
of
Token of
Feature activation+0.204
the
Token the
Feature activation+0.223
dispensary
Token dispensary
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.000
boldly
Token boldly
Feature activation+0.000
spoke
Token spoke
Feature activation+0.000
out
Token out
Feature activation+0.000
against
Token against
Feature activation+0.000
the
Token the
Feature activation+0.000
practice
Token practice
Feature activation+0.233
in
Token in
Feature activation+0.000
an
Token an
Feature activation+0.000
award
Token award
Feature activation+0.000
-
Token-
Feature activation+0.000
winning
Tokenwinning
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Sounds
TokenSounds
Feature activation+0.000
like
Token like
Feature activation+0.000
a
Token a
Feature activation+0.208
win
Token win
Feature activation+0.000
-
Token-
Feature activation+0.000
win
Tokenwin
Feature activation+0.000
to
Token to
Feature activation+0.000
me
Token me
Feature activation+0.000
There
TokenThere
Feature activation+0.000
is
Token is
Feature activation+0.000
no
Token no
Feature activation+0.000
known
Token known
Feature activation+0.000
cure
Token cure
Feature activation+0.000
for
Token for
Feature activation+0.356
Ebola
Token Ebola
Feature activation+0.000
,
Token,
Feature activation+0.000
which
Token which
Feature activation+0.000
begins
Token begins
Feature activation+0.000
with
Token with
Feature activation+0.000
âĢĵ
Token âĢĵ
Feature activation+0.000
and
Token and
Feature activation+0.000
getting
Token getting
Feature activation+0.000
dirty
Token dirty
Feature activation+0.000
!
Token!
Feature activation+0.000
Fest
Token Fest
Feature activation+0.211
ivities
Tokenivities
Feature activation+0.000
include
Token include
Feature activation+0.000
:
Token:
Feature activation+0.000
Once
TokenOnce
Feature activation+0.000
again
Token again
Feature activation+0.000

INTERVAL 0.000 - 0.183
CONTAINS 99.871%

between
Token between
Feature activation+0.000
the
Token the
Feature activation+0.000
British
Token British
Feature activation+0.000
Council
Token Council
Feature activation+0.000
,
Token,
Feature activation+0.000
BBC
Token BBC
Feature activation+0.000
World
Token World
Feature activation+0.000
Service
Token Service
Feature activation+0.000
and
Token and
Feature activation+0.000
BBC
Token BBC
Feature activation+0.000
Radio
Token Radio
Feature activation+0.000
the
Token the
Feature activation+0.000
Baltimore
Token Baltimore
Feature activation+0.000
Ravens
Token Ravens
Feature activation+0.000
to
Token to
Feature activation+0.000
move
Token move
Feature activation+0.000
from
Token from
Feature activation+0.000
the
Token the
Feature activation+0.000
28
Token 28
Feature activation+0.000
th
Tokenth
Feature activation+0.000
position
Token position
Feature activation+0.000
to
Token to
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
t
Tokent
Feature activation+0.000
a
Token a
Feature activation+0.000
0
Token 0
Feature activation+0.000
-
Token-
Feature activation+0.000
techn
Tokentechn
Feature activation+0.000
ique
Tokenique
Feature activation+0.000
guy
Token guy
Feature activation+0.000
.
Token.
Feature activation+0.000
He
Token He
Feature activation+0.000
is
Token is
Feature activation+0.000
as
Token as
Feature activation+0.000
low
Token low
Feature activation+0.000
crude
Token crude
Feature activation+0.000
prices
Token prices
Feature activation+0.000
pinch
Token pinch
Feature activation+0.000
the
Token the
Feature activation+0.000
OPEC
Token OPEC
Feature activation+0.000
nation
Token nation
Feature activation+0.000
âĢ
TokenâĢ
Feature activation+0.000
Ļ
TokenĻ
Feature activation+0.000
s
Tokens
Feature activation+0.000
GA
TokenGA
Feature activation+0.000
Tour
Token Tour
Feature activation+0.000
in
Token in
Feature activation+0.000
1986
Token 1986
Feature activation+0.000
at
Token at
Feature activation+0.000
age
Token age
Feature activation+0.000
22
Token 22
Feature activation+0.000
,
Token,
Feature activation+0.000
averaging
Token averaging
Feature activation+0.000
285
Token 285
Feature activation+0.000
yards
Token yards
Feature activation+0.000

BOTTOM ACTIVATIONS
MIN = 0.000

atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
Development
Token Development
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
&
Token &
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Defence
Token Defence
Feature activation+0.000
Research
Token Research
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
<|endoftext|>
Token<|endoftext|>
Feature activation+0.000
rik
Tokenrik
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
ar
Tokenar
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
said
Token said
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
A
Token A
Feature activation+0.000
atre
Tokenatre
Feature activation+0.000
,
Token,
Feature activation+0.000
a
Token a
Feature activation+0.000
former
Token former
Feature activation+0.000
boss
Token boss
Feature activation+0.000
of
Token of
Feature activation+0.000
the
Token the
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
K
TokenK
Feature activation+0.000
Saturday
Token Saturday
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
Ċ
TokenĊ
Feature activation+0.000
The
TokenThe
Feature activation+0.000
defence
Token defence
Feature activation+0.000
ministry
Token ministry
Feature activation+0.000
committee
Token committee
Feature activation+0.000
headed
Token headed
Feature activation+0.000
by
Token by
Feature activation+0.000
V
Token V
Feature activation+0.000
.
Token.
Feature activation+0.000