TOP ACTIVATIONS
MAX = 3.813

confused . He stops how ling and scratching . He looks
was the monster , how ling and grow ling ! The
, and the wind how ling outside . They felt cold
woods and practice its how ling . The wolf wanted to
, the wolf 's how ling got stronger and stronger .
too cold for a re ind eer like you !"
ogre ran away , how ling in pain .
, the wolf tried how ling again in the clearing .
â s how ling was so loud it made
deep breath and started how ling . But the wolf â
and the wind was how ling . The pilot was so
work on improving its how ling . It practiced in the
to show off its how ling and make the rabbit scared
, the pup started how ling ! It was very
and all the other re ind eers . The
not know what a re ind eer is . They ask
â s how ling was not very loud .
a picture of a re ind eer . She wanted a
, but it also so ot hed the pain . She
both design something and show me later ." L

MIN ACTIVATIONS
MIN = -3.624

awn â s fault . The
felt like it was her fault , like she had done
realised it wasn 't her fault after all . <|endoftext|> Once
" It 's not my fault . I can 't help
Ur ge your heart and mind to
it was Tim my 's fault but Tim my felt guilty
knew it was his own fault . He would have to
knew it was her own fault . The girl
" It 's not your fault , honey . Ben made
that it wasn 't her fault . Finally , the family
it was all Joe 's fault . Joe tried to talk
He realized it was his fault and he felt regret .
" It 's your own fault , you should 've been
bird . You admit your fault and ask for mercy .
realized it was his own fault for not val uing her
" It was not our fault , mom . The phone
. " It was my fault . I threw a rock
. It is not your fault . It is not anyone
he knew it was his fault . In the end ,
said it was not his fault . She said the bird

INTERVAL 1.954 - 3.813
CONTAINS 0.032%

but you have to serve me first ." Tim my didn
how to play hide and seek instead . Lily liked this
Soup is better . Let me have some !"
are unt ied . Let me help you tie them up
you to paint something for me . Maybe you can paint

INTERVAL 0.094 - 1.954
CONTAINS 40.179%

good at drawing . One day , Mark wants
and go to your room . You are in big trouble
Let 's do something else : we can splash in the
make some cocoa . When her mom my
They sat down and worked together . They tied the hair

INTERVAL -1.765 - 0.094
CONTAINS 59.725%

tire swing . She looked at her mom my and said
told her it was five dollars . L ily
One sunny day , Max and Lily went to the park
to pull his train away . He thought of a plan
The bird liked to fly around the sky , near the

INTERVAL -3.624 - -1.765
CONTAINS 0.064%

Tw inkle tw inkle little star , how I wonder what
. We can â t let you play until
something strange a towel ! The
special prize a big , shiny
a shirt and pants combo - and brushed his hair into
Token confused
Token ID10416
Feature activation+9.98e-02

Pos logit contributions
 fault+0.005
 pret+0.004
 enough+0.004
 puck+0.004
 loses+0.004
Neg logit contributions
date-0.006
 hurricane-0.005
 tornado-0.005
 year-0.005
cient-0.005
Token.
Token ID13
Feature activation+1.85e-01

Pos logit contributions
 hurricane+0.005
date+0.005
 tornado+0.004
 Amen+0.004
 television+0.004
Neg logit contributions
 fault-0.008
 pret-0.007
paper-0.007
 loses-0.007
 enough-0.007
Token He
Token ID679
Feature activation-2.53e-02

Pos logit contributions
 hurricane+0.012
date+0.012
 tornado+0.012
 year+0.011
 voice+0.011
Neg logit contributions
 fault-0.011
 pret-0.009
paper-0.009
 pounds-0.009
 loses-0.009
Token stops
Token ID9911
Feature activation-1.00e+00

Pos logit contributions
 fault+0.001
 pret+0.001
paper+0.001
 loses+0.001
 puck+0.001
Neg logit contributions
 hurricane-0.002
date-0.002
 tornado-0.002
 year-0.002
 television-0.002
Token how
Token ID703
Feature activation-2.14e-01

Pos logit contributions
 fault+0.050
 pret+0.050
 loses+0.045
paper+0.044
 puck+0.043
Neg logit contributions
date-0.076
 hurricane-0.071
ilt-0.069
 Amen-0.068
 tornado-0.068
Tokenling
Token ID1359
Feature activation+3.05e+00

Pos logit contributions
 fault+0.013
 pret+0.012
paper+0.011
 puck+0.011
 loses+0.011
Neg logit contributions
 hurricane-0.013
date-0.013
 tornado-0.013
 year-0.012
 television-0.012
Token and
Token ID290
Feature activation+2.76e-01

Pos logit contributions
date+0.212
 hurricane+0.212
 Amen+0.201
 tornado+0.200
 television+0.198
Neg logit contributions
 fault-0.155
 pret-0.144
 loses-0.135
 puck-0.134
 enough-0.133
Token scratching
Token ID34688
Feature activation+7.22e-01

Pos logit contributions
 hurricane+0.017
date+0.017
 tornado+0.016
 year+0.016
 television+0.016
Neg logit contributions
 fault-0.017
 pret-0.015
paper-0.015
 loses-0.014
 puck-0.014
Token.
Token ID13
Feature activation+1.87e-01

Pos logit contributions
 hurricane+0.044
date+0.044
 tornado+0.041
 Amen+0.041
 year+0.040
Neg logit contributions
 fault-0.044
 pret-0.040
paper-0.038
 loses-0.038
 enough-0.038
Token He
Token ID679
Feature activation-2.59e-02

Pos logit contributions
 hurricane+0.012
date+0.011
 tornado+0.011
 voice+0.011
 television+0.011
Neg logit contributions
 fault-0.011
 pret-0.010
 pounds-0.010
paper-0.010
 loses-0.009
Token looks
Token ID3073
Feature activation+4.37e-02

Pos logit contributions
 fault+0.001
paper+0.001
 pounds+0.001
 pret+0.001
eter+0.001
Neg logit contributions
 hurricane-0.002
 tornado-0.002
 voice-0.002
date-0.002
 television-0.002
Token was
Token ID373
Feature activation+4.05e-01

Pos logit contributions
 hurricane+0.019
date+0.019
 tornado+0.018
 year+0.018
 voice+0.018
Neg logit contributions
 fault-0.025
 pret-0.023
 puck-0.021
 pounds-0.021
ws-0.021
Token the
Token ID262
Feature activation+2.01e-01

Pos logit contributions
 hurricane+0.028
date+0.027
 tornado+0.026
 year+0.026
 television+0.026
Neg logit contributions
 fault-0.022
 pret-0.020
paper-0.019
 loses-0.018
 pounds-0.018
Token monster
Token ID9234
Feature activation-4.01e-01

Pos logit contributions
date+0.011
 Amen+0.010
han+0.010
 hurricane+0.010
ind+0.010
Neg logit contributions
 fault-0.014
 pret-0.013
 puck-0.012
 loses-0.012
 enough-0.012
Token,
Token ID11
Feature activation+2.81e-01

Pos logit contributions
 fault+0.027
 pret+0.024
paper+0.024
 loses+0.023
 enough+0.023
Neg logit contributions
 hurricane-0.022
date-0.022
 tornado-0.021
 Amen-0.021
 television-0.020
Token how
Token ID703
Feature activation+2.38e-01

Pos logit contributions
date+0.019
 hurricane+0.019
 Amen+0.018
 tornado+0.017
ind+0.017
Neg logit contributions
 fault-0.015
 pret-0.014
 puck-0.013
 enough-0.013
 loses-0.013
Tokenling
Token ID1359
Feature activation+3.04e+00

Pos logit contributions
 hurricane+0.015
date+0.014
 tornado+0.014
 television+0.013
 year+0.013
Neg logit contributions
 fault-0.015
 pret-0.013
paper-0.013
 loses-0.012
 puck-0.012
Token and
Token ID290
Feature activation+1.65e-01

Pos logit contributions
date+0.208
 Amen+0.207
 hurricane+0.203
ind+0.197
 television+0.195
Neg logit contributions
 fault-0.149
 loses-0.142
 pret-0.142
 enough-0.139
paper-0.133
Token grow
Token ID1663
Feature activation+1.11e-01

Pos logit contributions
date+0.010
 hurricane+0.009
 tornado+0.009
 Amen+0.009
 year+0.009
Neg logit contributions
 fault-0.011
 pret-0.010
paper-0.009
 loses-0.009
 puck-0.009
Tokenling
Token ID1359
Feature activation+1.58e+00

Pos logit contributions
 hurricane+0.005
date+0.005
 tornado+0.005
 year+0.005
 television+0.005
Neg logit contributions
 fault-0.008
 pret-0.008
paper-0.007
 puck-0.007
 loses-0.007
Token!
Token ID0
Feature activation-2.45e-01

Pos logit contributions
 hurricane+0.098
date+0.097
 tornado+0.091
 Amen+0.091
ind+0.089
Neg logit contributions
 fault-0.095
 pret-0.087
 enough-0.083
 loses-0.082
paper-0.082
Token The
Token ID383
Feature activation-3.73e-03

Pos logit contributions
 fault+0.014
 pret+0.012
paper+0.012
 puck+0.012
 loses+0.012
Neg logit contributions
 hurricane-0.016
date-0.016
 tornado-0.015
ind-0.015
 year-0.015
Token,
Token ID11
Feature activation+2.76e-01

Pos logit contributions
 fault+0.028
 pret+0.026
paper+0.025
 enough+0.025
 loses+0.025
Neg logit contributions
 hurricane-0.021
date-0.021
 Amen-0.020
han-0.020
 tornado-0.019
Token and
Token ID290
Feature activation+4.22e-01

Pos logit contributions
 hurricane+0.017
date+0.016
 tornado+0.016
 television+0.016
 voice+0.016
Neg logit contributions
 fault-0.017
 pret-0.015
paper-0.015
 pounds-0.014
 puck-0.014
Token the
Token ID262
Feature activation+5.30e-01

Pos logit contributions
 hurricane+0.024
 tornado+0.023
 television+0.022
date+0.022
 voice+0.022
Neg logit contributions
 fault-0.029
 pret-0.025
paper-0.025
 pounds-0.024
eter-0.024
Token wind
Token ID2344
Feature activation+1.46e+00

Pos logit contributions
date+0.031
han+0.029
iet+0.028
Friend+0.028
ind+0.028
Neg logit contributions
 fault-0.036
 pret-0.033
 puck-0.031
 cones-0.030
 loses-0.029
Token how
Token ID703
Feature activation+6.31e-02

Pos logit contributions
 hurricane+0.075
date+0.074
 tornado+0.069
 Amen+0.068
 television+0.067
Neg logit contributions
 fault-0.104
 pret-0.097
paper-0.093
 loses-0.091
 enough-0.090
Tokenling
Token ID1359
Feature activation+2.97e+00

Pos logit contributions
 hurricane+0.004
date+0.004
 tornado+0.003
 year+0.003
 television+0.003
Neg logit contributions
 fault-0.004
 pret-0.004
paper-0.004
 loses-0.003
 puck-0.003
Token outside
Token ID2354
Feature activation+1.38e+00

Pos logit contributions
 hurricane+0.195
date+0.195
 Amen+0.191
ind+0.183
 tornado+0.183
Neg logit contributions
 fault-0.159
 pret-0.146
 loses-0.138
paper-0.138
 enough-0.135
Token.
Token ID13
Feature activation+2.50e-01

Pos logit contributions
 hurricane+0.092
date+0.090
 tornado+0.087
 Amen+0.086
 television+0.085
Neg logit contributions
 fault-0.075
 pret-0.068
paper-0.065
 loses-0.063
 enough-0.062
Token They
Token ID1119
Feature activation+1.99e-01

Pos logit contributions
 hurricane+0.014
date+0.013
 tornado+0.013
 television+0.013
 voice+0.012
Neg logit contributions
 fault-0.016
 pret-0.015
 pounds-0.015
paper-0.015
 puck-0.015
Token felt
Token ID2936
Feature activation-7.11e-02

Pos logit contributions
 hurricane+0.014
date+0.013
 tornado+0.013
 television+0.013
 year+0.013
Neg logit contributions
 fault-0.011
 pret-0.010
paper-0.009
 loses-0.009
 puck-0.009
Token cold
Token ID4692
Feature activation-5.68e-01

Pos logit contributions
 fault+0.005
 pret+0.004
 loses+0.004
paper+0.004
 enough+0.004
Neg logit contributions
date-0.004
 hurricane-0.004
 tornado-0.004
 year-0.004
 television-0.004
Token woods
Token ID16479
Feature activation-3.43e-01

Pos logit contributions
 fault+0.025
 pret+0.024
paper+0.023
 puck+0.023
 cones+0.023
Neg logit contributions
date-0.014
 Amen-0.014
 messenger-0.013
 hurricane-0.013
Friend-0.013
Token and
Token ID290
Feature activation+5.44e-03

Pos logit contributions
 fault+0.020
 pret+0.018
paper+0.017
 loses+0.017
 enough+0.016
Neg logit contributions
 hurricane-0.022
date-0.022
 tornado-0.021
 television-0.020
 Amen-0.020
Token practice
Token ID3357
Feature activation+6.43e-01

Pos logit contributions
 hurricane+0.000
date+0.000
 tornado+0.000
 year+0.000
 television+0.000
Neg logit contributions
 fault-0.000
 pret-0.000
paper-0.000
 loses-0.000
 puck-0.000
Token its
Token ID663
Feature activation+3.78e-01

Pos logit contributions
 hurricane+0.036
date+0.035
 tornado+0.034
 television+0.033
 year+0.033
Neg logit contributions
 fault-0.042
 pret-0.038
paper-0.037
 loses-0.036
 puck-0.036
Token how
Token ID703
Feature activation+5.25e-02

Pos logit contributions
 hurricane+0.021
date+0.021
 tornado+0.020
 television+0.019
 year+0.019
Neg logit contributions
 fault-0.025
 pret-0.023
paper-0.022
 loses-0.022
 puck-0.022
Tokenling
Token ID1359
Feature activation+2.94e+00

Pos logit contributions
 hurricane+0.003
date+0.003
 tornado+0.003
 year+0.003
 television+0.003
Neg logit contributions
 fault-0.003
 pret-0.003
paper-0.003
 puck-0.003
 loses-0.003
Token.
Token ID13
Feature activation+3.46e-01

Pos logit contributions
date+0.194
 hurricane+0.193
 Amen+0.187
 television+0.183
 tornado+0.183
Neg logit contributions
 fault-0.159
 pret-0.147
 loses-0.139
paper-0.139
 enough-0.134
Token The
Token ID383
Feature activation+3.47e-01

Pos logit contributions
 hurricane+0.016
date+0.016
 tornado+0.015
 year+0.015
 Amen+0.014
Neg logit contributions
 fault-0.026
 pret-0.024
paper-0.024
 loses-0.024
 pounds-0.024
Token wolf
Token ID17481
Feature activation-3.90e-01

Pos logit contributions
date+0.018
ilt+0.017
han+0.017
aby+0.017
rim+0.016
Neg logit contributions
 cones-0.021
 enough-0.021
 fault-0.021
 puck-0.020
 pret-0.020
Token wanted
Token ID2227
Feature activation+2.06e-01

Pos logit contributions
 fault+0.025
 pret+0.022
paper+0.021
 pounds+0.021
 loses+0.020
Neg logit contributions
 hurricane-0.023
date-0.023
 year-0.022
 tornado-0.022
 musician-0.021
Token to
Token ID284
Feature activation-3.70e-01

Pos logit contributions
 hurricane+0.014
 tornado+0.013
date+0.013
 television+0.013
 year+0.013
Neg logit contributions
 fault-0.012
 pret-0.011
 pounds-0.010
paper-0.010
 loses-0.010
Token,
Token ID11
Feature activation+6.51e-01

Pos logit contributions
 hurricane+0.007
date+0.007
 tornado+0.006
 television+0.006
 Amen+0.006
Neg logit contributions
 fault-0.007
 pret-0.006
paper-0.006
 loses-0.006
 enough-0.006
Token the
Token ID262
Feature activation+1.62e-01

Pos logit contributions
 hurricane+0.041
date+0.041
 tornado+0.039
 television+0.038
 year+0.038
Neg logit contributions
 fault-0.038
 pret-0.034
paper-0.033
 loses-0.032
 enough-0.032
Token wolf
Token ID17481
Feature activation+3.14e-02

Pos logit contributions
date+0.010
aby+0.009
ilt+0.009
arten+0.009
ind+0.009
Neg logit contributions
 fault-0.008
 enough-0.008
 cones-0.008
 puck-0.008
 scar-0.007
Token's
Token ID338
Feature activation-5.05e-01

Pos logit contributions
 hurricane+0.002
date+0.002
 Amen+0.002
 television+0.002
 tornado+0.002
Neg logit contributions
 fault-0.002
 pret-0.002
paper-0.001
 loses-0.001
 enough-0.001
Token how
Token ID703
Feature activation+6.03e-02

Pos logit contributions
 fault+0.034
 pret+0.030
 enough+0.029
 puck+0.028
 loses+0.028
Neg logit contributions
date-0.028
 hurricane-0.028
 Amen-0.027
Friend-0.027
han-0.027
Tokenling
Token ID1359
Feature activation+2.91e+00

Pos logit contributions
 hurricane+0.004
date+0.003
 tornado+0.003
 year+0.003
 television+0.003
Neg logit contributions
 fault-0.004
 pret-0.004
paper-0.003
 loses-0.003
 puck-0.003
Token got
Token ID1392
Feature activation+5.82e-01

Pos logit contributions
 Amen+0.208
date+0.207
 hurricane+0.202
 television+0.197
ind+0.197
Neg logit contributions
 fault-0.131
 pret-0.126
 loses-0.123
 enough-0.121
 puck-0.114
Token stronger
Token ID7387
Feature activation-4.66e-01

Pos logit contributions
 hurricane+0.039
date+0.038
 tornado+0.036
 television+0.036
 year+0.036
Neg logit contributions
 fault-0.033
 pret-0.029
paper-0.028
 loses-0.027
 puck-0.027
Token and
Token ID290
Feature activation+3.68e-02

Pos logit contributions
 fault+0.029
 pret+0.027
paper+0.025
 puck+0.025
 pounds+0.025
Neg logit contributions
 hurricane-0.026
date-0.026
 tornado-0.025
 year-0.024
 voice-0.024
Token stronger
Token ID7387
Feature activation-3.94e-01

Pos logit contributions
 hurricane+0.003
 tornado+0.002
 others+0.002
 me+0.002
 voice+0.002
Neg logit contributions
 fault-0.002
 pret-0.002
 pounds-0.002
paper-0.002
ws-0.002
Token.
Token ID13
Feature activation-1.11e-02

Pos logit contributions
 fault+0.029
 pret+0.027
paper+0.025
 loses+0.025
 puck+0.025
Neg logit contributions
 hurricane-0.019
date-0.019
 tornado-0.018
 year-0.017
 television-0.017
Token too
Token ID1165
Feature activation-4.36e-01

Pos logit contributions
 fault+0.031
 pret+0.028
 enough+0.027
paper+0.026
 puck+0.026
Neg logit contributions
 hurricane-0.029
date-0.029
 tornado-0.027
 Amen-0.027
 television-0.027
Token cold
Token ID4692
Feature activation+4.04e-02

Pos logit contributions
 fault+0.029
 pret+0.026
 enough+0.025
paper+0.025
 loses+0.025
Neg logit contributions
date-0.025
 hurricane-0.024
 tornado-0.023
 Amen-0.023
 television-0.023
Token for
Token ID329
Feature activation+4.43e-01

Pos logit contributions
 hurricane+0.002
date+0.002
 tornado+0.002
 television+0.002
 year+0.002
Neg logit contributions
 fault-0.003
 pret-0.002
 puck-0.002
 loses-0.002
paper-0.002
Token a
Token ID257
Feature activation+8.83e-03

Pos logit contributions
date+0.030
 hurricane+0.029
 Amen+0.029
iet+0.029
han+0.028
Neg logit contributions
 fault-0.022
 enough-0.021
 pret-0.021
 pounds-0.019
paper-0.019
Token re
Token ID302
Feature activation+1.34e+00

Pos logit contributions
date+0.001
 hurricane+0.000
han+0.000
ind+0.000
iet+0.000
Neg logit contributions
 fault-0.001
 pret-0.001
 puck-0.000
paper-0.000
 pounds-0.000
Tokenind
Token ID521
Feature activation+2.91e+00

Pos logit contributions
 television+0.065
 hurricane+0.064
 tornado+0.061
date+0.059
Friend+0.057
Neg logit contributions
 fault-0.101
 pounds-0.095
eter-0.091
 loses-0.091
paper-0.090
Tokeneer
Token ID28153
Feature activation+4.89e-01

Pos logit contributions
 hurricane+0.188
date+0.185
 tornado+0.178
 television+0.174
 year+0.174
Neg logit contributions
 fault-0.168
 pret-0.152
paper-0.144
 loses-0.140
 puck-0.139
Token like
Token ID588
Feature activation-6.50e-01

Pos logit contributions
 hurricane+0.028
date+0.027
 tornado+0.026
 Amen+0.026
 television+0.026
Neg logit contributions
 fault-0.032
 pret-0.028
 enough-0.028
 loses-0.027
paper-0.027
Token you
Token ID345
Feature activation+5.13e-01

Pos logit contributions
 fault+0.035
 pret+0.031
 enough+0.030
paper+0.029
 puck+0.029
Neg logit contributions
date-0.044
 hurricane-0.043
 Amen-0.041
 tornado-0.041
iet-0.041
Token!"
Token ID2474
Feature activation-1.22e-01

Pos logit contributions
 hurricane+0.030
date+0.029
 tornado+0.028
 television+0.027
 Amen+0.027
Neg logit contributions
 fault-0.033
 pret-0.030
paper-0.029
 loses-0.028
 enough-0.028
Token
Token ID198
Feature activation-2.10e-01

Pos logit contributions
 fault+0.006
 pret+0.006
 Mir+0.005
 enough+0.005
 loses+0.005
Neg logit contributions
 hurricane-0.009
date-0.008
 tornado-0.008
ind-0.008
ilt-0.008
Token ogre
Token ID47656
Feature activation-3.21e-01

Pos logit contributions
date+0.008
ilt+0.007
ator+0.007
rim+0.006
han+0.006
Neg logit contributions
 puck-0.006
 enough-0.006
 scar-0.006
 bones-0.006
 pret-0.006
Token ran
Token ID4966
Feature activation-1.41e-01

Pos logit contributions
 fault+0.016
 pret+0.016
 loses+0.015
 enough+0.014
 puck+0.014
Neg logit contributions
 hurricane-0.022
date-0.021
 television-0.020
 tornado-0.020
 Amen-0.020
Token away
Token ID1497
Feature activation+3.45e-01

Pos logit contributions
 fault+0.009
 enough+0.009
 pret+0.008
paper+0.008
 puck+0.008
Neg logit contributions
 hurricane-0.008
date-0.008
 Amen-0.007
 tornado-0.007
ilt-0.007
Token,
Token ID11
Feature activation+4.99e-02

Pos logit contributions
 hurricane+0.022
date+0.022
 tornado+0.021
 Amen+0.021
 television+0.021
Neg logit contributions
 fault-0.020
 pret-0.018
 enough-0.017
paper-0.017
 loses-0.017
Token how
Token ID703
Feature activation-1.42e-01

Pos logit contributions
date+0.004
 hurricane+0.003
 tornado+0.003
iet+0.003
ilt+0.003
Neg logit contributions
 fault-0.003
 pret-0.002
 enough-0.002
 puck-0.002
 loses-0.002
Tokenling
Token ID1359
Feature activation+2.83e+00

Pos logit contributions
 fault+0.008
 pret+0.008
paper+0.007
 loses+0.007
 enough+0.007
Neg logit contributions
 hurricane-0.009
date-0.009
 tornado-0.008
 television-0.008
 year-0.008
Token in
Token ID287
Feature activation+6.65e-01

Pos logit contributions
date+0.191
 Amen+0.189
 hurricane+0.187
ind+0.186
 television+0.181
Neg logit contributions
 fault-0.137
 enough-0.130
 pret-0.128
 loses-0.124
 puck-0.120
Token pain
Token ID2356
Feature activation+3.92e-02

Pos logit contributions
 hurricane+0.045
date+0.045
 Amen+0.043
 tornado+0.042
cient+0.042
Neg logit contributions
 fault-0.036
 pret-0.032
 enough-0.031
 puck-0.029
 loses-0.029
Token.
Token ID13
Feature activation-7.84e-02

Pos logit contributions
 hurricane+0.003
date+0.003
 Amen+0.002
ind+0.002
 tornado+0.002
Neg logit contributions
 fault-0.002
 pret-0.002
 enough-0.002
paper-0.002
 loses-0.002
Token
Token ID220
Feature activation+3.91e-01

Pos logit contributions
 fault+0.005
 pret+0.004
paper+0.004
 loses+0.004
 puck+0.004
Neg logit contributions
 hurricane-0.005
date-0.005
 tornado-0.005
 television-0.004
ind-0.004
Token
Token ID198
Feature activation+5.70e-02

Pos logit contributions
 hurricane+0.021
date+0.020
 tornado+0.019
 television+0.019
 year+0.019
Neg logit contributions
 fault-0.027
 pret-0.025
paper-0.024
 loses-0.023
 puck-0.023
Token,
Token ID11
Feature activation-1.17e-02

Pos logit contributions
 hurricane+0.011
date+0.011
 tornado+0.011
 year+0.010
 television+0.010
Neg logit contributions
 fault-0.014
 pret-0.012
paper-0.012
 loses-0.012
 puck-0.012
Token the
Token ID262
Feature activation-1.13e-02

Pos logit contributions
 fault+0.001
 pret+0.001
 enough+0.001
paper+0.001
 puck+0.001
Neg logit contributions
 hurricane-0.001
date-0.001
 Amen-0.001
 tornado-0.001
ind-0.001
Token wolf
Token ID17481
Feature activation-4.94e-01

Pos logit contributions
 enough+0.001
 fault+0.001
 cones+0.001
 snowball+0.001
 bones+0.000
Neg logit contributions
date-0.001
ilt-0.001
arten-0.001
aby-0.001
han-0.001
Token tried
Token ID3088
Feature activation+5.39e-01

Pos logit contributions
 fault+0.025
 pret+0.024
 enough+0.023
 loses+0.023
 puck+0.022
Neg logit contributions
 hurricane-0.034
date-0.033
 Amen-0.032
 television-0.032
 tornado-0.032
Token how
Token ID703
Feature activation-1.14e-01

Pos logit contributions
 hurricane+0.036
 television+0.035
 tornado+0.034
date+0.034
 voice+0.033
Neg logit contributions
 fault-0.028
 pret-0.026
paper-0.025
 pounds-0.024
 loses-0.024
Tokenling
Token ID1359
Feature activation+2.82e+00

Pos logit contributions
 fault+0.006
 pret+0.006
paper+0.006
 puck+0.005
 pounds+0.005
Neg logit contributions
 hurricane-0.008
 tornado-0.007
date-0.007
 year-0.007
 television-0.007
Token again
Token ID757
Feature activation-3.41e-01

Pos logit contributions
 hurricane+0.180
date+0.179
 Amen+0.173
 tornado+0.170
ind+0.169
Neg logit contributions
 fault-0.159
 pret-0.147
paper-0.138
 loses-0.138
 enough-0.136
Token in
Token ID287
Feature activation+7.04e-01

Pos logit contributions
 fault+0.023
 pret+0.021
paper+0.020
 loses+0.020
 puck+0.020
Neg logit contributions
 hurricane-0.019
date-0.018
 tornado-0.017
 television-0.017
 year-0.017
Token the
Token ID262
Feature activation-6.33e-03

Pos logit contributions
date+0.046
 hurricane+0.046
ind+0.045
iet+0.044
 Amen+0.044
Neg logit contributions
 fault-0.040
 pret-0.035
 enough-0.035
 loses-0.032
 puck-0.032
Token clearing
Token ID17304
Feature activation+4.15e-01

Pos logit contributions
 enough+0.000
 fault+0.000
 puck+0.000
 harder+0.000
 pret+0.000
Neg logit contributions
aby-0.000
ator-0.000
date-0.000
han-0.000
lf-0.000
Token.
Token ID13
Feature activation-1.06e-01

Pos logit contributions
 hurricane+0.026
date+0.025
 Amen+0.025
ind+0.025
 tornado+0.024
Neg logit contributions
 fault-0.025
 pret-0.022
 enough-0.022
 loses-0.021
paper-0.021
Tokenâ
Token ID22940
Feature activation-4.93e-03

Pos logit contributions
 fault+0.013
 pret+0.012
 loses+0.012
 enough+0.012
paper+0.012
Neg logit contributions
 hurricane-0.018
date-0.017
 Amen-0.017
 television-0.017
 tornado-0.017
Token
Token ID26391
Feature activation-8.32e-02

Pos logit contributions
 fault+0.000
 pret+0.000
paper+0.000
 loses+0.000
 enough+0.000
Neg logit contributions
 hurricane-0.000
date-0.000
 tornado-0.000
 television-0.000
 Amen-0.000
Token
Token ID8151
Feature activation-1.94e+00

Pos logit contributions
 fault+0.005
 pret+0.005
paper+0.004
 loses+0.004
 puck+0.004
Neg logit contributions
 hurricane-0.005
date-0.005
 tornado-0.005
 television-0.005
 Amen-0.005
Tokens
Token ID82
Feature activation-4.16e-01

Pos logit contributions
 fault+0.112
 enough+0.093
paper+0.093
 loses+0.091
 puck+0.090
Neg logit contributions
 hurricane-0.130
 tornado-0.123
date-0.122
 television-0.122
 Amen-0.122
Token how
Token ID703
Feature activation-2.77e-01

Pos logit contributions
 fault+0.027
 pret+0.023
 enough+0.023
paper+0.022
 loses+0.022
Neg logit contributions
date-0.025
 hurricane-0.024
 Amen-0.024
han-0.023
ind-0.023
Tokenling
Token ID1359
Feature activation+2.78e+00

Pos logit contributions
 fault+0.017
 pret+0.015
paper+0.014
 loses+0.014
 enough+0.014
Neg logit contributions
 hurricane-0.017
date-0.017
 tornado-0.016
 television-0.016
 Amen-0.016
Token was
Token ID373
Feature activation+2.80e-01

Pos logit contributions
 Amen+0.196
date+0.194
 hurricane+0.189
 television+0.185
ind+0.184
Neg logit contributions
 fault-0.129
 pret-0.126
 loses-0.123
 enough-0.121
 puck-0.115
Token so
Token ID523
Feature activation+3.17e-01

Pos logit contributions
 hurricane+0.018
date+0.018
 tornado+0.017
 Amen+0.017
ind+0.017
Neg logit contributions
 fault-0.016
 pret-0.014
paper-0.013
 loses-0.013
 enough-0.013
Token loud
Token ID7812
Feature activation+7.53e-01

Pos logit contributions
date+0.018
 hurricane+0.018
 tornado+0.017
 television+0.017
 Amen+0.017
Neg logit contributions
 fault-0.021
 pret-0.019
paper-0.018
 loses-0.018
 enough-0.018
Token it
Token ID340
Feature activation+3.99e-02

Pos logit contributions
 hurricane+0.043
date+0.042
 tornado+0.040
 Amen+0.039
 year+0.039
Neg logit contributions
 fault-0.049
 pret-0.045
paper-0.043
 loses-0.042
 enough-0.042
Token made
Token ID925
Feature activation-4.51e-01

Pos logit contributions
 hurricane+0.003
date+0.003
 tornado+0.003
 Amen+0.003
 television+0.003
Neg logit contributions
 fault-0.002
 pret-0.002
paper-0.002
 loses-0.002
 enough-0.002
Token deep
Token ID2769
Feature activation+4.82e-01

Pos logit contributions
 fault+0.006
 pret+0.006
 puck+0.005
 loses+0.005
paper+0.005
Neg logit contributions
date-0.005
ind-0.005
ator-0.005
 hurricane-0.005
 Amen-0.005
Token breath
Token ID8033
Feature activation+5.51e-01

Pos logit contributions
 hurricane+0.029
date+0.029
 tornado+0.027
 television+0.027
ind+0.027
Neg logit contributions
 fault-0.029
 pret-0.027
paper-0.025
 enough-0.025
 puck-0.025
Token and
Token ID290
Feature activation-5.13e-01

Pos logit contributions
 hurricane+0.041
date+0.040
 tornado+0.039
 Amen+0.038
 television+0.038
Neg logit contributions
 fault-0.027
 pret-0.023
 enough-0.023
paper-0.022
 loses-0.021
Token started
Token ID2067
Feature activation+2.63e-01

Pos logit contributions
 fault+0.033
 pret+0.029
paper+0.028
 pounds+0.028
 loses+0.027
Neg logit contributions
 hurricane-0.030
date-0.029
 tornado-0.028
 year-0.028
 television-0.027
Token how
Token ID703
Feature activation-1.41e-01

Pos logit contributions
 hurricane+0.019
date+0.019
 tornado+0.018
 year+0.018
 Amen+0.018
Neg logit contributions
 fault-0.013
 pret-0.012
paper-0.011
 loses-0.011
 enough-0.011
Tokenling
Token ID1359
Feature activation+2.78e+00

Pos logit contributions
 fault+0.010
 pret+0.009
paper+0.009
 loses+0.008
 puck+0.008
Neg logit contributions
 hurricane-0.007
date-0.007
 tornado-0.007
 television-0.007
 year-0.007
Token.
Token ID13
Feature activation+1.01e-01

Pos logit contributions
 hurricane+0.179
date+0.178
 Amen+0.171
 tornado+0.169
 television+0.167
Neg logit contributions
 fault-0.154
 pret-0.144
paper-0.137
 loses-0.135
 enough-0.135
Token But
Token ID887
Feature activation+3.71e-01

Pos logit contributions
 hurricane+0.006
date+0.006
 tornado+0.006
 television+0.006
 year+0.005
Neg logit contributions
 fault-0.006
 pret-0.006
paper-0.006
 loses-0.005
 pounds-0.005
Token the
Token ID262
Feature activation+1.85e-01

Pos logit contributions
 hurricane+0.022
date+0.021
 tornado+0.021
 year+0.020
 television+0.020
Neg logit contributions
 fault-0.024
 pret-0.022
paper-0.021
 pounds-0.020
 loses-0.020
Token wolf
Token ID17481
Feature activation-3.22e-01

Pos logit contributions
date+0.013
ilt+0.012
ator+0.012
ind+0.012
wered+0.012
Neg logit contributions
 fault-0.009
 puck-0.009
 enough-0.008
 cones-0.008
 pret-0.008
Tokenâ
Token ID22940
Feature activation-1.98e-02

Pos logit contributions
 fault+0.017
 pret+0.015
 loses+0.014
paper+0.014
 enough+0.014
Neg logit contributions
 hurricane-0.023
date-0.022
 Amen-0.022
 television-0.022
 tornado-0.021
Token and
Token ID290
Feature activation-3.35e-01

Pos logit contributions
 fault+0.055
 pret+0.050
paper+0.047
 loses+0.046
 puck+0.046
Neg logit contributions
 hurricane-0.059
date-0.058
 tornado-0.056
 television-0.055
 year-0.055
Token the
Token ID262
Feature activation+2.05e-01

Pos logit contributions
 fault+0.021
 pret+0.019
paper+0.018
 loses+0.018
 puck+0.018
Neg logit contributions
 hurricane-0.020
date-0.019
 tornado-0.019
 year-0.018
 television-0.018
Token wind
Token ID2344
Feature activation+9.19e-01

Pos logit contributions
date+0.011
han+0.010
rim+0.010
ilt+0.010
iet+0.009
Neg logit contributions
 fault-0.014
 enough-0.013
 pret-0.013
paper-0.013
 loses-0.013
Token was
Token ID373
Feature activation-1.00e-01

Pos logit contributions
date+0.065
 hurricane+0.064
 Amen+0.064
 tornado+0.061
 television+0.061
Neg logit contributions
 fault-0.045
 pret-0.042
paper-0.041
 loses-0.039
 enough-0.038
Token how
Token ID703
Feature activation-2.12e-01

Pos logit contributions
 fault+0.006
 enough+0.006
 pret+0.006
paper+0.005
 puck+0.005
Neg logit contributions
date-0.006
 hurricane-0.006
Friend-0.006
 Amen-0.006
han-0.005
Tokenling
Token ID1359
Feature activation+2.74e+00

Pos logit contributions
 fault+0.015
 pret+0.014
paper+0.013
 loses+0.013
 puck+0.013
Neg logit contributions
 hurricane-0.011
date-0.010
 tornado-0.010
 year-0.010
 television-0.010
Token.
Token ID13
Feature activation+2.20e-01

Pos logit contributions
date+0.180
 hurricane+0.178
 Amen+0.171
 tornado+0.170
 television+0.169
Neg logit contributions
 fault-0.148
 pret-0.137
paper-0.130
 loses-0.129
 enough-0.127
Token The
Token ID383
Feature activation+3.56e-01

Pos logit contributions
 hurricane+0.009
date+0.009
 year+0.009
 voice+0.009
 Everyone+0.009
Neg logit contributions
 fault-0.017
 pret-0.016
paper-0.016
 pounds-0.015
 loses-0.015
Token pilot
Token ID8022
Feature activation-9.43e-01

Pos logit contributions
date+0.019
han+0.016
ilt+0.016
rim+0.016
iet+0.015
Neg logit contributions
 fault-0.025
 pret-0.024
 enough-0.024
paper-0.023
 loses-0.023
Token was
Token ID373
Feature activation-2.34e-01

Pos logit contributions
 fault+0.057
 pret+0.052
paper+0.049
 loses+0.048
 puck+0.047
Neg logit contributions
 hurricane-0.058
date-0.057
 tornado-0.055
 television-0.054
 year-0.053
Token so
Token ID523
Feature activation+1.28e-01

Pos logit contributions
 fault+0.016
 pret+0.014
 enough+0.014
 loses+0.014
paper+0.014
Neg logit contributions
 hurricane-0.013
date-0.013
 tornado-0.012
 television-0.012
 year-0.012
Token work
Token ID670
Feature activation-1.09e-01

Pos logit contributions
date+0.047
ind+0.045
han+0.045
iet+0.044
 hurricane+0.043
Neg logit contributions
 fault-0.036
 pret-0.034
 loses-0.032
 enough-0.031
 puck-0.031
Token on
Token ID319
Feature activation+3.54e-01

Pos logit contributions
 fault+0.006
 pret+0.006
paper+0.005
 loses+0.005
 puck+0.005
Neg logit contributions
 hurricane-0.007
date-0.007
 tornado-0.007
 television-0.006
 year-0.006
Token improving
Token ID10068
Feature activation-7.11e-02

Pos logit contributions
date+0.024
 hurricane+0.024
 Amen+0.023
iet+0.022
han+0.022
Neg logit contributions
 fault-0.020
 pret-0.017
 enough-0.017
paper-0.016
 loses-0.016
Token its
Token ID663
Feature activation-1.66e-01

Pos logit contributions
 fault+0.004
 pret+0.004
paper+0.003
 enough+0.003
 loses+0.003
Neg logit contributions
date-0.005
 hurricane-0.005
 Amen-0.004
 tornado-0.004
 year-0.004
Token how
Token ID703
Feature activation-3.65e-01

Pos logit contributions
 fault+0.011
 pret+0.010
 loses+0.009
 enough+0.009
paper+0.009
Neg logit contributions
date-0.009
 hurricane-0.009
 tornado-0.009
 Amen-0.009
han-0.009
Tokenling
Token ID1359
Feature activation+2.72e+00

Pos logit contributions
 fault+0.021
 pret+0.019
paper+0.018
 loses+0.018
 enough+0.018
Neg logit contributions
date-0.024
 hurricane-0.023
 Amen-0.022
 television-0.022
 tornado-0.022
Token.
Token ID13
Feature activation+4.98e-02

Pos logit contributions
date+0.179
 hurricane+0.175
 Amen+0.175
ind+0.168
 tornado+0.166
Neg logit contributions
 fault-0.149
 pret-0.138
 loses-0.131
paper-0.130
 enough-0.128
Token It
Token ID632
Feature activation-5.01e-01

Pos logit contributions
 hurricane+0.003
date+0.003
 tornado+0.003
 year+0.003
 television+0.003
Neg logit contributions
 fault-0.003
 pret-0.003
paper-0.003
 loses-0.003
 pounds-0.003
Token practiced
Token ID19893
Feature activation+6.34e-01

Pos logit contributions
 fault+0.031
 pret+0.027
paper+0.027
 pounds+0.026
 loses+0.026
Neg logit contributions
 hurricane-0.031
date-0.030
 tornado-0.029
 year-0.028
 voice-0.028
Token in
Token ID287
Feature activation+4.39e-01

Pos logit contributions
 hurricane+0.040
date+0.039
 tornado+0.038
 television+0.038
 year+0.037
Neg logit contributions
 fault-0.037
 pret-0.033
paper-0.032
 loses-0.031
 puck-0.031
Token the
Token ID262
Feature activation-2.12e-01

Pos logit contributions
 hurricane+0.027
date+0.027
 tornado+0.026
 Amen+0.025
 television+0.025
Neg logit contributions
 fault-0.027
 pret-0.024
paper-0.022
 loses-0.022
 enough-0.022
Token to
Token ID284
Feature activation-3.67e-01

Pos logit contributions
 hurricane+0.009
date+0.009
 tornado+0.009
 year+0.008
 television+0.008
Neg logit contributions
 fault-0.006
 pret-0.006
paper-0.005
 loses-0.005
 puck-0.005
Token show
Token ID905
Feature activation-1.38e-03

Pos logit contributions
 pret+0.017
 fault+0.017
 loses+0.016
 puck+0.016
 puts+0.016
Neg logit contributions
ind-0.026
iet-0.025
 hurricane-0.025
 tornado-0.024
 Amen-0.024
Token off
Token ID572
Feature activation-2.26e-01

Pos logit contributions
 fault+0.000
 enough+0.000
 pret+0.000
 loses+0.000
paper+0.000
Neg logit contributions
date-0.000
 hurricane-0.000
ilt-0.000
 Amen-0.000
ind-0.000
Token its
Token ID663
Feature activation-1.69e-01

Pos logit contributions
 fault+0.011
 enough+0.010
 pret+0.009
paper+0.009
 loses+0.009
Neg logit contributions
date-0.017
 hurricane-0.017
ilt-0.016
 Amen-0.016
 tornado-0.016
Token how
Token ID703
Feature activation-2.09e-01

Pos logit contributions
 fault+0.011
 pret+0.010
 teeth+0.009
 enough+0.009
 puck+0.009
Neg logit contributions
date-0.010
 hurricane-0.010
 Amen-0.009
han-0.009
 tornado-0.009
Tokenling
Token ID1359
Feature activation+2.72e+00

Pos logit contributions
 fault+0.011
 enough+0.011
 loses+0.011
paper+0.010
 pounds+0.010
Neg logit contributions
date-0.014
 hurricane-0.013
 television-0.013
ator-0.013
 tornado-0.013
Token and
Token ID290
Feature activation+2.14e-01

Pos logit contributions
date+0.185
 hurricane+0.183
 Amen+0.180
ind+0.175
 tornado+0.174
Neg logit contributions
 fault-0.135
 pret-0.132
 loses-0.124
 enough-0.124
paper-0.121
Token make
Token ID787
Feature activation+4.00e-01

Pos logit contributions
date+0.013
 hurricane+0.013
 tornado+0.012
iet+0.012
ind+0.012
Neg logit contributions
 fault-0.013
 pret-0.012
 enough-0.011
 loses-0.011
paper-0.011
Token the
Token ID262
Feature activation+3.64e-01

Pos logit contributions
 hurricane+0.027
date+0.026
 tornado+0.025
 year+0.025
 television+0.024
Neg logit contributions
 fault-0.022
 pret-0.020
paper-0.019
 loses-0.019
 puck-0.018
Token rabbit
Token ID22746
Feature activation+7.56e-01

Pos logit contributions
date+0.021
han+0.021
aby+0.020
 hurricane+0.020
 Amen+0.020
Neg logit contributions
 fault-0.022
 pret-0.020
 puck-0.020
 enough-0.020
 cones-0.019
Token scared
Token ID12008
Feature activation+1.45e-01

Pos logit contributions
 hurricane+0.045
date+0.044
 tornado+0.043
 television+0.042
 year+0.041
Neg logit contributions
 fault-0.047
 pret-0.043
paper-0.040
 loses-0.040
 puck-0.039
Token,
Token ID11
Feature activation+1.19e-01

Pos logit contributions
 fault+0.029
 pret+0.026
paper+0.025
 loses+0.024
 puck+0.024
Neg logit contributions
 hurricane-0.026
date-0.026
 tornado-0.025
 television-0.024
 year-0.024
Token the
Token ID262
Feature activation+1.26e-02

Pos logit contributions
date+0.008
 hurricane+0.008
ind+0.007
 Amen+0.007
 year+0.007
Neg logit contributions
 fault-0.007
 pret-0.006
 enough-0.006
paper-0.006
 loses-0.006
Token pup
Token ID15552
Feature activation+3.09e-01

Pos logit contributions
date+0.001
 hurricane+0.001
 Amen+0.001
ator+0.001
cient+0.001
Neg logit contributions
 fault-0.001
 pret-0.001
 puck-0.001
 cones-0.001
 enough-0.001
Token started
Token ID2067
Feature activation-1.37e-01

Pos logit contributions
 hurricane+0.022
 Amen+0.022
date+0.021
 television+0.021
 tornado+0.021
Neg logit contributions
 fault-0.015
 pret-0.014
 loses-0.014
 enough-0.014
 puck-0.014
Token how
Token ID703
Feature activation-2.66e-01

Pos logit contributions
 fault+0.007
 pret+0.006
paper+0.005
 loses+0.005
 puck+0.005
Neg logit contributions
 hurricane-0.010
date-0.010
 tornado-0.010
 television-0.010
 year-0.010
Tokenling
Token ID1359
Feature activation+2.70e+00

Pos logit contributions
 fault+0.017
 pret+0.016
paper+0.015
 loses+0.015
 puck+0.015
Neg logit contributions
 hurricane-0.015
date-0.015
 tornado-0.014
 television-0.014
 year-0.014
Token!
Token ID0
Feature activation-1.19e-01

Pos logit contributions
 Amen+0.183
 hurricane+0.180
date+0.176
ind+0.172
 television+0.170
Neg logit contributions
 fault-0.136
 pret-0.132
 enough-0.128
 loses-0.127
paper-0.125
Token It
Token ID632
Feature activation-7.53e-01

Pos logit contributions
 fault+0.007
 pret+0.007
paper+0.006
 puck+0.006
 loses+0.006
Neg logit contributions
 hurricane-0.007
date-0.007
 tornado-0.007
ind-0.007
 Amen-0.007
Token was
Token ID373
Feature activation-2.17e-01

Pos logit contributions
 fault+0.038
 pret+0.036
 loses+0.034
paper+0.034
 puck+0.033
Neg logit contributions
 hurricane-0.052
date-0.051
 tornado-0.049
 Amen-0.049
 television-0.049
Token very
Token ID845
Feature activation+5.60e-01

Pos logit contributions
 fault+0.013
 pret+0.012
paper+0.011
 loses+0.011
 puck+0.011
Neg logit contributions
 hurricane-0.013
date-0.013
 tornado-0.013
 television-0.012
 year-0.012
Token
Token ID220
Feature activation+1.42e-01

Pos logit contributions
date+0.035
 hurricane+0.033
 Amen+0.032
 tornado+0.032
 television+0.031
Neg logit contributions
 fault-0.035
 pret-0.031
 loses-0.031
 puck-0.031
 enough-0.031
Token and
Token ID290
Feature activation+4.60e-01

Pos logit contributions
 hurricane+0.023
date+0.023
 tornado+0.022
 Amen+0.022
ind+0.021
Neg logit contributions
 fault-0.018
 enough-0.016
 pret-0.016
 puck-0.015
paper-0.015
Token all
Token ID477
Feature activation+1.18e-01

Pos logit contributions
 hurricane+0.027
date+0.026
 tornado+0.025
 year+0.024
 television+0.024
Neg logit contributions
 fault-0.030
 pret-0.027
paper-0.026
 loses-0.025
 pounds-0.025
Token the
Token ID262
Feature activation+8.04e-01

Pos logit contributions
date+0.008
 hurricane+0.008
 tornado+0.008
 television+0.008
ind+0.007
Neg logit contributions
 fault-0.006
 enough-0.006
 pret-0.006
 puck-0.005
 loses-0.005
Token other
Token ID584
Feature activation+2.78e-01

Pos logit contributions
date+0.049
 hurricane+0.043
han+0.043
 tornado+0.042
iet+0.042
Neg logit contributions
 fault-0.051
 pret-0.049
 enough-0.046
 pounds-0.045
 cones-0.045
Token re
Token ID302
Feature activation+1.33e+00

Pos logit contributions
date+0.015
 hurricane+0.015
 tornado+0.014
 television+0.013
 Amen+0.013
Neg logit contributions
 fault-0.019
 pret-0.018
 enough-0.017
 pounds-0.017
paper-0.017
Tokenind
Token ID521
Feature activation+2.67e+00

Pos logit contributions
 hurricane+0.034
 tornado+0.032
date+0.030
 television+0.030
 lens+0.025
Neg logit contributions
 fault-0.131
 pounds-0.122
 pret-0.121
eter-0.120
paper-0.119
Tokeneers
Token ID47619
Feature activation+1.02e+00

Pos logit contributions
 hurricane+0.177
date+0.175
 tornado+0.168
 television+0.164
 year+0.161
Neg logit contributions
 fault-0.150
 pret-0.136
paper-0.128
 loses-0.125
 pounds-0.124
Token.
Token ID13
Feature activation+5.33e-02

Pos logit contributions
 hurricane+0.063
date+0.061
 tornado+0.060
ind+0.058
 television+0.058
Neg logit contributions
 fault-0.060
 pret-0.055
 enough-0.053
paper-0.052
 loses-0.052
Token
Token ID198
Feature activation+2.02e-01

Pos logit contributions
 hurricane+0.003
date+0.003
 tornado+0.003
 television+0.003
 year+0.003
Neg logit contributions
 fault-0.003
 pret-0.003
paper-0.003
 loses-0.003
 puck-0.003
Token
Token ID198
Feature activation+5.31e-02

Pos logit contributions
 hurricane+0.012
date+0.011
 tornado+0.011
 television+0.011
 lens+0.010
Neg logit contributions
 fault-0.013
 pret-0.012
 puck-0.011
 Mir-0.011
 loses-0.011
TokenThe
Token ID464
Feature activation+3.81e-01

Pos logit contributions
 hurricane+0.003
 television+0.003
 tornado+0.003
date+0.003
 lens+0.003
Neg logit contributions
 fault-0.003
 pret-0.003
 puck-0.003
 pounds-0.003
 loses-0.003
Token not
Token ID407
Feature activation+4.28e-01

Pos logit contributions
 fault+0.006
 pret+0.006
paper+0.005
 loses+0.005
 puck+0.005
Neg logit contributions
 hurricane-0.005
date-0.005
 tornado-0.005
 television-0.005
 year-0.005
Token know
Token ID760
Feature activation-3.27e-01

Pos logit contributions
 hurricane+0.029
date+0.028
 year+0.027
 voice+0.027
 musician+0.027
Neg logit contributions
 fault-0.024
 pret-0.020
 pounds-0.020
eter-0.020
paper-0.020
Token what
Token ID644
Feature activation+4.38e-01

Pos logit contributions
 fault+0.021
 pret+0.019
eter+0.019
 puck+0.019
ws+0.018
Neg logit contributions
 hurricane-0.019
 television-0.018
 tornado-0.018
 year-0.017
 others-0.017
Token a
Token ID257
Feature activation+2.30e-01

Pos logit contributions
 hurricane+0.027
date+0.026
 tornado+0.026
 year+0.025
 television+0.025
Neg logit contributions
 fault-0.027
 pret-0.023
paper-0.023
eter-0.023
 puck-0.022
Token re
Token ID302
Feature activation+1.33e+00

Pos logit contributions
date+0.013
 hurricane+0.013
han+0.013
ilt+0.012
wered+0.012
Neg logit contributions
 pret-0.015
 fault-0.014
 enough-0.013
 pounds-0.013
paper-0.013
Tokenind
Token ID521
Feature activation+2.67e+00

Pos logit contributions
 hurricane+0.039
date+0.037
 tornado+0.034
 television+0.032
Friend+0.028
Neg logit contributions
 fault-0.128
 pret-0.121
 pounds-0.120
paper-0.119
 loses-0.116
Tokeneer
Token ID28153
Feature activation+4.94e-01

Pos logit contributions
 hurricane+0.174
date+0.171
 tornado+0.164
 television+0.161
 year+0.160
Neg logit contributions
 fault-0.153
 pret-0.138
paper-0.130
 loses-0.127
 puck-0.126
Token is
Token ID318
Feature activation-7.40e-02

Pos logit contributions
 hurricane+0.031
date+0.030
 tornado+0.029
 television+0.028
 year+0.028
Neg logit contributions
 fault-0.029
 pret-0.027
paper-0.025
 loses-0.025
 puck-0.024
Token.
Token ID13
Feature activation+2.14e-01

Pos logit contributions
 fault+0.005
 pret+0.005
paper+0.005
 puck+0.005
 loses+0.005
Neg logit contributions
 hurricane-0.004
date-0.004
 tornado-0.003
 television-0.003
 year-0.003
Token They
Token ID1119
Feature activation+3.70e-01

Pos logit contributions
 hurricane+0.014
date+0.014
 tornado+0.013
 television+0.013
 year+0.013
Neg logit contributions
 fault-0.012
 pret-0.011
paper-0.010
 loses-0.010
 puck-0.010
Token ask
Token ID1265
Feature activation+1.11e-01

Pos logit contributions
date+0.029
 hurricane+0.027
 tornado+0.026
ind+0.026
 year+0.026
Neg logit contributions
 pret-0.016
 fault-0.016
 loses-0.016
 hides-0.015
 pulls-0.014
Tokenâ
Token ID22940
Feature activation-1.98e-02

Pos logit contributions
 fault+0.017
 pret+0.015
 loses+0.014
paper+0.014
 enough+0.014
Neg logit contributions
 hurricane-0.023
date-0.022
 Amen-0.022
 television-0.022
 tornado-0.021
Token
Token ID26391
Feature activation-1.93e-01

Pos logit contributions
 fault+0.001
 pret+0.001
paper+0.001
 loses+0.001
 enough+0.001
Neg logit contributions
 hurricane-0.001
date-0.001
 tornado-0.001
 television-0.001
 year-0.001
Token
Token ID8151
Feature activation-1.84e+00

Pos logit contributions
 fault+0.013
 pret+0.012
paper+0.011
 loses+0.011
 puck+0.011
Neg logit contributions
 hurricane-0.011
date-0.011
 tornado-0.010
 year-0.010
 television-0.010
Tokens
Token ID82
Feature activation-4.90e-01

Pos logit contributions
 fault+0.113
paper+0.095
 pret+0.094
 enough+0.093
 loses+0.093
Neg logit contributions
 hurricane-0.116
date-0.110
 tornado-0.110
 Amen-0.108
 television-0.108
Token how
Token ID703
Feature activation-4.70e-01

Pos logit contributions
 fault+0.031
 enough+0.026
 pret+0.026
 loses+0.025
paper+0.025
Neg logit contributions
date-0.031
 hurricane-0.030
 Amen-0.029
Friend-0.028
 television-0.028
Tokenling
Token ID1359
Feature activation+2.65e+00

Pos logit contributions
 fault+0.029
 pret+0.026
 loses+0.025
paper+0.025
 enough+0.025
Neg logit contributions
date-0.028
 hurricane-0.028
 television-0.027
 tornado-0.026
 Amen-0.026
Token was
Token ID373
Feature activation+1.61e-01

Pos logit contributions
 Amen+0.183
date+0.182
 hurricane+0.182
 television+0.175
 tornado+0.173
Neg logit contributions
 fault-0.126
 pret-0.124
 loses-0.119
 enough-0.117
 puck-0.113
Token not
Token ID407
Feature activation+3.69e-01

Pos logit contributions
 hurricane+0.010
date+0.010
 Amen+0.010
 tornado+0.010
 television+0.010
Neg logit contributions
 fault-0.009
 pret-0.008
 loses-0.008
 enough-0.008
paper-0.008
Token very
Token ID845
Feature activation+8.61e-01

Pos logit contributions
 hurricane+0.023
date+0.022
 tornado+0.022
 year+0.021
 television+0.021
Neg logit contributions
 fault-0.022
 pret-0.020
paper-0.019
 loses-0.019
 puck-0.019
Token loud
Token ID7812
Feature activation+6.31e-02

Pos logit contributions
date+0.050
 hurricane+0.048
 Amen+0.046
 tornado+0.046
 television+0.045
Neg logit contributions
 fault-0.057
 pret-0.051
 loses-0.050
paper-0.049
 enough-0.049
Token.
Token ID13
Feature activation-2.36e-01

Pos logit contributions
 hurricane+0.003
date+0.003
 tornado+0.003
 television+0.003
 year+0.003
Neg logit contributions
 fault-0.004
 pret-0.004
paper-0.004
 puck-0.004
 loses-0.004
Token a
Token ID257
Feature activation-1.53e-01

Pos logit contributions
 hurricane+0.005
date+0.005
 tornado+0.005
 television+0.005
 year+0.005
Neg logit contributions
 fault-0.005
 pret-0.004
paper-0.004
 loses-0.004
 puck-0.004
Token picture
Token ID4286
Feature activation-1.02e+00

Pos logit contributions
 pret+0.009
 fault+0.009
 puck+0.009
paper+0.008
 scar+0.008
Neg logit contributions
date-0.009
 hurricane-0.009
han-0.009
wered-0.008
ind-0.008
Token of
Token ID286
Feature activation+2.36e-01

Pos logit contributions
 fault+0.066
 pret+0.061
paper+0.059
 enough+0.058
 loses+0.057
Neg logit contributions
 hurricane-0.058
date-0.056
 tornado-0.053
 Amen-0.052
ind-0.052
Token a
Token ID257
Feature activation-1.18e-01

Pos logit contributions
date+0.017
 hurricane+0.017
 Amen+0.016
 year+0.016
 tornado+0.016
Neg logit contributions
 fault-0.012
 pret-0.011
 enough-0.010
paper-0.010
 puck-0.010
Token re
Token ID302
Feature activation+1.27e+00

Pos logit contributions
 pret+0.007
 fault+0.007
 puck+0.007
paper+0.006
 cones+0.006
Neg logit contributions
han-0.007
wered-0.007
date-0.007
iet-0.007
ilt-0.006
Tokenind
Token ID521
Feature activation+2.65e+00

Pos logit contributions
 hurricane+0.039
 tornado+0.035
date+0.033
 television+0.030
wered+0.028
Neg logit contributions
 fault-0.119
 pret-0.115
paper-0.114
 pounds-0.113
eter-0.112
Tokeneer
Token ID28153
Feature activation+4.44e-01

Pos logit contributions
 hurricane+0.173
date+0.170
 tornado+0.164
 television+0.159
 year+0.159
Neg logit contributions
 fault-0.151
 pret-0.137
paper-0.129
 loses-0.126
 puck-0.124
Token.
Token ID13
Feature activation+1.03e-01

Pos logit contributions
 hurricane+0.026
date+0.024
 tornado+0.023
han+0.023
 Amen+0.023
Neg logit contributions
 fault-0.029
 pret-0.027
 enough-0.026
paper-0.025
 loses-0.025
Token She
Token ID1375
Feature activation+1.03e-01

Pos logit contributions
 hurricane+0.006
date+0.006
 tornado+0.006
 year+0.006
ind+0.006
Neg logit contributions
 fault-0.006
 pret-0.006
paper-0.005
 puck-0.005
 loses-0.005
Token wanted
Token ID2227
Feature activation+3.44e-01

Pos logit contributions
 hurricane+0.006
date+0.006
 year+0.005
 television+0.005
 tornado+0.005
Neg logit contributions
 fault-0.007
paper-0.006
 pounds-0.006
 pret-0.006
eter-0.006
Token a
Token ID257
Feature activation-3.20e-01

Pos logit contributions
 hurricane+0.022
date+0.022
 television+0.022
 year+0.021
 tornado+0.021
Neg logit contributions
 fault-0.019
 pret-0.017
 pounds-0.016
 puck-0.016
paper-0.016
Token,
Token ID11
Feature activation+3.51e-01

Pos logit contributions
 hurricane+0.016
date+0.016
 tornado+0.016
 year+0.015
 television+0.015
Neg logit contributions
 fault-0.018
 pret-0.016
paper-0.016
 loses-0.015
 puck-0.015
Token but
Token ID475
Feature activation+4.55e-01

Pos logit contributions
 hurricane+0.026
date+0.025
 tornado+0.025
 television+0.024
 year+0.024
Neg logit contributions
 fault-0.017
 pret-0.015
paper-0.014
 loses-0.014
 puck-0.014
Token it
Token ID340
Feature activation+1.68e-02

Pos logit contributions
 hurricane+0.027
 tornado+0.025
date+0.025
 television+0.024
 year+0.024
Neg logit contributions
 fault-0.030
 pret-0.027
paper-0.026
 pounds-0.025
 loses-0.025
Token also
Token ID635
Feature activation+1.60e-01

Pos logit contributions
 hurricane+0.001
 year+0.001
 tornado+0.001
date+0.001
 voice+0.001
Neg logit contributions
 fault-0.001
 pounds-0.001
paper-0.001
 pret-0.001
 puck-0.001
Token so
Token ID523
Feature activation+6.72e-01

Pos logit contributions
 hurricane+0.009
 tornado+0.008
 voice+0.008
date+0.008
 year+0.008
Neg logit contributions
 fault-0.010
paper-0.010
 pret-0.010
 pounds-0.009
 puck-0.009
Tokenot
Token ID313
Feature activation+2.64e+00

Pos logit contributions
 hurricane+0.041
date+0.039
 tornado+0.038
 year+0.038
 television+0.037
Neg logit contributions
 fault-0.042
 pret-0.038
paper-0.036
 puck-0.035
 loses-0.035
Tokenhed
Token ID704
Feature activation+7.82e-01

Pos logit contributions
date+0.163
 hurricane+0.161
 year+0.155
 tornado+0.154
 musician+0.149
Neg logit contributions
 fault-0.158
 pret-0.138
paper-0.132
 puck-0.131
 pounds-0.131
Token the
Token ID262
Feature activation+1.11e-01

Pos logit contributions
 hurricane+0.044
date+0.043
 tornado+0.041
 television+0.040
 year+0.040
Neg logit contributions
 fault-0.052
 pret-0.048
paper-0.046
 loses-0.045
 puck-0.044
Token pain
Token ID2356
Feature activation-3.07e-01

Pos logit contributions
date+0.007
 hurricane+0.007
 Amen+0.007
iet+0.007
Friend+0.007
Neg logit contributions
 fault-0.007
 pret-0.006
 puck-0.005
paper-0.005
 pounds-0.005
Token.
Token ID13
Feature activation+2.77e-01

Pos logit contributions
 fault+0.022
 pret+0.020
paper+0.020
 loses+0.019
 puck+0.019
Neg logit contributions
 hurricane-0.015
date-0.015
 tornado-0.014
 television-0.014
 year-0.014
Token She
Token ID1375
Feature activation+3.10e-01

Pos logit contributions
 hurricane+0.017
date+0.017
 tornado+0.016
 year+0.016
 television+0.015
Neg logit contributions
 fault-0.017
 pret-0.015
paper-0.015
 loses-0.014
 puck-0.014
Token both
Token ID1111
Feature activation+9.37e-05

Pos logit contributions
 fault+0.027
 pret+0.024
 loses+0.022
 puck+0.022
 puts+0.022
Neg logit contributions
 hurricane-0.035
 tornado-0.034
 messenger-0.034
cient-0.032
iet-0.032
Token design
Token ID1486
Feature activation+1.65e-01

Pos logit contributions
 hurricane+0.000
 tornado+0.000
date+0.000
iet+0.000
han+0.000
Neg logit contributions
 fault-0.000
 pret-0.000
 loses-0.000
 puck-0.000
paper-0.000
Token something
Token ID1223
Feature activation-4.36e-01

Pos logit contributions
 television+0.010
date+0.010
 hurricane+0.010
 tornado+0.010
 me+0.010
Neg logit contributions
 fault-0.010
paper-0.008
 pret-0.008
 pounds-0.008
eter-0.008
Token and
Token ID290
Feature activation+6.97e-01

Pos logit contributions
 fault+0.024
 pret+0.022
paper+0.021
 puck+0.020
 loses+0.020
Neg logit contributions
 hurricane-0.029
date-0.028
 tornado-0.027
 television-0.027
 year-0.027
Token show
Token ID905
Feature activation+1.17e+00

Pos logit contributions
 hurricane+0.041
date+0.040
 tornado+0.039
 television+0.038
 year+0.038
Neg logit contributions
 fault-0.044
 pret-0.040
paper-0.038
 loses-0.037
 puck-0.037
Token me
Token ID502
Feature activation+2.64e+00

Pos logit contributions
 hurricane+0.079
date+0.077
 year+0.073
 ranch+0.073
 tornado+0.072
Neg logit contributions
 fault-0.069
 enough-0.063
 pret-0.061
 loses-0.059
ws-0.055
Token later
Token ID1568
Feature activation+3.08e-01

Pos logit contributions
 hurricane+0.205
date+0.201
 tornado+0.195
 Amen+0.191
 year+0.190
Neg logit contributions
 fault-0.119
 pret-0.105
paper-0.096
 loses-0.094
 enough-0.093
Token."
Token ID526
Feature activation-3.44e-02

Pos logit contributions
 hurricane+0.017
date+0.017
 tornado+0.016
 Amen+0.016
ind+0.015
Neg logit contributions
 fault-0.021
 pret-0.019
 enough-0.018
paper-0.018
 loses-0.018
Token
Token ID198
Feature activation+3.75e-02

Pos logit contributions
 fault+0.002
 pret+0.002
paper+0.002
 pounds+0.002
 loses+0.002
Neg logit contributions
 hurricane-0.002
date-0.002
 tornado-0.002
 television-0.002
 year-0.002
Token
Token ID198
Feature activation-3.34e-02

Pos logit contributions
 hurricane+0.002
date+0.002
 tornado+0.002
 television+0.002
ind+0.002
Neg logit contributions
 fault-0.002
 pret-0.002
 loses-0.002
 puck-0.002
paper-0.002
TokenL
Token ID43
Feature activation+2.88e-02

Pos logit contributions
 fault+0.002
 pret+0.002
 puck+0.002
 loses+0.002
paper+0.002
Neg logit contributions
 hurricane-0.002
date-0.002
 tornado-0.002
 television-0.002
cient-0.002
Tokenawn
Token ID3832
Feature activation+4.70e-01

Pos logit contributions
 fault+0.017
 pret+0.014
paper+0.013
 pounds+0.013
eter+0.013
Neg logit contributions
 hurricane-0.019
date-0.019
 tornado-0.018
 television-0.018
 Amen-0.018
Tokenâ
Token ID22940
Feature activation-5.19e-01

Pos logit contributions
 hurricane+0.031
date+0.030
 tornado+0.029
 Amen+0.028
 television+0.028
Neg logit contributions
 fault-0.027
 pret-0.024
paper-0.023
 loses-0.023
 enough-0.023
Token
Token ID26391
Feature activation-5.85e-01

Pos logit contributions
 fault+0.031
 pret+0.028
eter+0.027
 weighs+0.027
 enough+0.027
Neg logit contributions
 hurricane-0.032
 television-0.031
 tornado-0.031
 messenger-0.029
date-0.028
Token
Token ID8151
Feature activation-2.02e+00

Pos logit contributions
 fault+0.033
paper+0.029
 loses+0.029
 pret+0.028
eter+0.028
Neg logit contributions
 hurricane-0.040
 television-0.038
date-0.037
 tornado-0.037
wered-0.036
Tokens
Token ID82
Feature activation-5.45e-01

Pos logit contributions
 fault+0.112
paper+0.092
 enough+0.090
eter+0.090
 loses+0.089
Neg logit contributions
 hurricane-0.141
 tornado-0.135
 television-0.133
 Amen-0.132
date-0.130
Token fault
Token ID8046
Feature activation-3.62e+00

Pos logit contributions
 fault+0.034
 pret+0.028
 enough+0.028
 loses+0.027
paper+0.027
Neg logit contributions
date-0.034
 hurricane-0.034
 Amen-0.032
ind-0.032
 television-0.032
Token.
Token ID13
Feature activation-3.11e-01

Pos logit contributions
 fault+0.239
 pret+0.210
 enough+0.204
paper+0.202
 pounds+0.199
Neg logit contributions
 hurricane-0.208
date-0.207
 tornado-0.193
 Amen-0.189
 television-0.188
Token
Token ID220
Feature activation+2.52e-01

Pos logit contributions
 fault+0.019
 pret+0.017
paper+0.016
 loses+0.016
 puck+0.016
Neg logit contributions
 hurricane-0.019
date-0.019
 tornado-0.018
 television-0.018
 year-0.017
Token
Token ID198
Feature activation-1.86e-02

Pos logit contributions
 hurricane+0.014
date+0.014
 tornado+0.014
 television+0.013
 year+0.013
Neg logit contributions
 fault-0.017
 pret-0.015
paper-0.014
 loses-0.014
 puck-0.014
Token
Token ID198
Feature activation-1.14e-01

Pos logit contributions
 fault+0.001
 pret+0.001
 puck+0.001
paper+0.001
 loses+0.001
Neg logit contributions
 hurricane-0.001
date-0.001
 tornado-0.001
 television-0.001
 musician-0.001
TokenThe
Token ID464
Feature activation+2.40e-01

Pos logit contributions
 fault+0.007
 pret+0.007
 puck+0.007
paper+0.007
 loses+0.006
Neg logit contributions
 hurricane-0.006
date-0.006
 tornado-0.006
 television-0.006
 Amen-0.006
Token felt
Token ID2936
Feature activation-4.35e-01

Pos logit contributions
 pret+0.009
 fault+0.009
 loses+0.008
 enough+0.008
 puts+0.008
Neg logit contributions
date-0.014
 hurricane-0.014
 tornado-0.013
 television-0.013
han-0.013
Token like
Token ID588
Feature activation-1.43e+00

Pos logit contributions
 fault+0.026
 enough+0.025
 loses+0.023
 pret+0.023
 maybe+0.021
Neg logit contributions
date-0.028
 hurricane-0.027
ator-0.026
ind-0.026
 tornado-0.026
Token it
Token ID340
Feature activation-1.04e+00

Pos logit contributions
 fault+0.086
 pret+0.078
paper+0.074
 loses+0.073
 enough+0.072
Neg logit contributions
 hurricane-0.088
date-0.087
 tornado-0.083
 year-0.081
 Amen-0.081
Token was
Token ID373
Feature activation-5.18e-01

Pos logit contributions
 fault+0.060
 pret+0.054
paper+0.051
 loses+0.050
 puck+0.049
Neg logit contributions
 hurricane-0.067
date-0.066
 tornado-0.063
 television-0.062
 year-0.061
Token her
Token ID607
Feature activation-5.66e-01

Pos logit contributions
 fault+0.031
 enough+0.027
 pret+0.025
 loses+0.023
paper+0.022
Neg logit contributions
date-0.035
 Amen-0.034
 hurricane-0.034
cient-0.034
Friend-0.034
Token fault
Token ID8046
Feature activation-3.49e+00

Pos logit contributions
 fault+0.035
 pret+0.028
 enough+0.027
eter+0.026
 loses+0.025
Neg logit contributions
date-0.037
 hurricane-0.037
 Amen-0.036
han-0.036
Friend-0.035
Token,
Token ID11
Feature activation-1.98e-02

Pos logit contributions
 fault+0.237
 enough+0.204
 pret+0.204
paper+0.196
 loses+0.193
Neg logit contributions
date-0.197
 hurricane-0.196
 tornado-0.180
Friend-0.180
 Amen-0.180
Token like
Token ID588
Feature activation-6.22e-01

Pos logit contributions
 fault+0.001
 pret+0.001
paper+0.001
 loses+0.001
 puck+0.001
Neg logit contributions
 hurricane-0.001
date-0.001
 tornado-0.001
 year-0.001
 Amen-0.001
Token she
Token ID673
Feature activation-9.35e-01

Pos logit contributions
 fault+0.036
 pret+0.032
 enough+0.031
paper+0.030
 loses+0.030
Neg logit contributions
date-0.040
 hurricane-0.040
 Amen-0.038
 tornado-0.038
iet-0.037
Token had
Token ID550
Feature activation-2.25e-01

Pos logit contributions
 fault+0.050
 pret+0.045
 loses+0.043
paper+0.042
 enough+0.041
Neg logit contributions
 hurricane-0.065
date-0.064
 tornado-0.061
 television-0.060
 Amen-0.060
Token done
Token ID1760
Feature activation-9.68e-01

Pos logit contributions
 fault+0.013
 pret+0.011
paper+0.011
 loses+0.011
 enough+0.011
Neg logit contributions
 hurricane-0.015
date-0.015
 tornado-0.014
 Amen-0.014
 television-0.014
Token realised
Token ID19169
Feature activation+3.47e-02

Pos logit contributions
 fault+0.015
 pret+0.014
paper+0.013
 pounds+0.013
 puck+0.013
Neg logit contributions
 hurricane-0.014
date-0.013
 year-0.013
 tornado-0.013
iet-0.012
Token it
Token ID340
Feature activation-1.33e-01

Pos logit contributions
date+0.002
 hurricane+0.002
 Amen+0.002
Friend+0.002
ind+0.002
Neg logit contributions
 fault-0.002
 enough-0.002
 pret-0.002
 loses-0.002
paper-0.002
Token wasn
Token ID2492
Feature activation-6.48e-01

Pos logit contributions
 fault+0.007
 pret+0.006
paper+0.006
 loses+0.006
 enough+0.006
Neg logit contributions
 hurricane-0.009
date-0.009
 tornado-0.009
 Amen-0.008
 television-0.008
Token't
Token ID470
Feature activation+7.40e-01

Pos logit contributions
 fault+0.049
 enough+0.045
 loses+0.044
 pret+0.043
paper+0.042
Neg logit contributions
date-0.029
 lens-0.027
iet-0.027
 hurricane-0.027
ind-0.026
Token her
Token ID607
Feature activation-3.38e-01

Pos logit contributions
 hurricane+0.046
date+0.046
 tornado+0.044
 television+0.043
 year+0.043
Neg logit contributions
 fault-0.044
 pret-0.040
paper-0.038
 loses-0.037
 enough-0.037
Token fault
Token ID8046
Feature activation-3.29e+00

Pos logit contributions
 fault+0.022
 pret+0.021
paper+0.021
 puck+0.020
 loses+0.020
Neg logit contributions
 hurricane-0.018
date-0.017
 tornado-0.017
 television-0.017
 year-0.016
Token after
Token ID706
Feature activation+2.70e-01

Pos logit contributions
 fault+0.212
 pret+0.195
paper+0.185
 loses+0.181
 puck+0.180
Neg logit contributions
 hurricane-0.190
date-0.186
 tornado-0.178
 television-0.174
 year-0.174
Token all
Token ID477
Feature activation-5.46e-01

Pos logit contributions
date+0.015
 hurricane+0.015
 Amen+0.015
 tornado+0.014
ind+0.014
Neg logit contributions
 fault-0.018
 pret-0.016
 enough-0.016
 loses-0.015
paper-0.015
Token.
Token ID13
Feature activation-3.89e-01

Pos logit contributions
 fault+0.037
 pret+0.033
 enough+0.033
 loses+0.032
paper+0.031
Neg logit contributions
date-0.030
 hurricane-0.030
 Amen-0.028
ind-0.028
 tornado-0.027
Token<|endoftext|>
Token ID50256
Feature activation+2.93e-02

Pos logit contributions
 fault+0.025
 pret+0.022
paper+0.021
 loses+0.021
 enough+0.021
Neg logit contributions
 hurricane-0.023
date-0.022
 tornado-0.022
 television-0.021
ind-0.021
TokenOnce
Token ID7454
Feature activation-1.08e+00

Pos logit contributions
 hurricane+0.002
date+0.002
 tornado+0.002
ind+0.002
 television+0.002
Neg logit contributions
 fault-0.002
 pret-0.002
paper-0.001
 enough-0.001
 puck-0.001
Token "
Token ID366
Feature activation-2.47e-02

Pos logit contributions
 fault+0.005
 pret+0.004
paper+0.004
 loses+0.004
 puck+0.004
Neg logit contributions
 hurricane-0.004
date-0.004
 tornado-0.004
 year-0.004
 television-0.004
TokenIt
Token ID1026
Feature activation-7.37e-01

Pos logit contributions
 fault+0.002
 pret+0.002
paper+0.002
 loses+0.002
 puck+0.002
Neg logit contributions
 hurricane-0.001
date-0.001
 tornado-0.001
 television-0.001
 year-0.001
Token's
Token ID338
Feature activation-6.91e-01

Pos logit contributions
 fault+0.036
 pret+0.032
paper+0.030
 loses+0.029
 puck+0.029
Neg logit contributions
 hurricane-0.054
date-0.053
 tornado-0.051
 television-0.050
 year-0.050
Token not
Token ID407
Feature activation+3.49e-01

Pos logit contributions
 fault+0.039
 enough+0.034
 pret+0.033
 pounds+0.030
 loses+0.030
Neg logit contributions
date-0.049
 hurricane-0.047
 Amen-0.045
Friend-0.044
 musician-0.044
Token my
Token ID616
Feature activation+9.18e-02

Pos logit contributions
 hurricane+0.020
 television+0.019
 tornado+0.019
date+0.018
 year+0.018
Neg logit contributions
 fault-0.022
 pret-0.021
paper-0.020
 loses-0.020
 puck-0.020
Token fault
Token ID8046
Feature activation-3.28e+00

Pos logit contributions
date+0.006
 hurricane+0.006
 Amen+0.006
 tornado+0.006
 year+0.006
Neg logit contributions
 fault-0.005
 pret-0.005
paper-0.004
 loses-0.004
 puck-0.004
Token.
Token ID13
Feature activation-2.74e-02

Pos logit contributions
 fault+0.224
 pret+0.204
paper+0.195
 loses+0.191
 puck+0.190
Neg logit contributions
 hurricane-0.178
date-0.175
 tornado-0.166
 television-0.162
 year-0.162
Token I
Token ID314
Feature activation+9.53e-02

Pos logit contributions
 fault+0.001
 pret+0.001
eter+0.001
 loses+0.001
 puck+0.001
Neg logit contributions
 hurricane-0.002
date-0.002
 tornado-0.002
iet-0.002
 ranch-0.002
Token can
Token ID460
Feature activation-4.65e-01

Pos logit contributions
 hurricane+0.007
date+0.007
 tornado+0.007
 television+0.007
ind+0.007
Neg logit contributions
 fault-0.005
 pret-0.004
 loses-0.004
 enough-0.003
paper-0.003
Token't
Token ID470
Feature activation+1.11e+00

Pos logit contributions
 fault+0.026
 loses+0.026
onna+0.025
 pret+0.025
 puts+0.024
Neg logit contributions
 hurricane-0.027
iet-0.026
date-0.026
 courtroom-0.026
 ranch-0.026
Token help
Token ID1037
Feature activation+6.62e-01

Pos logit contributions
 Amen+0.078
cient+0.078
 lens+0.078
iet+0.077
 courtroom+0.077
Neg logit contributions
 loses-0.047
 fault-0.043
 pret-0.041
 sometime-0.040
 puts-0.039
Token
Token ID95
Feature activation-2.73e-01

Pos logit contributions
 fault+0.013
 pret+0.012
paper+0.012
 pounds+0.011
 puck+0.011
Neg logit contributions
 hurricane-0.012
date-0.012
 tornado-0.011
 year-0.011
 television-0.011
Token
Token ID26391
Feature activation-1.65e-01

Pos logit contributions
 fault+0.018
 pret+0.017
paper+0.016
 puck+0.016
 loses+0.016
Neg logit contributions
 hurricane-0.016
date-0.015
 tornado-0.015
 television-0.014
 year-0.014
Token
Token ID129
Feature activation+5.31e-01

Pos logit contributions
 fault+0.010
 pret+0.009
paper+0.008
 loses+0.008
 puck+0.008
Neg logit contributions
 hurricane-0.010
date-0.010
 tornado-0.010
 year-0.010
 television-0.010
Token
Token ID241
Feature activation-1.98e-01

Pos logit contributions
date+0.029
 hurricane+0.029
 tornado+0.027
 messenger+0.025
 year+0.025
Neg logit contributions
 fault-0.038
 pret-0.036
paper-0.033
 loses-0.033
ws-0.032
TokenUr
Token ID16692
Feature activation-8.14e-01

Pos logit contributions
 fault+0.013
 pret+0.012
paper+0.011
 loses+0.011
 puck+0.011
Neg logit contributions
 hurricane-0.011
date-0.011
 year-0.010
 tornado-0.010
 television-0.010
Tokenge
Token ID469
Feature activation-3.23e+00

Pos logit contributions
 fault+0.045
 pret+0.042
paper+0.040
 loses+0.039
 pounds+0.038
Neg logit contributions
 hurricane-0.055
 tornado-0.054
date-0.053
 television-0.051
 ranch-0.050
Token your
Token ID534
Feature activation-2.85e-01

Pos logit contributions
 fault+0.209
 pret+0.189
paper+0.181
 enough+0.181
 loses+0.178
Neg logit contributions
 hurricane-0.187
date-0.184
 tornado-0.174
 Amen-0.173
 television-0.170
Token heart
Token ID2612
Feature activation-1.76e-01

Pos logit contributions
 fault+0.018
 pret+0.016
 enough+0.015
 teeth+0.015
 loses+0.015
Neg logit contributions
date-0.018
 hurricane-0.017
 Amen-0.017
 tornado-0.017
ilt-0.016
Token and
Token ID290
Feature activation-4.09e-02

Pos logit contributions
 fault+0.011
 pret+0.010
paper+0.009
 loses+0.009
 enough+0.009
Neg logit contributions
 hurricane-0.011
date-0.011
 tornado-0.010
 television-0.010
 Amen-0.010
Token mind
Token ID2000
Feature activation+5.29e-01

Pos logit contributions
 fault+0.002
 pret+0.002
paper+0.002
 loses+0.002
 enough+0.002
Neg logit contributions
 hurricane-0.003
date-0.003
 tornado-0.003
 television-0.002
 year-0.002
Token to
Token ID284
Feature activation-3.69e-01

Pos logit contributions
date+0.033
 hurricane+0.033
 tornado+0.032
 Amen+0.031
 television+0.030
Neg logit contributions
 fault-0.033
 pret-0.029
 enough-0.028
paper-0.028
 loses-0.028
Token it
Token ID340
Feature activation-3.46e-01

Pos logit contributions
 fault+0.006
 pret+0.006
paper+0.006
 enough+0.005
 loses+0.005
Neg logit contributions
 hurricane-0.006
date-0.006
 tornado-0.006
 year-0.006
 television-0.006
Token was
Token ID373
Feature activation-1.61e-02

Pos logit contributions
 fault+0.021
 pret+0.019
paper+0.018
 pounds+0.018
 puck+0.018
Neg logit contributions
 hurricane-0.021
date-0.021
 tornado-0.020
 year-0.020
 television-0.020
Token Tim
Token ID5045
Feature activation-4.61e-01

Pos logit contributions
 fault+0.001
 pret+0.001
 enough+0.001
 loses+0.001
paper+0.001
Neg logit contributions
date-0.001
 hurricane-0.001
 tornado-0.001
 year-0.001
 Amen-0.001
Tokenmy
Token ID1820
Feature activation+7.43e-01

Pos logit contributions
 pret+0.032
 fault+0.032
 puck+0.031
 loses+0.030
 pounds+0.030
Neg logit contributions
date-0.021
 year-0.021
 hurricane-0.021
 telephone-0.020
 voice-0.020
Token's
Token ID338
Feature activation-4.54e-01

Pos logit contributions
 hurricane+0.043
date+0.042
 year+0.042
 tornado+0.041
 voice+0.041
Neg logit contributions
 fault-0.045
 pret-0.043
 puck-0.040
paper-0.040
ws-0.039
Token fault
Token ID8046
Feature activation-3.19e+00

Pos logit contributions
ws+0.031
paper+0.031
 pret+0.030
 weighs+0.030
 pounds+0.030
Neg logit contributions
 voice-0.022
 hurricane-0.021
 television-0.021
 year-0.020
 tornado-0.020
Token but
Token ID475
Feature activation+4.16e-01

Pos logit contributions
 fault+0.190
 pret+0.181
paper+0.169
 loses+0.166
ws+0.164
Neg logit contributions
 hurricane-0.195
 tornado-0.186
date-0.186
 television-0.184
 year-0.181
Token Tim
Token ID5045
Feature activation-2.68e-01

Pos logit contributions
 hurricane+0.025
date+0.023
 tornado+0.023
 television+0.023
 year+0.022
Neg logit contributions
 fault-0.027
 pret-0.024
paper-0.023
 pounds-0.023
 puck-0.023
Tokenmy
Token ID1820
Feature activation+1.03e+00

Pos logit contributions
 fault+0.018
 pret+0.017
 puck+0.016
 pounds+0.016
paper+0.016
Neg logit contributions
date-0.014
 hurricane-0.014
 year-0.013
 voice-0.013
ind-0.013
Token felt
Token ID2936
Feature activation-3.75e-02

Pos logit contributions
 hurricane+0.061
date+0.060
 tornado+0.057
 year+0.057
 television+0.056
Neg logit contributions
 fault-0.063
 pret-0.057
paper-0.055
 puck-0.054
 pounds-0.054
Token guilty
Token ID6717
Feature activation-8.57e-01

Pos logit contributions
 fault+0.003
 pret+0.002
 loses+0.002
 enough+0.002
paper+0.002
Neg logit contributions
 hurricane-0.002
date-0.002
 tornado-0.002
 television-0.002
 year-0.002
Token knew
Token ID2993
Feature activation+1.55e-01

Pos logit contributions
 fault+0.035
 pret+0.032
paper+0.030
 loses+0.029
 puck+0.029
Neg logit contributions
 hurricane-0.037
date-0.036
 tornado-0.035
 television-0.034
 year-0.034
Token it
Token ID340
Feature activation-2.97e-02

Pos logit contributions
 hurricane+0.011
date+0.010
 tornado+0.010
 television+0.010
 year+0.010
Neg logit contributions
 fault-0.008
 pret-0.008
paper-0.007
 loses-0.007
 puck-0.007
Token was
Token ID373
Feature activation+1.43e-02

Pos logit contributions
 fault+0.002
 pret+0.002
paper+0.002
 puck+0.002
 loses+0.002
Neg logit contributions
 hurricane-0.002
date-0.002
 tornado-0.002
 year-0.002
 television-0.002
Token his
Token ID465
Feature activation-2.26e-01

Pos logit contributions
date+0.001
 hurricane+0.001
 tornado+0.001
cient+0.001
ind+0.001
Neg logit contributions
 fault-0.001
 enough-0.001
 pret-0.001
 pounds-0.001
 puck-0.001
Token own
Token ID898
Feature activation+6.45e-01

Pos logit contributions
 fault+0.015
 pret+0.014
paper+0.014
 loses+0.013
 puck+0.013
Neg logit contributions
 hurricane-0.012
date-0.012
 tornado-0.011
 television-0.011
 year-0.011
Token fault
Token ID8046
Feature activation-3.15e+00

Pos logit contributions
 hurricane+0.034
date+0.033
 tornado+0.031
 television+0.030
 Amen+0.030
Neg logit contributions
 fault-0.046
 pret-0.042
paper-0.040
 loses-0.039
 puck-0.039
Token.
Token ID13
Feature activation-1.03e-01

Pos logit contributions
 fault+0.221
 pret+0.198
paper+0.191
 enough+0.190
 puck+0.188
Neg logit contributions
 hurricane-0.166
date-0.164
 tornado-0.154
 Amen-0.149
 television-0.148
Token He
Token ID679
Feature activation-1.19e-01

Pos logit contributions
 fault+0.007
 pret+0.006
paper+0.006
 loses+0.006
 pounds+0.006
Neg logit contributions
 hurricane-0.006
date-0.006
 Amen-0.005
 television-0.005
 tornado-0.005
Token would
Token ID561
Feature activation+5.46e-01

Pos logit contributions
 fault+0.008
 pret+0.007
paper+0.007
 loses+0.006
 pounds+0.006
Neg logit contributions
 hurricane-0.007
date-0.007
 tornado-0.007
 year-0.007
 Amen-0.006
Token have
Token ID423
Feature activation-3.36e-01

Pos logit contributions
date+0.035
 hurricane+0.035
 tornado+0.034
ind+0.033
iet+0.033
Neg logit contributions
 fault-0.031
 pret-0.028
 enough-0.027
 loses-0.026
 puts-0.026
Token to
Token ID284
Feature activation-2.99e-03

Pos logit contributions
 pret+0.019
 fault+0.018
paper+0.018
 pounds+0.018
 loses+0.018
Neg logit contributions
 hurricane-0.019
date-0.019
 year-0.018
 me-0.018
 others-0.018
Token knew
Token ID2993
Feature activation-3.41e-02

Pos logit contributions
 hurricane+0.062
date+0.060
 tornado+0.059
 television+0.058
han+0.057
Neg logit contributions
 fault-0.050
 pret-0.047
 loses-0.044
 enough-0.044
paper-0.042
Token it
Token ID340
Feature activation+2.53e-01

Pos logit contributions
 fault+0.002
 pret+0.002
 enough+0.002
paper+0.002
 loses+0.002
Neg logit contributions
 hurricane-0.002
date-0.002
 tornado-0.002
 television-0.002
 Amen-0.002
Token was
Token ID373
Feature activation-6.52e-02

Pos logit contributions
 hurricane+0.015
date+0.015
 tornado+0.015
 year+0.015
 television+0.014
Neg logit contributions
 fault-0.015
 pret-0.014
paper-0.013
 puck-0.013
 pounds-0.013
Token her
Token ID607
Feature activation-1.88e-01

Pos logit contributions
 fault+0.004
 enough+0.003
 pret+0.003
 pounds+0.003
paper+0.003
Neg logit contributions
date-0.004
 hurricane-0.004
cient-0.004
Friend-0.004
 tornado-0.004
Token own
Token ID898
Feature activation+5.55e-01

Pos logit contributions
 fault+0.013
 pret+0.011
 enough+0.011
 loses+0.010
paper+0.010
Neg logit contributions
date-0.010
 hurricane-0.010
Friend-0.010
iet-0.010
 Amen-0.010
Token fault
Token ID8046
Feature activation-3.12e+00

Pos logit contributions
 hurricane+0.031
date+0.031
Friend+0.029
 Amen+0.029
 tornado+0.029
Neg logit contributions
 fault-0.039
 pret-0.033
 enough-0.032
 pounds-0.031
 puck-0.031
Token.
Token ID13
Feature activation-2.51e-01

Pos logit contributions
 fault+0.216
 pret+0.192
paper+0.185
 enough+0.184
 loses+0.182
Neg logit contributions
 hurricane-0.169
date-0.167
 tornado-0.158
 Amen-0.153
 television-0.152
Token
Token ID198
Feature activation-1.05e-01

Pos logit contributions
 fault+0.015
 pret+0.013
 enough+0.012
eter+0.012
 puck+0.012
Neg logit contributions
 hurricane-0.016
date-0.015
 tornado-0.015
ind-0.015
 television-0.014
Token
Token ID198
Feature activation-1.73e-01

Pos logit contributions
 fault+0.007
 pret+0.006
 Mir+0.006
 puck+0.006
ends+0.006
Neg logit contributions
 hurricane-0.006
 tornado-0.005
date-0.005
 television-0.005
cient-0.005
TokenThe
Token ID464
Feature activation+2.81e-01

Pos logit contributions
 fault+0.009
ends+0.008
 puck+0.008
 pret+0.008
 Mir+0.008
Neg logit contributions
 hurricane-0.011
 television-0.010
 tornado-0.010
cient-0.010
 lens-0.010
Token girl
Token ID2576
Feature activation-2.65e-02

Pos logit contributions
date+0.015
cient+0.014
ilt+0.014
han+0.014
ind+0.014
Neg logit contributions
 fault-0.018
 pret-0.017
 puck-0.017
 enough-0.016
 cones-0.016
Token "
Token ID366
Feature activation+1.34e-01

Pos logit contributions
 hurricane+0.018
date+0.018
 tornado+0.017
 year+0.016
ind+0.016
Neg logit contributions
 fault-0.016
 pret-0.015
paper-0.014
 loses-0.014
 puck-0.014
TokenIt
Token ID1026
Feature activation-6.23e-01

Pos logit contributions
 hurricane+0.007
date+0.006
 tornado+0.006
 year+0.006
 Amen+0.006
Neg logit contributions
 fault-0.010
 pret-0.009
paper-0.009
 loses-0.009
 pounds-0.009
Token's
Token ID338
Feature activation-6.47e-01

Pos logit contributions
 fault+0.038
paper+0.034
 pounds+0.034
 puck+0.034
 pret+0.034
Neg logit contributions
 hurricane-0.037
date-0.036
 tornado-0.036
 television-0.036
 Amen-0.035
Token not
Token ID407
Feature activation+4.05e-01

Pos logit contributions
 fault+0.024
 enough+0.018
 pret+0.018
 loses+0.016
eter+0.015
Neg logit contributions
date-0.057
 hurricane-0.057
 tornado-0.054
 television-0.053
cient-0.053
Token your
Token ID534
Feature activation+3.28e-01

Pos logit contributions
 hurricane+0.022
 television+0.021
 tornado+0.021
date+0.020
 year+0.020
Neg logit contributions
 fault-0.026
 pret-0.025
paper-0.025
 puck-0.024
 loses-0.024
Token fault
Token ID8046
Feature activation-3.06e+00

Pos logit contributions
 hurricane+0.034
date+0.033
 tornado+0.033
 television+0.033
 year+0.032
Neg logit contributions
 fault-0.005
 pret-0.004
paper-0.004
 loses-0.003
 puck-0.003
Token,
Token ID11
Feature activation+3.64e-01

Pos logit contributions
 fault+0.229
 pret+0.220
paper+0.210
 puck+0.207
 loses+0.205
Neg logit contributions
 hurricane-0.140
date-0.131
 tornado-0.131
 television-0.130
 year-0.127
Token honey
Token ID12498
Feature activation+1.77e-01

Pos logit contributions
 hurricane+0.024
 tornado+0.023
 television+0.022
 year+0.022
 voice+0.022
Neg logit contributions
 fault-0.021
 pret-0.019
paper-0.019
ws-0.018
eter-0.018
Token.
Token ID13
Feature activation+1.36e-01

Pos logit contributions
 hurricane+0.007
date+0.007
 tornado+0.007
 year+0.007
 television+0.007
Neg logit contributions
 fault-0.014
 pret-0.013
paper-0.013
 puck-0.013
 loses-0.012
Token Ben
Token ID3932
Feature activation+4.98e-01

Pos logit contributions
 hurricane+0.008
date+0.008
 tornado+0.008
 television+0.007
 Amen+0.007
Neg logit contributions
 fault-0.008
 pret-0.008
paper-0.007
 puck-0.007
 loses-0.007
Token made
Token ID925
Feature activation-3.19e-01

Pos logit contributions
 hurricane+0.034
date+0.034
 tornado+0.033
 year+0.032
 television+0.032
Neg logit contributions
 fault-0.026
 pret-0.023
paper-0.022
 puck-0.022
 loses-0.021
Token that
Token ID326
Feature activation-2.80e-02

Pos logit contributions
 fault+0.047
 pret+0.042
 enough+0.042
 loses+0.039
 pounds+0.039
Neg logit contributions
date-0.051
 hurricane-0.051
 tornado-0.048
ind-0.048
 Amen-0.047
Token it
Token ID340
Feature activation+2.32e-01

Pos logit contributions
 fault+0.002
 pret+0.001
 enough+0.001
paper+0.001
 loses+0.001
Neg logit contributions
 hurricane-0.002
date-0.002
 tornado-0.002
ind-0.002
 Amen-0.002
Token wasn
Token ID2492
Feature activation-5.25e-01

Pos logit contributions
 hurricane+0.016
date+0.016
 tornado+0.015
 Amen+0.015
 television+0.015
Neg logit contributions
 fault-0.012
 pret-0.011
paper-0.010
 loses-0.010
 enough-0.010
Token't
Token ID470
Feature activation+7.81e-01

Pos logit contributions
 fault+0.041
 enough+0.038
 loses+0.037
 pret+0.037
paper+0.036
Neg logit contributions
date-0.022
 hurricane-0.021
iet-0.021
 lens-0.020
ind-0.020
Token her
Token ID607
Feature activation-2.58e-01

Pos logit contributions
 hurricane+0.048
 tornado+0.045
date+0.045
 television+0.044
 year+0.044
Neg logit contributions
 fault-0.047
 pret-0.044
paper-0.042
 loses-0.041
 puck-0.041
Token fault
Token ID8046
Feature activation-3.06e+00

Pos logit contributions
 fault+0.017
 pret+0.016
paper+0.015
 loses+0.015
 puck+0.015
Neg logit contributions
 hurricane-0.014
date-0.014
 tornado-0.013
 television-0.013
 year-0.013
Token.
Token ID13
Feature activation-1.75e-01

Pos logit contributions
 fault+0.212
 pret+0.196
paper+0.187
 loses+0.184
 puck+0.183
Neg logit contributions
 hurricane-0.161
date-0.157
 tornado-0.150
 television-0.146
 year-0.146
Token Finally
Token ID9461
Feature activation+3.69e-01

Pos logit contributions
 fault+0.012
 pret+0.011
paper+0.010
 loses+0.010
 puck+0.010
Neg logit contributions
 hurricane-0.010
date-0.010
 tornado-0.009
 television-0.009
 year-0.009
Token,
Token ID11
Feature activation+5.88e-01

Pos logit contributions
 hurricane+0.023
date+0.023
 tornado+0.022
ind+0.022
 Amen+0.021
Neg logit contributions
 fault-0.022
 pret-0.020
 enough-0.019
paper-0.019
 loses-0.018
Token the
Token ID262
Feature activation+1.94e-01

Pos logit contributions
 hurricane+0.039
date+0.038
ind+0.038
 tornado+0.038
 Amen+0.037
Neg logit contributions
 fault-0.032
 pret-0.028
 enough-0.028
paper-0.027
 loses-0.026
Token family
Token ID1641
Feature activation+2.36e-01

Pos logit contributions
ind+0.011
iet+0.010
cient+0.010
date+0.010
rim+0.010
Neg logit contributions
 fault-0.013
 puck-0.012
 pret-0.012
 enough-0.012
 cones-0.011
Token it
Token ID340
Feature activation-1.23e-01

Pos logit contributions
 hurricane+0.004
date+0.004
 tornado+0.004
 television+0.004
 year+0.004
Neg logit contributions
 fault-0.003
 pret-0.003
paper-0.002
 enough-0.002
 loses-0.002
Token was
Token ID373
Feature activation+7.44e-02

Pos logit contributions
 fault+0.008
 pret+0.007
paper+0.007
 puck+0.007
 pounds+0.006
Neg logit contributions
 hurricane-0.007
date-0.007
 tornado-0.007
 year-0.007
 television-0.007
Token all
Token ID477
Feature activation-7.45e-01

Pos logit contributions
date+0.005
 hurricane+0.005
iet+0.005
ind+0.005
Friend+0.005
Neg logit contributions
 fault-0.004
 enough-0.004
 pret-0.004
paper-0.003
 loses-0.003
Token Joe
Token ID5689
Feature activation+1.60e-02

Pos logit contributions
 fault+0.045
 pret+0.042
paper+0.040
 puck+0.039
 loses+0.039
Neg logit contributions
 hurricane-0.046
date-0.044
 tornado-0.043
 television-0.042
 year-0.042
Token's
Token ID338
Feature activation-4.21e-01

Pos logit contributions
 hurricane+0.001
 year+0.001
date+0.001
 tornado+0.001
 voice+0.001
Neg logit contributions
 fault-0.001
 pret-0.001
paper-0.001
 puck-0.001
ws-0.001
Token fault
Token ID8046
Feature activation-3.04e+00

Pos logit contributions
 fault+0.029
 pret+0.025
paper+0.024
 enough+0.024
 pounds+0.023
Neg logit contributions
date-0.024
 hurricane-0.023
 tornado-0.022
 Amen-0.022
iet-0.022
Token.
Token ID13
Feature activation-3.35e-02

Pos logit contributions
 fault+0.192
 pret+0.173
paper+0.165
 loses+0.161
 enough+0.161
Neg logit contributions
 hurricane-0.180
date-0.178
 tornado-0.169
 television-0.165
 year-0.164
Token Joe
Token ID5689
Feature activation+2.52e-01

Pos logit contributions
 fault+0.002
 pret+0.002
 puck+0.002
 loses+0.002
paper+0.002
Neg logit contributions
 hurricane-0.002
date-0.002
 tornado-0.002
ind-0.002
 television-0.002
Token tried
Token ID3088
Feature activation+6.12e-01

Pos logit contributions
 hurricane+0.017
date+0.017
 tornado+0.016
 television+0.016
ind+0.016
Neg logit contributions
 fault-0.013
 pret-0.012
 enough-0.012
 loses-0.011
paper-0.011
Token to
Token ID284
Feature activation+2.75e-01

Pos logit contributions
 hurricane+0.042
date+0.040
 television+0.040
 tornado+0.040
 year+0.039
Neg logit contributions
 fault-0.031
 pret-0.029
 pounds-0.028
paper-0.028
 loses-0.027
Token talk
Token ID1561
Feature activation+4.45e-02

Pos logit contributions
cient+0.018
date+0.018
 hurricane+0.018
ind+0.018
 tornado+0.017
Neg logit contributions
 fault-0.014
 pret-0.013
 puts-0.012
 loses-0.012
 enough-0.012
Token He
Token ID679
Feature activation-1.49e-01

Pos logit contributions
 hurricane+0.003
date+0.003
 tornado+0.003
 television+0.003
 year+0.003
Neg logit contributions
 fault-0.002
 pret-0.002
paper-0.002
 loses-0.002
 puck-0.002
Token realized
Token ID6939
Feature activation-2.41e-01

Pos logit contributions
 fault+0.009
 pret+0.008
 loses+0.007
paper+0.007
 enough+0.007
Neg logit contributions
 hurricane-0.010
date-0.009
 tornado-0.009
 television-0.009
 year-0.009
Token it
Token ID340
Feature activation-1.22e-01

Pos logit contributions
 fault+0.014
 pret+0.013
paper+0.012
 enough+0.012
 loses+0.012
Neg logit contributions
date-0.015
 hurricane-0.015
 tornado-0.014
 Amen-0.014
 television-0.014
Token was
Token ID373
Feature activation+5.22e-02

Pos logit contributions
 fault+0.008
 pret+0.007
paper+0.007
 pounds+0.007
eter+0.007
Neg logit contributions
 hurricane-0.007
date-0.007
 tornado-0.007
 year-0.007
 television-0.006
Token his
Token ID465
Feature activation-9.97e-02

Pos logit contributions
date+0.004
 hurricane+0.004
 tornado+0.003
 Amen+0.003
 year+0.003
Neg logit contributions
 fault-0.003
 pret-0.003
 enough-0.002
paper-0.002
 loses-0.002
Token fault
Token ID8046
Feature activation-3.04e+00

Pos logit contributions
 fault+0.005
 pret+0.005
paper+0.005
 loses+0.004
 puck+0.004
Neg logit contributions
 hurricane-0.007
date-0.007
 tornado-0.007
 television-0.006
 year-0.006
Token and
Token ID290
Feature activation+5.90e-01

Pos logit contributions
 fault+0.193
 pret+0.180
paper+0.170
 loses+0.167
 puck+0.164
Neg logit contributions
 hurricane-0.177
date-0.171
 tornado-0.167
 television-0.163
 year-0.163
Token he
Token ID339
Feature activation-2.00e-01

Pos logit contributions
date+0.046
 hurricane+0.045
 tornado+0.043
 television+0.043
ind+0.042
Neg logit contributions
 fault-0.026
 pret-0.024
 loses-0.022
 enough-0.022
paper-0.022
Token felt
Token ID2936
Feature activation+2.39e-01

Pos logit contributions
 fault+0.012
 pret+0.011
paper+0.010
 loses+0.010
 puck+0.010
Neg logit contributions
 hurricane-0.013
date-0.012
 tornado-0.012
 television-0.012
 year-0.012
Token regret
Token ID13721
Feature activation-6.54e-01

Pos logit contributions
date+0.014
 hurricane+0.014
 tornado+0.013
ind+0.013
 television+0.013
Neg logit contributions
 fault-0.015
 pret-0.014
 enough-0.013
 loses-0.013
paper-0.013
Token.
Token ID13
Feature activation-1.12e-01

Pos logit contributions
 fault+0.046
 pret+0.042
paper+0.040
 loses+0.040
 puck+0.039
Neg logit contributions
 hurricane-0.034
date-0.033
 tornado-0.032
 television-0.031
 year-0.031
Token "
Token ID366
Feature activation-3.31e-01

Pos logit contributions
 hurricane+0.024
date+0.023
ind+0.023
 tornado+0.023
ilt+0.022
Neg logit contributions
 fault-0.020
 pret-0.017
 enough-0.017
paper-0.016
 loses-0.016
TokenIt
Token ID1026
Feature activation-8.46e-01

Pos logit contributions
 fault+0.021
 pret+0.020
paper+0.019
eter+0.019
 loses+0.018
Neg logit contributions
 hurricane-0.019
 tornado-0.018
date-0.018
 television-0.018
 ranch-0.017
Token's
Token ID338
Feature activation-5.84e-01

Pos logit contributions
 fault+0.040
 pret+0.035
paper+0.032
 loses+0.032
 enough+0.032
Neg logit contributions
 hurricane-0.064
date-0.063
 tornado-0.060
 television-0.060
 year-0.059
Token your
Token ID534
Feature activation+4.45e-01

Pos logit contributions
 fault+0.031
 enough+0.026
 pret+0.026
eter+0.023
 loses+0.023
Neg logit contributions
date-0.042
 hurricane-0.041
Friend-0.039
 tornado-0.039
iet-0.039
Token own
Token ID898
Feature activation+9.10e-01

Pos logit contributions
 hurricane+0.025
date+0.025
 tornado+0.024
 Amen+0.023
Friend+0.023
Neg logit contributions
 fault-0.031
 pret-0.027
 enough-0.025
paper-0.025
 puck-0.025
Token fault
Token ID8046
Feature activation-3.03e+00

Pos logit contributions
date+0.050
 hurricane+0.050
Friend+0.048
han+0.047
 Amen+0.047
Neg logit contributions
 fault-0.064
 pret-0.055
 enough-0.053
 pounds-0.050
 puck-0.050
Token,
Token ID11
Feature activation+1.99e-01

Pos logit contributions
 fault+0.191
 pret+0.166
 enough+0.159
paper+0.158
 pounds+0.155
Neg logit contributions
 hurricane-0.186
date-0.184
 tornado-0.174
ind-0.170
Friend-0.169
Token you
Token ID345
Feature activation+1.05e+00

Pos logit contributions
 hurricane+0.013
date+0.013
 tornado+0.013
 Amen+0.012
 year+0.012
Neg logit contributions
 fault-0.011
 pret-0.010
paper-0.009
 enough-0.009
 puck-0.009
Token should
Token ID815
Feature activation+2.76e-01

Pos logit contributions
 hurricane+0.081
date+0.080
 tornado+0.077
 television+0.077
ind+0.076
Neg logit contributions
 fault-0.050
 pret-0.042
 loses-0.042
 enough-0.040
paper-0.037
Token've
Token ID1053
Feature activation+6.98e-03

Pos logit contributions
date+0.021
 tornado+0.020
cient+0.020
 hurricane+0.020
Friend+0.020
Neg logit contributions
 fault-0.014
 enough-0.011
 puts-0.010
 pret-0.010
 loses-0.010
Token been
Token ID587
Feature activation+1.46e-01

Pos logit contributions
 hurricane+0.000
date+0.000
 tornado+0.000
 television+0.000
ind+0.000
Neg logit contributions
 fault-0.000
 pret-0.000
 enough-0.000
 loses-0.000
 puck-0.000
Token bird
Token ID6512
Feature activation-4.37e-01

Pos logit contributions
 hurricane+0.010
date+0.010
 tornado+0.009
 television+0.009
 year+0.009
Neg logit contributions
 fault-0.013
 pret-0.012
paper-0.011
 enough-0.011
 loses-0.011
Token.
Token ID13
Feature activation-2.92e-01

Pos logit contributions
 fault+0.032
 pret+0.029
 enough+0.028
 loses+0.028
paper+0.028
Neg logit contributions
 hurricane-0.022
date-0.021
 tornado-0.020
 television-0.020
 Amen-0.019
Token You
Token ID921
Feature activation+5.21e-01

Pos logit contributions
 fault+0.019
 pret+0.017
paper+0.016
 loses+0.016
 puck+0.016
Neg logit contributions
 hurricane-0.017
date-0.016
 tornado-0.016
 Amen-0.015
 television-0.015
Token admit
Token ID9159
Feature activation+5.16e-02

Pos logit contributions
 hurricane+0.039
date+0.038
 tornado+0.037
 television+0.036
 year+0.036
Neg logit contributions
 fault-0.025
 pret-0.022
paper-0.021
 loses-0.020
 puck-0.020
Token your
Token ID534
Feature activation+2.01e-01

Pos logit contributions
date+0.003
 hurricane+0.003
ilt+0.003
 year+0.003
 Amen+0.003
Neg logit contributions
 fault-0.003
 pret-0.003
 enough-0.003
 loses-0.003
paper-0.002
Token fault
Token ID8046
Feature activation-3.02e+00

Pos logit contributions
date+0.012
 hurricane+0.011
han+0.010
Friend+0.010
ilt+0.010
Neg logit contributions
 fault-0.015
 pret-0.012
 enough-0.011
 teeth-0.011
 loses-0.011
Token and
Token ID290
Feature activation+6.22e-01

Pos logit contributions
 fault+0.218
 pret+0.200
paper+0.192
 loses+0.188
 puck+0.187
Neg logit contributions
 hurricane-0.152
date-0.149
 tornado-0.141
 television-0.137
 year-0.136
Token ask
Token ID1265
Feature activation+6.98e-01

Pos logit contributions
 hurricane+0.039
date+0.039
 tornado+0.037
 year+0.036
 television+0.036
Neg logit contributions
 fault-0.037
 pret-0.034
paper-0.031
 loses-0.031
 enough-0.031
Token for
Token ID329
Feature activation+7.55e-01

Pos logit contributions
 hurricane+0.040
date+0.039
 tornado+0.037
 Amen+0.036
ilt+0.035
Neg logit contributions
 fault-0.046
 pret-0.043
 enough-0.042
 loses-0.040
 pounds-0.040
Token mercy
Token ID17703
Feature activation-1.91e-01

Pos logit contributions
date+0.048
 hurricane+0.048
iet+0.046
 tornado+0.045
 lens+0.044
Neg logit contributions
 fault-0.043
 pret-0.040
 enough-0.040
 loses-0.038
 puck-0.037
Token.
Token ID13
Feature activation-3.01e-01

Pos logit contributions
 fault+0.014
 pret+0.013
paper+0.013
 puck+0.013
 loses+0.013
Neg logit contributions
 hurricane-0.009
date-0.009
 year-0.008
 tornado-0.008
 television-0.008
Token realized
Token ID6939
Feature activation-6.20e-02

Pos logit contributions
 hurricane+0.005
date+0.005
 tornado+0.005
 television+0.005
 Amen+0.005
Neg logit contributions
 fault-0.005
 pret-0.004
 loses-0.004
 enough-0.004
paper-0.004
Token it
Token ID340
Feature activation-5.03e-02

Pos logit contributions
 fault+0.004
 enough+0.003
 pret+0.003
paper+0.003
 loses+0.003
Neg logit contributions
date-0.004
 hurricane-0.004
 Amen-0.004
 tornado-0.004
ind-0.004
Token was
Token ID373
Feature activation+6.73e-02

Pos logit contributions
 fault+0.003
 pret+0.003
paper+0.002
 loses+0.002
 puck+0.002
Neg logit contributions
 hurricane-0.003
date-0.003
 tornado-0.003
 television-0.003
 year-0.003
Token his
Token ID465
Feature activation-3.56e-02

Pos logit contributions
date+0.005
 hurricane+0.005
 tornado+0.004
 Amen+0.004
iet+0.004
Neg logit contributions
 fault-0.004
 enough-0.003
 pret-0.003
 loses-0.003
paper-0.003
Token own
Token ID898
Feature activation+8.68e-01

Pos logit contributions
 fault+0.002
 pret+0.001
 enough+0.001
 loses+0.001
paper+0.001
Neg logit contributions
date-0.003
 hurricane-0.003
 tornado-0.003
 Amen-0.003
 year-0.003
Token fault
Token ID8046
Feature activation-3.00e+00

Pos logit contributions
date+0.052
 hurricane+0.051
 Amen+0.050
ind+0.049
Friend+0.049
Neg logit contributions
 fault-0.059
 pret-0.048
 enough-0.047
 loses-0.045
paper-0.045
Token for
Token ID329
Feature activation+4.46e-01

Pos logit contributions
 fault+0.177
 enough+0.153
 pret+0.146
 pounds+0.145
paper+0.140
Neg logit contributions
date-0.197
 hurricane-0.195
Friend-0.183
 Amen-0.183
 tornado-0.181
Token not
Token ID407
Feature activation+4.61e-01

Pos logit contributions
 hurricane+0.030
date+0.030
 tornado+0.028
 television+0.028
 Amen+0.028
Neg logit contributions
 fault-0.025
 pret-0.022
paper-0.021
 loses-0.020
 enough-0.020
Token val
Token ID1188
Feature activation+4.68e-01

Pos logit contributions
 hurricane+0.030
date+0.029
 tornado+0.028
 year+0.028
 television+0.027
Neg logit contributions
 fault-0.026
 pret-0.024
paper-0.023
 loses-0.022
 puck-0.022
Tokenuing
Token ID4250
Feature activation+1.05e-01

Pos logit contributions
 hurricane+0.035
date+0.034
 tornado+0.034
 Amen+0.033
 year+0.033
Neg logit contributions
 fault-0.023
 pret-0.020
paper-0.019
 pounds-0.019
 loses-0.019
Token her
Token ID607
Feature activation-8.55e-03

Pos logit contributions
 hurricane+0.007
date+0.007
 tornado+0.006
 Amen+0.006
 year+0.006
Neg logit contributions
 fault-0.006
 pret-0.005
paper-0.005
 loses-0.005
 enough-0.005
Token"
Token ID1
Feature activation-1.43e-01

Pos logit contributions
 hurricane+0.012
date+0.011
 tornado+0.011
 television+0.011
 ranch+0.010
Neg logit contributions
ends-0.009
 pret-0.009
 loses-0.008
 Mir-0.008
 puck-0.008
TokenIt
Token ID1026
Feature activation-7.56e-01

Pos logit contributions
 fault+0.010
 pret+0.009
paper+0.009
 loses+0.008
 puck+0.008
Neg logit contributions
 hurricane-0.008
date-0.008
 tornado-0.007
 television-0.007
 year-0.007
Token was
Token ID373
Feature activation-1.64e-01

Pos logit contributions
 fault+0.041
 pret+0.036
paper+0.035
 puck+0.034
 pounds+0.034
Neg logit contributions
 hurricane-0.051
date-0.050
 tornado-0.049
 television-0.048
 year-0.048
Token not
Token ID407
Feature activation+2.43e-01

Pos logit contributions
 fault+0.009
 pret+0.008
 enough+0.008
 loses+0.008
 puck+0.008
Neg logit contributions
date-0.011
 hurricane-0.011
 year-0.010
 Amen-0.010
Friend-0.010
Token our
Token ID674
Feature activation+1.70e-01

Pos logit contributions
 hurricane+0.014
 television+0.014
 tornado+0.013
date+0.013
 voice+0.013
Neg logit contributions
 fault-0.015
 pret-0.014
paper-0.013
 loses-0.013
eter-0.013
Token fault
Token ID8046
Feature activation-2.99e+00

Pos logit contributions
date+0.008
wered+0.007
 Amen+0.007
ilt+0.007
Friend+0.007
Neg logit contributions
 fault-0.015
 pret-0.013
 enough-0.012
 hides-0.012
 teeth-0.011
Token,
Token ID11
Feature activation+1.32e-01

Pos logit contributions
 fault+0.189
 pret+0.171
paper+0.163
 loses+0.160
 puck+0.158
Neg logit contributions
 hurricane-0.178
date-0.175
 tornado-0.167
 television-0.162
 year-0.162
Token mom
Token ID1995
Feature activation+2.32e-01

Pos logit contributions
date+0.009
 hurricane+0.009
 Amen+0.008
 year+0.008
 tornado+0.008
Neg logit contributions
 fault-0.007
 pret-0.007
paper-0.006
 loses-0.006
 enough-0.006
Token.
Token ID13
Feature activation-1.28e-01

Pos logit contributions
 hurricane+0.013
date+0.013
 tornado+0.013
 year+0.012
 television+0.012
Neg logit contributions
 fault-0.015
 pret-0.014
paper-0.013
 puck-0.013
 loses-0.013
Token The
Token ID383
Feature activation+3.00e-01

Pos logit contributions
 fault+0.007
 pret+0.006
 loses+0.006
paper+0.006
 enough+0.006
Neg logit contributions
 hurricane-0.009
date-0.009
 tornado-0.008
 year-0.008
 television-0.008
Token phone
Token ID3072
Feature activation+3.28e-01

Pos logit contributions
date+0.019
cient+0.017
ilt+0.017
ir+0.017
Bruce+0.017
Neg logit contributions
 fault-0.021
 pret-0.019
drivers-0.015
 enough-0.015
 hides-0.015
Token.
Token ID13
Feature activation-1.26e-02

Pos logit contributions
 fault+0.048
 pret+0.045
paper+0.043
 loses+0.043
 puck+0.043
Neg logit contributions
 hurricane-0.021
date-0.020
 voice-0.019
 television-0.019
 tornado-0.019
Token "
Token ID366
Feature activation+1.84e-01

Pos logit contributions
 fault+0.001
 pret+0.001
paper+0.001
 loses+0.001
 pounds+0.001
Neg logit contributions
 hurricane-0.001
date-0.001
 tornado-0.001
 television-0.001
 voice-0.001
TokenIt
Token ID1026
Feature activation-5.61e-01

Pos logit contributions
 hurricane+0.009
date+0.008
 tornado+0.008
 television+0.008
 year+0.008
Neg logit contributions
 fault-0.014
 pret-0.013
paper-0.013
 loses-0.012
 puck-0.012
Token was
Token ID373
Feature activation-3.32e-02

Pos logit contributions
 fault+0.034
 pounds+0.030
paper+0.030
 pret+0.030
 puck+0.029
Neg logit contributions
 hurricane-0.034
date-0.033
 tornado-0.032
 voice-0.032
 television-0.032
Token my
Token ID616
Feature activation+2.43e-01

Pos logit contributions
 fault+0.002
 pret+0.002
 enough+0.002
 puck+0.001
 loses+0.001
Neg logit contributions
date-0.002
 hurricane-0.002
 tornado-0.002
 year-0.002
 Amen-0.002
Token fault
Token ID8046
Feature activation-2.99e+00

Pos logit contributions
 hurricane+0.015
date+0.015
 tornado+0.014
 television+0.014
 voice+0.014
Neg logit contributions
 fault-0.014
 pret-0.013
paper-0.013
 pounds-0.012
 loses-0.012
Token.
Token ID13
Feature activation+1.09e-01

Pos logit contributions
 fault+0.233
 pret+0.217
paper+0.208
 loses+0.205
 puck+0.203
Neg logit contributions
 hurricane-0.132
date-0.127
 tornado-0.121
 television-0.118
 year-0.117
Token I
Token ID314
Feature activation+3.08e-01

Pos logit contributions
 hurricane+0.007
date+0.007
 tornado+0.007
 year+0.007
ind+0.007
Neg logit contributions
 fault-0.006
 pret-0.006
paper-0.005
 loses-0.005
 puck-0.005
Token threw
Token ID9617
Feature activation-2.54e-03

Pos logit contributions
 hurricane+0.019
date+0.019
 year+0.018
 tornado+0.018
 Amen+0.017
Neg logit contributions
 fault-0.019
 pret-0.017
paper-0.016
 pounds-0.016
 puck-0.016
Token a
Token ID257
Feature activation-1.46e-03

Pos logit contributions
 fault+0.000
 pret+0.000
paper+0.000
 loses+0.000
 puck+0.000
Neg logit contributions
 hurricane-0.000
date-0.000
 tornado-0.000
 television-0.000
 year-0.000
Token rock
Token ID3881
Feature activation+9.84e-02

Pos logit contributions
 fault+0.000
 pret+0.000
 puck+0.000
paper+0.000
 loses+0.000
Neg logit contributions
date-0.000
 Amen-0.000
 hurricane-0.000
ind-0.000
ator-0.000
Token.
Token ID13
Feature activation+1.24e-01

Pos logit contributions
 hurricane+0.015
date+0.015
 tornado+0.014
 Amen+0.014
 television+0.014
Neg logit contributions
 fault-0.014
 pret-0.013
paper-0.012
 loses-0.012
 enough-0.012
Token It
Token ID632
Feature activation-7.46e-01

Pos logit contributions
 hurricane+0.008
date+0.008
 tornado+0.007
 Amen+0.007
 television+0.007
Neg logit contributions
 fault-0.007
 pret-0.007
paper-0.006
 puck-0.006
 pounds-0.006
Token is
Token ID318
Feature activation-1.35e-01

Pos logit contributions
 fault+0.041
paper+0.036
 pounds+0.035
 pret+0.035
 puck+0.035
Neg logit contributions
 hurricane-0.050
 Amen-0.047
 tornado-0.047
 voice-0.047
date-0.047
Token not
Token ID407
Feature activation+3.88e-01

Pos logit contributions
 fault+0.008
 enough+0.007
 pret+0.007
 loses+0.006
ws+0.006
Neg logit contributions
date-0.009
 hurricane-0.009
 year-0.008
han-0.008
Friend-0.008
Token your
Token ID534
Feature activation+4.54e-01

Pos logit contributions
 hurricane+0.022
 tornado+0.020
date+0.020
 television+0.020
 year+0.019
Neg logit contributions
 fault-0.025
 pret-0.024
paper-0.023
 puck-0.022
 loses-0.022
Token fault
Token ID8046
Feature activation-2.97e+00

Pos logit contributions
date+0.032
 hurricane+0.031
 Amen+0.029
Friend+0.029
iet+0.029
Neg logit contributions
 fault-0.027
 pret-0.022
 enough-0.020
paper-0.020
 loses-0.020
Token.
Token ID13
Feature activation+2.59e-01

Pos logit contributions
 fault+0.210
 pret+0.195
paper+0.186
 loses+0.183
 puck+0.182
Neg logit contributions
 hurricane-0.152
date-0.148
 tornado-0.142
 television-0.138
 year-0.138
Token It
Token ID632
Feature activation-5.78e-01

Pos logit contributions
 hurricane+0.016
date+0.015
 tornado+0.015
 Amen+0.015
 television+0.015
Neg logit contributions
 fault-0.016
 pret-0.014
paper-0.014
 puck-0.013
 pounds-0.013
Token is
Token ID318
Feature activation+5.38e-03

Pos logit contributions
 fault+0.031
paper+0.028
 pret+0.028
 pounds+0.027
 puck+0.027
Neg logit contributions
 hurricane-0.038
 tornado-0.036
 year-0.036
date-0.036
 Amen-0.036
Token not
Token ID407
Feature activation+4.78e-01

Pos logit contributions
date+0.000
 hurricane+0.000
Friend+0.000
 year+0.000
han+0.000
Neg logit contributions
 fault-0.000
 enough-0.000
 pret-0.000
 loses-0.000
ws-0.000
Token anyone
Token ID2687
Feature activation-1.05e+00

Pos logit contributions
 hurricane+0.027
 tornado+0.025
date+0.025
 television+0.025
 year+0.024
Neg logit contributions
 fault-0.031
 pret-0.029
paper-0.028
 puck-0.027
 loses-0.027
Token he
Token ID339
Feature activation-7.50e-01

Pos logit contributions
 fault+0.016
 pret+0.014
paper+0.013
 loses+0.013
 puck+0.013
Neg logit contributions
 hurricane-0.020
date-0.020
 tornado-0.019
 television-0.019
 year-0.019
Token knew
Token ID2993
Feature activation+1.71e-01

Pos logit contributions
 fault+0.047
 pret+0.042
paper+0.041
 pounds+0.039
 loses+0.039
Neg logit contributions
 hurricane-0.045
date-0.044
 year-0.042
 tornado-0.042
 voice-0.041
Token it
Token ID340
Feature activation-4.27e-02

Pos logit contributions
 hurricane+0.012
date+0.011
 tornado+0.011
 year+0.011
 television+0.011
Neg logit contributions
 fault-0.009
 pret-0.009
paper-0.008
 loses-0.008
 puck-0.008
Token was
Token ID373
Feature activation-3.70e-02

Pos logit contributions
 fault+0.003
 pret+0.003
 pounds+0.003
paper+0.003
 puck+0.003
Neg logit contributions
 hurricane-0.002
date-0.002
 year-0.002
 tornado-0.002
 television-0.002
Token his
Token ID465
Feature activation-2.71e-01

Pos logit contributions
 fault+0.002
 enough+0.002
 pret+0.002
paper+0.002
 loses+0.002
Neg logit contributions
date-0.002
 hurricane-0.002
 tornado-0.002
ind-0.002
iet-0.002
Token fault
Token ID8046
Feature activation-2.95e+00

Pos logit contributions
 fault+0.018
 pret+0.017
paper+0.016
 loses+0.016
 puck+0.016
Neg logit contributions
 hurricane-0.015
date-0.015
 tornado-0.014
 television-0.014
 year-0.014
Token.
Token ID13
Feature activation-1.11e-01

Pos logit contributions
 fault+0.202
 pret+0.182
paper+0.174
 enough+0.173
 puck+0.171
Neg logit contributions
 hurricane-0.160
date-0.159
 tornado-0.149
 television-0.144
 Amen-0.144
Token In
Token ID554
Feature activation-1.75e-01

Pos logit contributions
 fault+0.007
 pret+0.006
paper+0.006
 loses+0.006
 pounds+0.006
Neg logit contributions
 hurricane-0.007
date-0.006
 year-0.006
 Amen-0.006
 voice-0.006
Token the
Token ID262
Feature activation-2.44e-01

Pos logit contributions
 fault+0.010
 enough+0.009
 pret+0.009
 puck+0.008
 loses+0.008
Neg logit contributions
date-0.012
 hurricane-0.011
 tornado-0.011
ind-0.011
cient-0.011
Token end
Token ID886
Feature activation+1.82e-01

Pos logit contributions
 fault+0.018
 pret+0.016
 puck+0.016
 enough+0.016
 harder+0.015
Neg logit contributions
ind-0.012
cient-0.011
aby-0.011
han-0.011
 hurricane-0.011
Token,
Token ID11
Feature activation-3.71e-02

Pos logit contributions
 hurricane+0.010
date+0.010
 tornado+0.009
 television+0.009
 year+0.009
Neg logit contributions
 fault-0.013
 pret-0.011
paper-0.011
 loses-0.011
 puck-0.011
Token said
Token ID531
Feature activation+2.85e-01

Pos logit contributions
 fault+0.017
paper+0.015
 pret+0.015
 pounds+0.015
eter+0.014
Neg logit contributions
 hurricane-0.013
date-0.012
 voice-0.012
 tornado-0.012
 television-0.012
Token it
Token ID340
Feature activation+3.04e-02

Pos logit contributions
 hurricane+0.016
date+0.015
 tornado+0.015
 television+0.014
 year+0.014
Neg logit contributions
 fault-0.019
 pret-0.018
paper-0.017
 puck-0.017
 loses-0.017
Token was
Token ID373
Feature activation+1.21e-01

Pos logit contributions
 voice+0.001
date+0.001
 year+0.001
 hurricane+0.001
 television+0.001
Neg logit contributions
 fault-0.002
 pounds-0.002
 puck-0.002
 pret-0.002
 loses-0.002
Token not
Token ID407
Feature activation+2.92e-01

Pos logit contributions
date+0.010
 hurricane+0.010
 year+0.010
 tornado+0.010
Friend+0.009
Neg logit contributions
 fault-0.005
 enough-0.004
 pret-0.004
 teeth-0.004
ws-0.003
Token his
Token ID465
Feature activation-2.49e-01

Pos logit contributions
 hurricane+0.016
 television+0.015
 tornado+0.015
 voice+0.015
date+0.015
Neg logit contributions
 fault-0.018
 pret-0.017
paper-0.017
 puck-0.017
 loses-0.017
Token fault
Token ID8046
Feature activation-2.95e+00

Pos logit contributions
 fault+0.010
 pret+0.008
 loses+0.007
 enough+0.007
paper+0.007
Neg logit contributions
date-0.021
 hurricane-0.021
 tornado-0.020
 year-0.020
 Amen-0.020
Token.
Token ID13
Feature activation+2.27e-01

Pos logit contributions
 fault+0.244
 pret+0.235
paper+0.224
 loses+0.220
 puck+0.220
Neg logit contributions
 hurricane-0.113
date-0.106
 tornado-0.103
 television-0.102
 year-0.100
Token She
Token ID1375
Feature activation-1.42e-02

Pos logit contributions
 hurricane+0.012
date+0.012
 tornado+0.012
 year+0.011
 television+0.011
Neg logit contributions
 fault-0.015
 pret-0.014
paper-0.013
 loses-0.013
 puck-0.013
Token said
Token ID531
Feature activation-2.89e-02

Pos logit contributions
 fault+0.001
paper+0.001
 pounds+0.001
 pret+0.001
 puck+0.001
Neg logit contributions
 hurricane-0.001
 year-0.001
 voice-0.001
Friend-0.001
date-0.001
Token the
Token ID262
Feature activation+2.20e-01

Pos logit contributions
 fault+0.002
 pret+0.002
eter+0.002
 puck+0.002
ws+0.002
Neg logit contributions
 hurricane-0.002
 tornado-0.002
 television-0.002
 voice-0.002
 Amen-0.002
Token bird
Token ID6512
Feature activation-3.28e-01

Pos logit contributions
date+0.013
ind+0.012
iet+0.012
 hurricane+0.011
Friend+0.011
Neg logit contributions
 fault-0.016
 pret-0.014
 enough-0.012
 loses-0.012
 puck-0.012
Token but
Token ID475
Feature activation+4.01e-01

Pos logit contributions
 hurricane+0.006
date+0.006
 tornado+0.006
 year+0.006
 television+0.006
Neg logit contributions
 fault-0.005
 pret-0.004
paper-0.004
 loses-0.004
 puck-0.004
Token you
Token ID345
Feature activation+1.14e+00

Pos logit contributions
date+0.025
 hurricane+0.025
 tornado+0.023
 year+0.023
 television+0.023
Neg logit contributions
 fault-0.024
 pret-0.022
paper-0.021
 enough-0.020
 loses-0.020
Token have
Token ID423
Feature activation+3.56e-01

Pos logit contributions
 hurricane+0.071
date+0.070
 tornado+0.066
 television+0.065
 year+0.065
Neg logit contributions
 fault-0.069
 pret-0.062
paper-0.059
 loses-0.059
 enough-0.058
Token to
Token ID284
Feature activation+2.60e-01

Pos logit contributions
 hurricane+0.022
 year+0.021
date+0.021
 me+0.021
 tornado+0.021
Neg logit contributions
 pret-0.020
 fault-0.019
paper-0.018
 pounds-0.017
 puck-0.017
Token serve
Token ID4691
Feature activation+5.13e-01

Pos logit contributions
 hurricane+0.019
 tornado+0.019
date+0.019
 television+0.018
 musician+0.018
Neg logit contributions
 fault-0.013
 enough-0.011
 pret-0.011
ws-0.011
 puts-0.010
Token me
Token ID502
Feature activation+1.98e+00

Pos logit contributions
 hurricane+0.032
date+0.030
 Amen+0.029
 tornado+0.029
iet+0.028
Neg logit contributions
 fault-0.031
 enough-0.029
 pret-0.028
ws-0.027
 loses-0.027
Token first
Token ID717
Feature activation+1.08e+00

Pos logit contributions
 hurricane+0.149
date+0.146
 tornado+0.142
 Amen+0.138
 television+0.137
Neg logit contributions
 fault-0.093
 pret-0.081
paper-0.076
 enough-0.076
 loses-0.074
Token."
Token ID526
Feature activation-2.40e-01

Pos logit contributions
 hurricane+0.064
date+0.062
 tornado+0.060
 television+0.058
 Amen+0.058
Neg logit contributions
 fault-0.069
 pret-0.062
paper-0.059
 enough-0.059
 loses-0.058
Token Tim
Token ID5045
Feature activation-4.50e-01

Pos logit contributions
 fault+0.013
 pret+0.012
paper+0.011
 loses+0.011
 puck+0.011
Neg logit contributions
 hurricane-0.016
date-0.016
 tornado-0.016
 television-0.015
 year-0.015
Tokenmy
Token ID1820
Feature activation+7.13e-01

Pos logit contributions
 fault+0.031
 pret+0.028
 loses+0.026
paper+0.026
 pounds+0.026
Neg logit contributions
date-0.024
 hurricane-0.024
 year-0.024
 voice-0.023
iet-0.023
Token didn
Token ID1422
Feature activation+1.43e-01

Pos logit contributions
date+0.040
 year+0.040
 hurricane+0.040
 voice+0.039
Friend+0.038
Neg logit contributions
 fault-0.048
 pret-0.040
paper-0.040
 pounds-0.039
 loses-0.039
Token how
Token ID703
Feature activation+5.92e-02

Pos logit contributions
 hurricane+0.030
date+0.030
 television+0.029
 tornado+0.029
 year+0.028
Neg logit contributions
 fault-0.023
 pret-0.022
paper-0.021
 loses-0.020
 pounds-0.020
Token to
Token ID284
Feature activation+7.95e-01

Pos logit contributions
 hurricane+0.004
date+0.004
 tornado+0.004
 television+0.004
 year+0.004
Neg logit contributions
 fault-0.003
 pret-0.003
paper-0.002
 loses-0.002
 puck-0.002
Token play
Token ID711
Feature activation+1.02e+00

Pos logit contributions
 hurricane+0.054
 Amen+0.053
date+0.052
 tornado+0.051
 messenger+0.051
Neg logit contributions
 fault-0.042
 pret-0.041
 puck-0.039
 loses-0.037
 puts-0.036
Token hide
Token ID7808
Feature activation+7.67e-01

Pos logit contributions
 hurricane+0.061
date+0.058
 tornado+0.057
 Amen+0.056
ind+0.055
Neg logit contributions
 fault-0.064
 pret-0.060
paper-0.056
 puck-0.055
 loses-0.055
Token and
Token ID290
Feature activation-7.74e-01

Pos logit contributions
 hurricane+0.045
date+0.044
 tornado+0.042
 Amen+0.041
 television+0.041
Neg logit contributions
 fault-0.049
 pret-0.045
paper-0.043
 loses-0.042
 puck-0.041
Token seek
Token ID5380
Feature activation+2.15e+00

Pos logit contributions
 fault+0.047
 pret+0.042
paper+0.040
 loses+0.039
 puck+0.039
Neg logit contributions
 hurricane-0.048
date-0.047
 tornado-0.045
 television-0.044
 year-0.044
Token instead
Token ID2427
Feature activation+1.00e-01

Pos logit contributions
date+0.103
 hurricane+0.103
 tornado+0.097
 year+0.095
 television+0.095
Neg logit contributions
 fault-0.158
 pret-0.145
paper-0.139
 loses-0.137
 puck-0.136
Token.
Token ID13
Feature activation-9.26e-02

Pos logit contributions
 hurricane+0.005
 Amen+0.005
date+0.005
ind+0.005
 tornado+0.005
Neg logit contributions
 fault-0.007
 pret-0.006
 enough-0.006
 loses-0.006
paper-0.006
Token Lily
Token ID20037
Feature activation-2.02e-01

Pos logit contributions
 fault+0.006
 pret+0.005
paper+0.005
 pounds+0.005
 loses+0.005
Neg logit contributions
date-0.006
 hurricane-0.005
 television-0.005
 tornado-0.005
 year-0.005
Token liked
Token ID8288
Feature activation+2.13e-01

Pos logit contributions
 fault+0.015
 pounds+0.013
paper+0.013
 pret+0.012
 loses+0.012
Neg logit contributions
 year-0.010
 voice-0.010
date-0.009
 me-0.009
Friend-0.009
Token this
Token ID428
Feature activation-5.75e-01

Pos logit contributions
 hurricane+0.012
 television+0.012
 tornado+0.012
date+0.011
 year+0.011
Neg logit contributions
 fault-0.014
 pret-0.012
 pounds-0.012
 loses-0.012
eter-0.012
Token Soup
Token ID34011
Feature activation+2.57e-01

Pos logit contributions
 hurricane+0.007
date+0.007
 tornado+0.007
 Amen+0.007
 television+0.007
Neg logit contributions
 fault-0.007
 pret-0.006
paper-0.006
 loses-0.006
 puck-0.006
Token is
Token ID318
Feature activation-2.58e-01

Pos logit contributions
 hurricane+0.016
 year+0.016
date+0.016
 tornado+0.015
 television+0.015
Neg logit contributions
 fault-0.015
 pret-0.013
paper-0.012
 puck-0.012
eter-0.012
Token better
Token ID1365
Feature activation-6.09e-01

Pos logit contributions
 fault+0.015
 pret+0.014
 enough+0.013
paper+0.012
 puck+0.012
Neg logit contributions
date-0.017
 hurricane-0.017
 tornado-0.016
 year-0.016
 Amen-0.015
Token.
Token ID13
Feature activation+2.50e-01

Pos logit contributions
 fault+0.030
 pret+0.026
paper+0.025
 puck+0.024
 loses+0.024
Neg logit contributions
 hurricane-0.044
date-0.044
 tornado-0.042
 year-0.042
 television-0.041
Token Let
Token ID3914
Feature activation+1.94e-01

Pos logit contributions
 hurricane+0.016
date+0.016
 tornado+0.015
 Amen+0.015
 television+0.015
Neg logit contributions
 fault-0.014
 pret-0.013
paper-0.012
 loses-0.012
 puck-0.012
Token me
Token ID502
Feature activation+2.09e+00

Pos logit contributions
 hurricane+0.014
date+0.014
 Amen+0.014
ind+0.014
 tornado+0.013
Neg logit contributions
 fault-0.009
 enough-0.009
 pret-0.008
 loses-0.008
 puts-0.008
Token have
Token ID423
Feature activation+1.40e-01

Pos logit contributions
 hurricane+0.125
date+0.120
 tornado+0.119
 television+0.113
 year+0.113
Neg logit contributions
 fault-0.131
 pret-0.121
paper-0.113
 loses-0.112
 enough-0.111
Token some
Token ID617
Feature activation+1.62e-01

Pos logit contributions
 me+0.009
 year+0.009
 hurricane+0.009
 television+0.008
 others+0.008
Neg logit contributions
 fault-0.008
 pret-0.007
paper-0.007
drivers-0.007
 puck-0.006
Token!"
Token ID2474
Feature activation-9.60e-02

Pos logit contributions
 hurricane+0.011
date+0.011
 tornado+0.010
 television+0.010
ind+0.010
Neg logit contributions
 fault-0.009
 pret-0.008
paper-0.007
 enough-0.007
 loses-0.007
Token
Token ID198
Feature activation+2.12e-01

Pos logit contributions
 fault+0.007
 pret+0.007
paper+0.006
 pounds+0.006
 puck+0.006
Neg logit contributions
 hurricane-0.005
date-0.004
 tornado-0.004
 television-0.004
 year-0.004
Token
Token ID198
Feature activation+1.08e-01

Pos logit contributions
 hurricane+0.013
date+0.013
 tornado+0.012
 television+0.011
 year+0.011
Neg logit contributions
 fault-0.013
 pret-0.012
 loses-0.012
paper-0.012
 puck-0.011
Token are
Token ID389
Feature activation-4.42e-01

Pos logit contributions
 fault+0.028
 pret+0.025
paper+0.024
 puck+0.023
 loses+0.023
Neg logit contributions
 hurricane-0.041
date-0.040
 tornado-0.039
 year-0.038
 television-0.038
Token unt
Token ID1418
Feature activation-6.00e-01

Pos logit contributions
 fault+0.028
 pret+0.026
paper+0.025
 loses+0.024
 puck+0.024
Neg logit contributions
 hurricane-0.025
date-0.024
 tornado-0.024
 television-0.023
 year-0.023
Tokenied
Token ID798
Feature activation+1.06e+00

Pos logit contributions
 fault+0.045
 pret+0.042
paper+0.040
 loses+0.039
 puck+0.039
Neg logit contributions
 hurricane-0.028
date-0.028
 tornado-0.026
 year-0.026
 television-0.025
Token.
Token ID13
Feature activation-1.85e-01

Pos logit contributions
 hurricane+0.056
date+0.055
 tornado+0.052
 Amen+0.051
 television+0.049
Neg logit contributions
 fault-0.073
 pret-0.067
 enough-0.065
paper-0.065
 loses-0.064
Token Let
Token ID3914
Feature activation+1.02e-01

Pos logit contributions
 fault+0.010
 pret+0.010
paper+0.009
 loses+0.009
 puck+0.009
Neg logit contributions
 hurricane-0.012
date-0.012
 tornado-0.011
 year-0.011
 television-0.011
Token me
Token ID502
Feature activation+1.98e+00

Pos logit contributions
 Amen+0.008
 hurricane+0.008
ind+0.008
date+0.008
wered+0.008
Neg logit contributions
 enough-0.004
 fault-0.004
 loses-0.003
 puts-0.003
 pret-0.003
Token help
Token ID1037
Feature activation+9.87e-01

Pos logit contributions
 hurricane+0.122
 tornado+0.116
date+0.114
 television+0.110
 year+0.109
Neg logit contributions
 fault-0.122
 pret-0.114
paper-0.107
 loses-0.107
 enough-0.106
Token you
Token ID345
Feature activation+1.08e+00

Pos logit contributions
 hurricane+0.063
date+0.060
iet+0.058
 tornado+0.058
 Amen+0.057
Neg logit contributions
 fault-0.056
 pret-0.054
paper-0.052
 loses-0.052
 enough-0.051
Token tie
Token ID9839
Feature activation+2.66e-01

Pos logit contributions
 hurricane+0.057
date+0.056
 tornado+0.053
 television+0.052
 year+0.052
Neg logit contributions
 fault-0.075
 pret-0.069
paper-0.065
 loses-0.064
 puck-0.064
Token them
Token ID606
Feature activation+6.68e-01

Pos logit contributions
 hurricane+0.016
date+0.016
 tornado+0.015
 year+0.015
 Amen+0.015
Neg logit contributions
 fault-0.017
 pret-0.015
paper-0.014
 loses-0.014
 enough-0.014
Token up
Token ID510
Feature activation+8.15e-01

Pos logit contributions
date+0.032
 hurricane+0.031
 television+0.030
ind+0.030
 tornado+0.030
Neg logit contributions
 fault-0.046
 pret-0.044
 pounds-0.042
 puck-0.041
 loses-0.041
Token you
Token ID345
Feature activation+1.01e+00

Pos logit contributions
 hurricane+0.048
date+0.048
 tornado+0.046
 Amen+0.045
 year+0.045
Neg logit contributions
 fault-0.030
 pret-0.027
paper-0.025
 loses-0.024
 enough-0.024
Token to
Token ID284
Feature activation+1.69e-01

Pos logit contributions
date+0.061
 hurricane+0.060
 year+0.059
 television+0.059
 voice+0.058
Neg logit contributions
 fault-0.058
 pret-0.056
 pounds-0.052
 puck-0.050
paper-0.050
Token paint
Token ID7521
Feature activation+4.34e-01

Pos logit contributions
ind+0.012
 hurricane+0.012
iet+0.012
 Amen+0.012
 tornado+0.012
Neg logit contributions
 fault-0.008
 pret-0.008
onna-0.008
 loses-0.007
 puck-0.007
Token something
Token ID1223
Feature activation-2.28e-01

Pos logit contributions
 hurricane+0.026
date+0.026
 tornado+0.025
 year+0.024
 television+0.024
Neg logit contributions
 fault-0.027
 pret-0.024
paper-0.023
 loses-0.023
 puck-0.023
Token for
Token ID329
Feature activation+9.00e-01

Pos logit contributions
 fault+0.014
 pret+0.013
paper+0.012
 loses+0.012
 enough+0.012
Neg logit contributions
 hurricane-0.014
date-0.014
 tornado-0.013
 Amen-0.013
iet-0.013
Token me
Token ID502
Feature activation+2.08e+00

Pos logit contributions
 hurricane+0.060
date+0.057
 tornado+0.056
cient+0.055
iet+0.055
Neg logit contributions
 enough-0.046
 fault-0.046
 pret-0.045
 pounds-0.044
 loses-0.043
Token.
Token ID13
Feature activation+2.10e-01

Pos logit contributions
 hurricane+0.105
date+0.102
 tornado+0.097
 television+0.093
 year+0.092
Neg logit contributions
 fault-0.151
 pret-0.139
paper-0.133
 loses-0.132
 enough-0.130
Token Maybe
Token ID6674
Feature activation-7.82e-02

Pos logit contributions
 hurricane+0.014
date+0.013
 tornado+0.013
 year+0.013
 television+0.013
Neg logit contributions
 fault-0.012
 pret-0.011
paper-0.010
 loses-0.010
 puck-0.010
Token you
Token ID345
Feature activation+1.43e+00

Pos logit contributions
 fault+0.005
 pret+0.005
paper+0.004
 loses+0.004
 puck+0.004
Neg logit contributions
 hurricane-0.004
date-0.004
 tornado-0.004
 television-0.004
 year-0.004
Token can
Token ID460
Feature activation+5.01e-02

Pos logit contributions
 hurricane+0.107
date+0.106
 tornado+0.102
 television+0.100
 year+0.100
Neg logit contributions
 fault-0.067
 pret-0.059
paper-0.055
 loses-0.053
 puck-0.053
Token paint
Token ID7521
Feature activation+4.73e-01

Pos logit contributions
 hurricane+0.004
 Amen+0.003
 tornado+0.003
iet+0.003
 messenger+0.003
Neg logit contributions
 fault-0.003
 pret-0.002
 loses-0.002
 puts-0.002
paper-0.002
Token good
Token ID922
Feature activation+4.52e-02

Pos logit contributions
date+0.074
 hurricane+0.073
 tornado+0.071
 television+0.067
han+0.067
Neg logit contributions
 fault-0.070
 pret-0.067
paper-0.063
 loses-0.061
ends-0.061
Token at
Token ID379
Feature activation-2.17e-05

Pos logit contributions
 hurricane+0.003
date+0.003
 tornado+0.002
 television+0.002
 year+0.002
Neg logit contributions
 fault-0.003
 pret-0.003
paper-0.003
 loses-0.003
 puck-0.003
Token drawing
Token ID8263
Feature activation-1.33e-01

Pos logit contributions
 fault+0.000
 pret+0.000
 pounds+0.000
 loses+0.000
paper+0.000
Neg logit contributions
 hurricane-0.000
date-0.000
 television-0.000
 tornado-0.000
 year-0.000
Token.
Token ID13
Feature activation+3.63e-01

Pos logit contributions
 fault+0.009
 pret+0.009
paper+0.009
 enough+0.008
 puck+0.008
Neg logit contributions
 hurricane-0.007
date-0.007
 tornado-0.006
 Amen-0.006
ind-0.006
Token
Token ID198
Feature activation+8.12e-01

Pos logit contributions
 hurricane+0.019
date+0.019
 tornado+0.018
 year+0.018
 Amen+0.018
Neg logit contributions
 fault-0.025
 pret-0.023
paper-0.022
 loses-0.022
 pounds-0.022
Token
Token ID198
Feature activation+5.92e-01

Pos logit contributions
 hurricane+0.045
date+0.044
 tornado+0.042
 television+0.041
ind+0.040
Neg logit contributions
 fault-0.053
 pret-0.049
paper-0.047
 loses-0.046
 puck-0.046
TokenOne
Token ID3198
Feature activation-3.47e-02

Pos logit contributions
 hurricane+0.026
date+0.026
 tornado+0.024
 television+0.023
 year+0.023
Neg logit contributions
 fault-0.046
 pret-0.043
paper-0.041
 loses-0.040
 puck-0.040
Token day
Token ID1110
Feature activation+3.06e-01

Pos logit contributions
 fault+0.003
 pret+0.003
paper+0.003
 loses+0.003
 puck+0.003
Neg logit contributions
 hurricane-0.001
date-0.001
 tornado-0.001
 Amen-0.001
 television-0.001
Token,
Token ID11
Feature activation+2.95e-01

Pos logit contributions
 hurricane+0.013
 tornado+0.012
 year+0.011
 voice+0.011
date+0.011
Neg logit contributions
 fault-0.028
eter-0.024
ws-0.024
 pret-0.023
 puck-0.023
Token Mark
Token ID2940
Feature activation+2.72e-01

Pos logit contributions
 hurricane+0.018
 tornado+0.018
 year+0.017
 voice+0.017
 television+0.017
Neg logit contributions
 fault-0.020
ws-0.017
eter-0.016
 pret-0.016
onna-0.016
Token wants
Token ID3382
Feature activation+7.55e-02

Pos logit contributions
 hurricane+0.017
date+0.017
 year+0.017
 musician+0.017
ind+0.017
Neg logit contributions
 fault-0.015
 pounds-0.013
paper-0.013
 pret-0.012
eter-0.012
Token and
Token ID290
Feature activation+6.92e-01

Pos logit contributions
 fault+0.023
 pret+0.020
 enough+0.020
 loses+0.019
paper+0.019
Neg logit contributions
 hurricane-0.025
date-0.024
 tornado-0.023
 Amen-0.022
Friend-0.022
Token go
Token ID467
Feature activation+9.26e-02

Pos logit contributions
 hurricane+0.041
date+0.039
 tornado+0.038
 television+0.037
 voice+0.036
Neg logit contributions
 fault-0.044
 pret-0.040
paper-0.039
 pounds-0.037
 puck-0.037
Token to
Token ID284
Feature activation+4.11e-01

Pos logit contributions
 hurricane+0.006
date+0.006
 tornado+0.006
iet+0.006
 Amen+0.006
Neg logit contributions
 fault-0.005
 enough-0.005
 pret-0.004
 loses-0.004
paper-0.004
Token your
Token ID534
Feature activation+1.06e+00

Pos logit contributions
 hurricane+0.024
date+0.023
 tornado+0.023
 television+0.022
 year+0.022
Neg logit contributions
 fault-0.026
 pret-0.024
paper-0.023
 pounds-0.022
 puck-0.022
Token room
Token ID2119
Feature activation+1.29e+00

Pos logit contributions
 hurricane+0.040
date+0.038
 tornado+0.036
 television+0.035
 year+0.035
Neg logit contributions
 fault-0.089
 pret-0.084
paper-0.082
 pounds-0.080
 loses-0.080
Token.
Token ID13
Feature activation+2.85e-01

Pos logit contributions
 hurricane+0.064
date+0.063
 tornado+0.060
 year+0.059
 television+0.058
Neg logit contributions
 fault-0.092
 pret-0.085
paper-0.082
 puck-0.080
 loses-0.080
Token You
Token ID921
Feature activation+9.03e-01

Pos logit contributions
 hurricane+0.018
date+0.018
 tornado+0.017
 year+0.017
 television+0.017
Neg logit contributions
 fault-0.016
 pret-0.015
paper-0.014
 loses-0.014
 puck-0.014
Token are
Token ID389
Feature activation+1.96e-02

Pos logit contributions
 hurricane+0.058
date+0.057
 tornado+0.055
 year+0.054
 television+0.053
Neg logit contributions
 fault-0.052
 pret-0.047
paper-0.045
 puck-0.044
 pounds-0.044
Token in
Token ID287
Feature activation+2.75e-01

Pos logit contributions
date+0.001
 hurricane+0.001
 year+0.001
iet+0.001
 tornado+0.001
Neg logit contributions
 fault-0.001
 pret-0.001
 enough-0.001
 loses-0.001
 puck-0.001
Token big
Token ID1263
Feature activation+4.50e-02

Pos logit contributions
 hurricane+0.020
date+0.020
 tornado+0.019
 television+0.019
 year+0.019
Neg logit contributions
 fault-0.014
 pret-0.012
paper-0.011
 loses-0.011
 enough-0.011
Token trouble
Token ID5876
Feature activation-1.48e+00

Pos logit contributions
 hurricane+0.003
date+0.003
 tornado+0.003
 television+0.003
 voice+0.003
Neg logit contributions
 fault-0.002
 pret-0.002
paper-0.002
 puck-0.002
eter-0.002
TokenLet
Token ID5756
Feature activation+5.32e-02

Pos logit contributions
 hurricane+0.001
 television+0.001
 tornado+0.001
date+0.001
 ranch+0.001
Neg logit contributions
 fault-0.001
 pret-0.001
paper-0.001
eter-0.001
 puck-0.001
Token's
Token ID338
Feature activation-3.39e-01

Pos logit contributions
 hurricane+0.004
 Amen+0.004
date+0.004
ind+0.004
 tornado+0.004
Neg logit contributions
 enough-0.002
 fault-0.002
 loses-0.002
 pret-0.002
 puts-0.002
Token do
Token ID466
Feature activation+5.25e-02

Pos logit contributions
 fault+0.015
 enough+0.015
 puts+0.014
 pret+0.013
 puck+0.013
Neg logit contributions
ind-0.025
iet-0.024
 year-0.023
aby-0.023
Friend-0.023
Token something
Token ID1223
Feature activation-6.68e-02

Pos logit contributions
 hurricane+0.003
 tornado+0.003
 television+0.003
date+0.003
 year+0.003
Neg logit contributions
 fault-0.003
 pret-0.003
paper-0.003
eter-0.003
 puck-0.003
Token else
Token ID2073
Feature activation-1.22e-01

Pos logit contributions
 fault+0.004
 pret+0.004
paper+0.004
 puck+0.003
eter+0.003
Neg logit contributions
 hurricane-0.004
date-0.004
 television-0.004
 tornado-0.004
 year-0.004
Token:
Token ID25
Feature activation+8.42e-01

Pos logit contributions
 fault+0.008
 pret+0.008
paper+0.007
 puck+0.007
ws+0.007
Neg logit contributions
 hurricane-0.006
date-0.006
 television-0.006
 tornado-0.006
 year-0.006
Token we
Token ID356
Feature activation+2.69e-01

Pos logit contributions
 hurricane+0.055
date+0.053
 tornado+0.052
 television+0.052
 year+0.050
Neg logit contributions
 fault-0.048
 pret-0.043
paper-0.041
 loses-0.040
 pounds-0.039
Token can
Token ID460
Feature activation-1.23e-01

Pos logit contributions
 hurricane+0.023
ind+0.022
 tornado+0.022
date+0.022
han+0.022
Neg logit contributions
 fault-0.009
 pret-0.008
 enough-0.008
 loses-0.007
ws-0.007
Token splash
Token ID22870
Feature activation+1.92e-01

Pos logit contributions
 fault+0.006
 pret+0.005
 enough+0.005
 puck+0.005
 loses+0.005
Neg logit contributions
 hurricane-0.008
iet-0.008
 tornado-0.008
ind-0.008
date-0.008
Token in
Token ID287
Feature activation+6.09e-02

Pos logit contributions
 hurricane+0.011
date+0.010
 tornado+0.010
 television+0.010
 year+0.010
Neg logit contributions
 fault-0.013
 pret-0.012
paper-0.011
 loses-0.011
 puck-0.011
Token the
Token ID262
Feature activation-2.71e-01

Pos logit contributions
 hurricane+0.004
date+0.004
 tornado+0.004
 television+0.003
 year+0.003
Neg logit contributions
 fault-0.004
 pret-0.003
paper-0.003
 loses-0.003
 puck-0.003
Token make
Token ID787
Feature activation-6.62e-02

Pos logit contributions
 pret+0.010
 puts+0.009
 fault+0.008
ws+0.008
 enough+0.008
Neg logit contributions
 hurricane-0.013
ind-0.013
 tornado-0.013
iet-0.012
han-0.012
Token some
Token ID617
Feature activation+1.45e-01

Pos logit contributions
 fault+0.004
 pret+0.003
 loses+0.003
paper+0.003
 puck+0.003
Neg logit contributions
date-0.004
 television-0.004
 hurricane-0.004
 year-0.004
 tornado-0.004
Token cocoa
Token ID35845
Feature activation+4.88e-02

Pos logit contributions
 hurricane+0.010
date+0.010
 tornado+0.010
 Amen+0.010
 year+0.010
Neg logit contributions
 fault-0.007
 pret-0.007
paper-0.006
 loses-0.006
 puck-0.006
Token.
Token ID13
Feature activation+3.66e-01

Pos logit contributions
 hurricane+0.002
date+0.002
 television+0.002
 tornado+0.002
 year+0.002
Neg logit contributions
 fault-0.004
 pret-0.003
 loses-0.003
paper-0.003
 puck-0.003
Token
Token ID220
Feature activation+8.13e-01

Pos logit contributions
 hurricane+0.020
date+0.020
 tornado+0.019
 year+0.018
 television+0.018
Neg logit contributions
 fault-0.025
 pret-0.023
paper-0.022
 loses-0.021
 puck-0.021
Token
Token ID198
Feature activation+5.21e-01

Pos logit contributions
 hurricane+0.043
date+0.042
 tornado+0.040
 television+0.039
 year+0.039
Neg logit contributions
 fault-0.056
 pret-0.051
paper-0.049
 loses-0.048
 puck-0.048
Token
Token ID198
Feature activation+3.21e-01

Pos logit contributions
 hurricane+0.030
date+0.029
 tornado+0.028
 television+0.027
 Amen+0.027
Neg logit contributions
 fault-0.033
 pret-0.031
paper-0.029
 loses-0.029
 puck-0.028
TokenWhen
Token ID2215
Feature activation-1.62e-01

Pos logit contributions
 hurricane+0.023
 tornado+0.021
 Amen+0.021
date+0.021
 television+0.021
Neg logit contributions
 fault-0.015
 pret-0.015
 puck-0.014
 loses-0.013
paper-0.013
Token her
Token ID607
Feature activation+3.88e-01

Pos logit contributions
 fault+0.010
 pret+0.009
 pounds+0.009
paper+0.009
 loses+0.008
Neg logit contributions
 hurricane-0.010
date-0.010
 tornado-0.010
 television-0.010
 year-0.010
Token mom
Token ID1995
Feature activation+2.61e-01

Pos logit contributions
 hurricane+0.018
date+0.017
 tornado+0.016
 television+0.016
 year+0.016
Neg logit contributions
 fault-0.030
 pret-0.028
paper-0.027
 loses-0.026
 puck-0.026
Tokenmy
Token ID1820
Feature activation+8.52e-01

Pos logit contributions
date+0.013
 year+0.012
 hurricane+0.012
 television+0.012
Friend+0.012
Neg logit contributions
 fault-0.020
 pret-0.017
 loses-0.017
 pounds-0.017
 puck-0.017
Token They
Token ID1119
Feature activation+7.95e-02

Pos logit contributions
 fault+0.002
 pret+0.002
paper+0.002
 loses+0.002
 puck+0.002
Neg logit contributions
 hurricane-0.003
date-0.002
 tornado-0.002
 year-0.002
 television-0.002
Token sat
Token ID3332
Feature activation-5.11e-01

Pos logit contributions
 hurricane+0.005
date+0.005
 tornado+0.005
 television+0.005
 year+0.005
Neg logit contributions
 fault-0.004
 pret-0.004
paper-0.004
 loses-0.004
 enough-0.004
Token down
Token ID866
Feature activation-4.42e-01

Pos logit contributions
 fault+0.032
 pret+0.029
paper+0.028
 loses+0.027
 puck+0.027
Neg logit contributions
 hurricane-0.030
date-0.030
 tornado-0.028
 television-0.028
 year-0.028
Token and
Token ID290
Feature activation-3.99e-02

Pos logit contributions
 fault+0.026
 pret+0.024
paper+0.023
 loses+0.022
 puck+0.022
Neg logit contributions
 hurricane-0.028
date-0.027
 tornado-0.026
 television-0.026
 year-0.026
Token worked
Token ID3111
Feature activation+5.22e-01

Pos logit contributions
 fault+0.003
 pounds+0.002
paper+0.002
 pret+0.002
 puck+0.002
Neg logit contributions
 hurricane-0.002
date-0.002
 television-0.002
 tornado-0.002
 Amen-0.002
Token together
Token ID1978
Feature activation+9.46e-01

Pos logit contributions
 hurricane+0.037
date+0.036
 year+0.035
 tornado+0.035
 television+0.035
Neg logit contributions
 fault-0.027
 pret-0.024
paper-0.023
 loses-0.022
 puck-0.022
Token.
Token ID13
Feature activation+8.10e-02

Pos logit contributions
 hurricane+0.051
date+0.051
 tornado+0.048
 year+0.048
 television+0.048
Neg logit contributions
 fault-0.063
 pret-0.059
paper-0.056
 loses-0.055
 puck-0.055
Token They
Token ID1119
Feature activation+8.32e-02

Pos logit contributions
 hurricane+0.005
date+0.005
 tornado+0.005
 year+0.005
 television+0.005
Neg logit contributions
 fault-0.005
 pret-0.004
paper-0.004
 loses-0.004
 puck-0.004
Token tied
Token ID8165
Feature activation-4.98e-02

Pos logit contributions
 hurricane+0.005
date+0.005
 year+0.005
 tornado+0.005
Friend+0.005
Neg logit contributions
 fault-0.005
 pret-0.005
 pounds-0.004
paper-0.004
 puck-0.004
Token the
Token ID262
Feature activation+3.43e-01

Pos logit contributions
 fault+0.003
 loses+0.003
 pret+0.003
 pounds+0.003
 weighs+0.003
Neg logit contributions
 hurricane-0.002
 television-0.002
 tornado-0.002
 voice-0.002
date-0.002
Token hair
Token ID4190
Feature activation+8.88e-01

Pos logit contributions
 hurricane+0.021
date+0.021
 Amen+0.020
 tornado+0.020
Friend+0.019
Neg logit contributions
 fault-0.021
 pret-0.020
paper-0.018
 puck-0.018
 loses-0.017
Token tire
Token ID15867
Feature activation-9.00e-02

Pos logit contributions
date+0.013
cient+0.011
ilt+0.011
 Amen+0.011
iet+0.011
Neg logit contributions
 fault-0.014
 cones-0.014
 puck-0.014
 pret-0.013
 loses-0.012
Token swing
Token ID9628
Feature activation+4.52e-01

Pos logit contributions
 fault+0.005
 cones+0.005
 enough+0.005
 puck+0.005
paper+0.005
Neg logit contributions
 Amen-0.005
 hurricane-0.005
ind-0.005
iet-0.005
date-0.005
Token.
Token ID13
Feature activation+2.59e-02

Pos logit contributions
 Amen+0.027
 hurricane+0.027
ind+0.025
 messenger+0.025
 tornado+0.024
Neg logit contributions
 fault-0.028
 enough-0.027
paper-0.025
 pret-0.025
 puck-0.025
Token She
Token ID1375
Feature activation+2.06e-01

Pos logit contributions
 hurricane+0.002
date+0.002
 tornado+0.002
ind+0.002
iet+0.002
Neg logit contributions
 fault-0.002
 pret-0.001
paper-0.001
 enough-0.001
eter-0.001
Token looked
Token ID3114
Feature activation+3.02e-01

Pos logit contributions
 hurricane+0.014
date+0.013
 tornado+0.013
 television+0.013
 Amen+0.013
Neg logit contributions
 fault-0.012
 pret-0.011
paper-0.010
 loses-0.010
 puck-0.010
Token at
Token ID379
Feature activation-2.61e-01

Pos logit contributions
date+0.018
 hurricane+0.017
 Amen+0.017
 tornado+0.016
Friend+0.016
Neg logit contributions
 fault-0.019
 pret-0.018
 enough-0.018
 loses-0.017
paper-0.016
Token her
Token ID607
Feature activation-9.53e-04

Pos logit contributions
 fault+0.015
 pret+0.013
 enough+0.013
paper+0.013
 loses+0.012
Neg logit contributions
date-0.017
 hurricane-0.017
 Amen-0.016
 tornado-0.016
 year-0.016
Token mom
Token ID1995
Feature activation-1.24e-01

Pos logit contributions
 fault+0.000
 pret+0.000
 puck+0.000
 loses+0.000
paper+0.000
Neg logit contributions
date-0.000
 hurricane-0.000
 Amen-0.000
 tornado-0.000
Friend-0.000
Tokenmy
Token ID1820
Feature activation+2.09e-01

Pos logit contributions
 fault+0.006
 pret+0.006
 enough+0.005
paper+0.005
 pounds+0.005
Neg logit contributions
 hurricane-0.009
date-0.009
 tornado-0.009
 Amen-0.008
 messenger-0.008
Token and
Token ID290
Feature activation-1.55e-01

Pos logit contributions
 hurricane+0.015
date+0.015
 tornado+0.015
 Amen+0.014
Friend+0.014
Neg logit contributions
 fault-0.010
 pret-0.009
 enough-0.009
paper-0.008
 pounds-0.008
Token said
Token ID531
Feature activation+4.66e-01

Pos logit contributions
 fault+0.008
 pret+0.008
 enough+0.007
paper+0.007
 loses+0.007
Neg logit contributions
 hurricane-0.011
date-0.011
 tornado-0.010
 Amen-0.010
 year-0.010
Token told
Token ID1297
Feature activation-1.97e-01

Pos logit contributions
 fault+0.008
 pounds+0.007
paper+0.007
 pret+0.007
 loses+0.007
Neg logit contributions
Friend-0.004
date-0.004
 year-0.004
 hurricane-0.004
 voice-0.004
Token her
Token ID607
Feature activation-3.18e-01

Pos logit contributions
 pret+0.009
 fault+0.009
 enough+0.009
paper+0.008
 puts+0.008
Neg logit contributions
 hurricane-0.014
date-0.014
 tornado-0.013
 Amen-0.013
ind-0.013
Token it
Token ID340
Feature activation-1.05e-01

Pos logit contributions
 fault+0.021
 pret+0.020
paper+0.019
 loses+0.019
 puck+0.019
Neg logit contributions
 hurricane-0.017
date-0.016
 tornado-0.016
 television-0.016
 year-0.016
Token was
Token ID373
Feature activation-8.31e-03

Pos logit contributions
 fault+0.007
 pret+0.007
 puck+0.006
 pounds+0.006
ws+0.006
Neg logit contributions
 hurricane-0.005
 year-0.005
date-0.005
 tornado-0.005
 voice-0.005
Token five
Token ID1936
Feature activation-3.81e-01

Pos logit contributions
 fault+0.000
 pret+0.000
paper+0.000
 enough+0.000
 loses+0.000
Neg logit contributions
 hurricane-0.001
date-0.001
 tornado-0.001
 television-0.001
 year-0.001
Token dollars
Token ID5054
Feature activation-5.13e-02

Pos logit contributions
 fault+0.014
 pret+0.011
paper+0.011
ws+0.010
 loses+0.010
Neg logit contributions
 year-0.032
 hurricane-0.032
date-0.031
ind-0.030
 voice-0.030
Token.
Token ID13
Feature activation+6.37e-02

Pos logit contributions
 fault+0.003
 pret+0.003
paper+0.003
 loses+0.003
 puck+0.003
Neg logit contributions
 hurricane-0.003
date-0.003
 tornado-0.003
 year-0.003
 television-0.003
Token
Token ID198
Feature activation+1.59e-01

Pos logit contributions
 hurricane+0.004
date+0.004
 tornado+0.004
 year+0.004
ind+0.004
Neg logit contributions
 fault-0.004
 pret-0.004
paper-0.003
 puck-0.003
 loses-0.003
Token
Token ID198
Feature activation-6.50e-03

Pos logit contributions
 hurricane+0.011
 tornado+0.011
date+0.010
 television+0.010
 ranch+0.010
Neg logit contributions
 fault-0.008
 pret-0.007
paper-0.007
 loses-0.007
 puck-0.007
TokenL
Token ID43
Feature activation-2.46e-01

Pos logit contributions
 puck+0.000
 pret+0.000
 fault+0.000
paper+0.000
 loses+0.000
Neg logit contributions
 hurricane-0.000
 tornado-0.000
 television-0.000
date-0.000
 Amen-0.000
Tokenily
Token ID813
Feature activation-4.54e-01

Pos logit contributions
 fault+0.018
 pret+0.016
paper+0.016
 puck+0.015
eter+0.015
Neg logit contributions
date-0.013
 year-0.012
 hurricane-0.012
ilt-0.012
iet-0.012
TokenOne
Token ID3198
Feature activation-4.38e-01

Pos logit contributions
 fault+0.004
 pret+0.004
 puck+0.003
 loses+0.003
 enough+0.003
Neg logit contributions
 hurricane-0.003
date-0.003
 television-0.003
 tornado-0.003
 messenger-0.003
Token sunny
Token ID27737
Feature activation+2.32e-01

Pos logit contributions
 fault+0.039
 pret+0.036
paper+0.035
 loses+0.034
 puck+0.034
Neg logit contributions
 hurricane-0.015
date-0.015
 tornado-0.013
 television-0.013
 year-0.013
Token day
Token ID1110
Feature activation+1.34e-01

Pos logit contributions
cient+0.013
 television+0.013
 hurricane+0.012
 Amen+0.012
 sincere+0.012
Neg logit contributions
 fault-0.013
 pret-0.012
 enough-0.012
 counts-0.012
 cones-0.012
Token,
Token ID11
Feature activation-1.02e-01

Pos logit contributions
 hurricane+0.008
date+0.008
 tornado+0.007
 television+0.007
 year+0.007
Neg logit contributions
 fault-0.009
 pret-0.008
paper-0.008
 loses-0.008
 puck-0.007
Token Max
Token ID5436
Feature activation+2.48e-01

Pos logit contributions
 fault+0.005
 pret+0.005
paper+0.004
 loses+0.004
 puck+0.004
Neg logit contributions
 hurricane-0.007
date-0.007
 tornado-0.007
 year-0.007
 television-0.007
Token and
Token ID290
Feature activation-3.99e-01

Pos logit contributions
date+0.013
 hurricane+0.013
 year+0.012
 tornado+0.012
 musician+0.012
Neg logit contributions
 fault-0.017
 pret-0.015
 pounds-0.015
paper-0.015
 puck-0.015
Token Lily
Token ID20037
Feature activation-9.39e-02

Pos logit contributions
 fault+0.025
 pounds+0.021
paper+0.021
 pret+0.021
 loses+0.021
Neg logit contributions
 hurricane-0.025
date-0.024
 year-0.024
 voice-0.023
 tornado-0.023
Token went
Token ID1816
Feature activation-4.12e-01

Pos logit contributions
 fault+0.007
 pret+0.006
 pounds+0.006
paper+0.006
 puck+0.006
Neg logit contributions
 hurricane-0.005
date-0.005
 year-0.004
 tornado-0.004
 television-0.004
Token to
Token ID284
Feature activation-1.97e-01

Pos logit contributions
 fault+0.021
 pret+0.018
paper+0.017
 loses+0.017
 puck+0.017
Neg logit contributions
 hurricane-0.030
date-0.029
 tornado-0.028
 year-0.028
 television-0.028
Token the
Token ID262
Feature activation+2.35e-01

Pos logit contributions
 fault+0.014
 pounds+0.013
paper+0.013
 pret+0.013
eter+0.013
Neg logit contributions
 hurricane-0.010
 television-0.010
date-0.010
 tornado-0.009
 voice-0.009
Token park
Token ID3952
Feature activation+1.08e-01

Pos logit contributions
 Amen+0.010
date+0.009
ilt+0.009
 hurricane+0.009
han+0.009
Neg logit contributions
 fault-0.018
 pret-0.016
 enough-0.016
paper-0.016
 puck-0.016
Token to
Token ID284
Feature activation-4.62e-01

Pos logit contributions
date+0.004
 year+0.004
 others+0.004
 voice+0.004
 hurricane+0.004
Neg logit contributions
 pret-0.005
 fault-0.005
 pounds-0.005
 puck-0.005
ws-0.004
Token pull
Token ID2834
Feature activation-1.32e-01

Pos logit contributions
 fault+0.023
 enough+0.021
 loses+0.021
 pret+0.020
paper+0.019
Neg logit contributions
 hurricane-0.033
 tornado-0.032
cient-0.032
iet-0.032
date-0.032
Token his
Token ID465
Feature activation-2.00e-01

Pos logit contributions
 fault+0.007
 enough+0.006
 pret+0.006
paper+0.006
 loses+0.006
Neg logit contributions
date-0.009
 hurricane-0.009
 Amen-0.009
 tornado-0.009
cient-0.009
Token train
Token ID4512
Feature activation-2.01e-01

Pos logit contributions
 fault+0.013
 pret+0.012
paper+0.012
 loses+0.012
 puck+0.012
Neg logit contributions
 hurricane-0.011
date-0.011
 tornado-0.010
 television-0.010
 year-0.010
Token away
Token ID1497
Feature activation+8.96e-01

Pos logit contributions
 fault+0.013
 enough+0.012
paper+0.012
 pret+0.011
 puck+0.011
Neg logit contributions
 Amen-0.012
 hurricane-0.011
ind-0.011
Friend-0.011
iet-0.011
Token.
Token ID13
Feature activation-1.67e-01

Pos logit contributions
 hurricane+0.039
date+0.038
 tornado+0.036
 television+0.035
 year+0.035
Neg logit contributions
 fault-0.070
 pret-0.065
paper-0.063
 loses-0.062
 puck-0.061
Token He
Token ID679
Feature activation-3.79e-02

Pos logit contributions
 fault+0.010
 pret+0.009
paper+0.008
 puck+0.008
 enough+0.008
Neg logit contributions
 hurricane-0.011
date-0.010
ind-0.010
ilt-0.010
 tornado-0.010
Token thought
Token ID1807
Feature activation+7.26e-01

Pos logit contributions
 fault+0.003
 pret+0.002
paper+0.002
 pounds+0.002
 loses+0.002
Neg logit contributions
 hurricane-0.002
date-0.002
 year-0.002
 tornado-0.002
 voice-0.002
Token of
Token ID286
Feature activation+2.49e-01

Pos logit contributions
 hurricane+0.044
 tornado+0.041
 television+0.040
date+0.040
 year+0.039
Neg logit contributions
 fault-0.046
 pret-0.043
eter-0.040
paper-0.040
 loses-0.039
Token a
Token ID257
Feature activation-1.49e-01

Pos logit contributions
 hurricane+0.018
date+0.018
 tornado+0.017
 television+0.017
 year+0.017
Neg logit contributions
 fault-0.012
 pret-0.011
paper-0.010
 loses-0.010
 puck-0.010
Token plan
Token ID1410
Feature activation+1.46e-01

Pos logit contributions
 fault+0.011
 pret+0.010
paper+0.010
 loses+0.009
 pounds+0.009
Neg logit contributions
 hurricane-0.008
date-0.007
 tornado-0.007
 year-0.007
 television-0.007
Token The
Token ID383
Feature activation+1.25e-01

Pos logit contributions
 hurricane+0.012
date+0.012
 tornado+0.011
 television+0.011
 year+0.011
Neg logit contributions
 fault-0.012
 pret-0.011
paper-0.011
 loses-0.011
 pounds-0.010
Token bird
Token ID6512
Feature activation-8.70e-01

Pos logit contributions
date+0.008
han+0.007
ind+0.007
iet+0.007
Friend+0.007
Neg logit contributions
 fault-0.007
 pret-0.006
 enough-0.006
paper-0.006
 cones-0.006
Token liked
Token ID8288
Feature activation+3.73e-01

Pos logit contributions
 fault+0.059
 pret+0.055
paper+0.053
 loses+0.052
 enough+0.051
Neg logit contributions
 hurricane-0.047
date-0.046
 tornado-0.044
 television-0.042
 Amen-0.041
Token to
Token ID284
Feature activation-7.03e-01

Pos logit contributions
 hurricane+0.025
 tornado+0.024
 television+0.024
date+0.024
 year+0.023
Neg logit contributions
 fault-0.020
 pret-0.018
 pounds-0.017
 loses-0.017
 puck-0.017
Token fly
Token ID6129
Feature activation-7.99e-01

Pos logit contributions
 pret+0.041
 fault+0.037
 puck+0.035
 puts+0.035
ws+0.035
Neg logit contributions
ind-0.044
han-0.043
date-0.042
 hurricane-0.041
 courtroom-0.041
Token around
Token ID1088
Feature activation+9.03e-02

Pos logit contributions
 fault+0.050
 pret+0.048
paper+0.046
 loses+0.045
 enough+0.045
Neg logit contributions
 hurricane-0.047
date-0.046
 tornado-0.043
 Amen-0.043
 television-0.043
Token the
Token ID262
Feature activation-5.62e-01

Pos logit contributions
date+0.006
 hurricane+0.006
han+0.006
 messenger+0.006
 Amen+0.006
Neg logit contributions
 pret-0.005
 fault-0.004
 enough-0.004
paper-0.004
 loses-0.004
Token sky
Token ID6766
Feature activation+8.58e-02

Pos logit contributions
 fault+0.045
paper+0.040
 pret+0.040
 enough+0.039
 puck+0.039
Neg logit contributions
 Amen-0.025
date-0.025
han-0.023
 hurricane-0.023
Friend-0.021
Token,
Token ID11
Feature activation+2.31e-01

Pos logit contributions
 hurricane+0.005
date+0.005
 tornado+0.004
 television+0.004
 year+0.004
Neg logit contributions
 fault-0.006
 pret-0.005
paper-0.005
 loses-0.005
 enough-0.005
Token near
Token ID1474
Feature activation+2.82e-01

Pos logit contributions
date+0.014
 hurricane+0.014
 tornado+0.013
 Amen+0.013
ind+0.013
Neg logit contributions
 fault-0.014
 pret-0.013
paper-0.012
 loses-0.012
 puck-0.012
Token the
Token ID262
Feature activation+7.83e-02

Pos logit contributions
 hurricane+0.017
date+0.017
 tornado+0.016
 Amen+0.015
 television+0.015
Neg logit contributions
 fault-0.018
 pret-0.016
paper-0.015
 loses-0.015
 enough-0.015
TokenTw
Token ID5080
Feature activation-9.80e-01

Pos logit contributions
 fault+0.013
 pret+0.012
paper+0.012
 loses+0.012
 puck+0.011
Neg logit contributions
 hurricane-0.010
date-0.010
 tornado-0.009
 year-0.009
 television-0.009
Tokeninkle
Token ID19894
Feature activation-1.08e+00

Pos logit contributions
 fault+0.086
 pret+0.080
 pounds+0.076
ws+0.075
 loses+0.074
Neg logit contributions
 hurricane-0.038
date-0.036
 tornado-0.033
 television-0.032
 Amen-0.030
Token tw
Token ID665
Feature activation-1.31e+00

Pos logit contributions
 fault+0.068
 pret+0.062
paper+0.059
 loses+0.058
 puck+0.057
Neg logit contributions
 hurricane-0.064
date-0.063
 tornado-0.060
 television-0.059
 year-0.059
Tokeninkle
Token ID19894
Feature activation-1.25e+00

Pos logit contributions
 fault+0.113
 pret+0.107
paper+0.101
 loses+0.101
 pounds+0.100
Neg logit contributions
 hurricane-0.051
date-0.047
 tornado-0.045
 television-0.042
 Amen-0.041
Token little
Token ID1310
Feature activation-8.56e-01

Pos logit contributions
 fault+0.075
 pret+0.069
paper+0.065
 enough+0.065
 loses+0.064
Neg logit contributions
 hurricane-0.077
date-0.076
 tornado-0.072
 television-0.070
 Amen-0.070
Token star
Token ID3491
Feature activation-1.79e+00

Pos logit contributions
 fault+0.058
 pret+0.051
paper+0.050
 puck+0.049
 loses+0.048
Neg logit contributions
 hurricane-0.046
date-0.045
 voice-0.044
 tornado-0.044
 television-0.044
Token,
Token ID11
Feature activation-3.58e-01

Pos logit contributions
 fault+0.119
 pret+0.105
paper+0.103
 puck+0.102
 loses+0.098
Neg logit contributions
 hurricane-0.096
date-0.095
 year-0.093
 voice-0.092
 tornado-0.092
Token how
Token ID703
Feature activation-7.98e-02

Pos logit contributions
 fault+0.020
 pret+0.019
paper+0.017
 loses+0.017
 puck+0.017
Neg logit contributions
 hurricane-0.023
date-0.023
 tornado-0.022
 year-0.022
 television-0.021
Token I
Token ID314
Feature activation+6.75e-01

Pos logit contributions
 fault+0.005
 pret+0.004
paper+0.004
 puck+0.004
 pounds+0.004
Neg logit contributions
 hurricane-0.005
date-0.005
 tornado-0.005
 year-0.005
 television-0.004
Token wonder
Token ID4240
Feature activation+8.31e-01

Pos logit contributions
 hurricane+0.044
date+0.042
 tornado+0.042
 television+0.040
ind+0.040
Neg logit contributions
 fault-0.038
 pret-0.038
 loses-0.034
paper-0.033
ws-0.032
Token what
Token ID644
Feature activation+4.02e-01

Pos logit contributions
 hurricane+0.058
date+0.058
 tornado+0.056
 television+0.054
 year+0.054
Neg logit contributions
 fault-0.043
 pret-0.038
paper-0.036
 loses-0.035
 puck-0.035
Token.
Token ID13
Feature activation-1.26e-01

Pos logit contributions
 hurricane+0.015
date+0.015
 tornado+0.014
 year+0.014
 television+0.014
Neg logit contributions
 fault-0.015
 pret-0.014
paper-0.013
 loses-0.013
 puck-0.013
Token We
Token ID775
Feature activation-3.01e-01

Pos logit contributions
 fault+0.005
eter+0.005
 Maybe+0.005
 pret+0.005
 loses+0.004
Neg logit contributions
 hurricane-0.010
date-0.010
ilt-0.010
iet-0.009
 lens-0.009
Token can
Token ID460
Feature activation-6.25e-01

Pos logit contributions
 fault+0.013
 pret+0.011
 loses+0.010
 enough+0.010
ws+0.009
Neg logit contributions
 hurricane-0.024
date-0.024
ind-0.024
aby-0.024
 television-0.023
Tokenâ
Token ID22940
Feature activation-3.29e-02

Pos logit contributions
 fault+0.030
 loses+0.027
 pret+0.026
 puck+0.026
 puts+0.025
Neg logit contributions
 messenger-0.042
iet-0.042
 hurricane-0.042
 tornado-0.041
 doorway-0.041
Token
Token ID26391
Feature activation+2.01e-01

Pos logit contributions
 fault+0.002
 pret+0.002
ws+0.002
 loses+0.002
 enough+0.002
Neg logit contributions
 hurricane-0.002
 tornado-0.002
 television-0.002
ind-0.002
 lens-0.002
Token
Token ID8151
Feature activation-1.82e+00

Pos logit contributions
date+0.014
 hurricane+0.014
 tornado+0.013
 year+0.013
 television+0.013
Neg logit contributions
 fault-0.011
 pret-0.010
paper-0.009
 puck-0.009
 pounds-0.009
Tokent
Token ID83
Feature activation+5.60e-01

Pos logit contributions
 fault+0.098
paper+0.082
 loses+0.081
 enough+0.081
 pret+0.080
Neg logit contributions
 hurricane-0.126
date-0.120
 tornado-0.119
 Amen-0.118
 television-0.116
Token let
Token ID1309
Feature activation-8.58e-01

Pos logit contributions
cient+0.042
 lens+0.042
iet+0.041
 doorway+0.041
 tornado+0.041
Neg logit contributions
 fault-0.022
 enough-0.020
 loses-0.020
 puck-0.019
onna-0.019
Token you
Token ID345
Feature activation+7.91e-01

Pos logit contributions
 fault+0.062
 enough+0.057
 pret+0.056
 loses+0.054
paper+0.053
Neg logit contributions
date-0.044
 hurricane-0.042
 Amen-0.041
 lens-0.040
 year-0.039
Token play
Token ID711
Feature activation+3.13e-01

Pos logit contributions
 hurricane+0.052
 Amen+0.051
 tornado+0.051
ind+0.050
date+0.050
Neg logit contributions
 fault-0.043
 loses-0.039
 pret-0.038
 enough-0.037
 puck-0.037
Token until
Token ID1566
Feature activation-3.25e-01

Pos logit contributions
 hurricane+0.019
date+0.019
 tornado+0.018
 Amen+0.018
 year+0.018
Neg logit contributions
 fault-0.019
 pret-0.017
paper-0.016
 loses-0.016
 puck-0.016
Token something
Token ID1223
Feature activation-2.86e-01

Pos logit contributions
 hurricane+0.022
date+0.022
 tornado+0.021
ind+0.021
 year+0.021
Neg logit contributions
 fault-0.015
 pret-0.013
paper-0.012
 enough-0.012
 puck-0.012
Token strange
Token ID6283
Feature activation-7.69e-02

Pos logit contributions
 fault+0.019
 pret+0.017
paper+0.016
 loses+0.016
eter+0.016
Neg logit contributions
 hurricane-0.016
date-0.016
 tornado-0.015
 television-0.015
 year-0.015
Token
Token ID6184
Feature activation-2.02e-01

Pos logit contributions
 fault+0.006
 pret+0.005
paper+0.005
 enough+0.005
 loses+0.005
Neg logit contributions
 hurricane-0.004
date-0.004
 tornado-0.003
 Amen-0.003
ind-0.003
Token
Token ID95
Feature activation-3.40e-01

Pos logit contributions
 fault+0.012
 pret+0.012
paper+0.011
 puck+0.011
 loses+0.011
Neg logit contributions
 hurricane-0.012
date-0.012
 tornado-0.011
 year-0.011
 television-0.011
Token
Token ID26391
Feature activation-1.35e-01

Pos logit contributions
 fault+0.022
 pret+0.020
 puck+0.019
 loses+0.019
paper+0.019
Neg logit contributions
 hurricane-0.020
 tornado-0.019
date-0.018
 television-0.018
 year-0.017
Token
Token ID447
Feature activation-2.05e+00

Pos logit contributions
 fault+0.008
 pret+0.007
paper+0.007
 loses+0.007
 puck+0.007
Neg logit contributions
 hurricane-0.009
date-0.008
 tornado-0.008
 television-0.008
 year-0.008
Token
Token ID251
Feature activation-1.11e-01

Pos logit contributions
 fault+0.157
 pret+0.150
paper+0.146
 loses+0.143
 puck+0.141
Neg logit contributions
 hurricane-0.089
date-0.087
 tornado-0.082
 television-0.081
 year-0.078
Token a
Token ID257
Feature activation-6.91e-01

Pos logit contributions
 pret+0.004
 enough+0.004
 cones+0.004
 puck+0.004
 fault+0.004
Neg logit contributions
ind-0.009
 hurricane-0.008
date-0.008
iet-0.008
 tornado-0.008
Token towel
Token ID24808
Feature activation-2.98e-01

Pos logit contributions
 pret+0.041
 puck+0.041
 fault+0.038
 pounds+0.037
 enough+0.036
Neg logit contributions
han-0.039
date-0.038
ind-0.038
wered-0.037
ilt-0.036
Token!
Token ID0
Feature activation-1.03e-01

Pos logit contributions
 fault+0.015
 enough+0.015
paper+0.014
 pret+0.014
 puck+0.013
Neg logit contributions
ind-0.020
 Amen-0.020
 hurricane-0.019
iet-0.019
date-0.019
Token The
Token ID383
Feature activation+5.48e-03

Pos logit contributions
 fault+0.006
 pret+0.006
paper+0.005
 loses+0.005
 pounds+0.005
Neg logit contributions
 hurricane-0.006
date-0.006
 tornado-0.006
 television-0.006
 year-0.006
Token special
Token ID2041
Feature activation+8.45e-01

Pos logit contributions
 fault+0.006
 pret+0.006
 puck+0.006
 enough+0.005
 loses+0.005
Neg logit contributions
date-0.005
han-0.005
iet-0.005
wered-0.005
ind-0.005
Token prize
Token ID11596
Feature activation-7.05e-02

Pos logit contributions
iet+0.047
Friend+0.047
date+0.046
ind+0.046
cient+0.045
Neg logit contributions
 fault-0.056
 pret-0.051
 enough-0.048
 puck-0.047
paper-0.045
Token
Token ID6184
Feature activation-2.19e-01

Pos logit contributions
 fault+0.005
 enough+0.004
 pret+0.004
paper+0.004
 loses+0.004
Neg logit contributions
 hurricane-0.004
date-0.004
 Amen-0.004
ind-0.004
iet-0.004
Token
Token ID95
Feature activation-3.82e-01

Pos logit contributions
 fault+0.013
 pret+0.013
paper+0.012
 puck+0.012
 loses+0.012
Neg logit contributions
 hurricane-0.012
date-0.012
iet-0.012
 Amen-0.012
ind-0.011
Token
Token ID26391
Feature activation-2.55e-01

Pos logit contributions
 fault+0.024
 pret+0.023
 loses+0.021
paper+0.021
 weighs+0.021
Neg logit contributions
 hurricane-0.022
 tornado-0.020
 television-0.020
date-0.020
ind-0.020
Token
Token ID447
Feature activation-1.99e+00

Pos logit contributions
 fault+0.016
paper+0.014
 loses+0.014
 pret+0.014
eter+0.013
Neg logit contributions
 hurricane-0.016
date-0.015
 television-0.015
 tornado-0.014
 Amen-0.014
Token
Token ID250
Feature activation+3.98e-02

Pos logit contributions
 fault+0.132
 pret+0.131
paper+0.130
 loses+0.126
 puck+0.124
Neg logit contributions
 hurricane-0.098
date-0.094
 television-0.093
 tornado-0.091
ind-0.090
Token a
Token ID257
Feature activation-1.03e+00

Pos logit contributions
 hurricane+0.002
date+0.002
 tornado+0.002
 television+0.002
 year+0.002
Neg logit contributions
 fault-0.003
 pret-0.003
paper-0.002
 loses-0.002
 puck-0.002
Token big
Token ID1263
Feature activation+3.78e-02

Pos logit contributions
 pret+0.058
 fault+0.057
 puck+0.056
 enough+0.053
 pounds+0.053
Neg logit contributions
iet-0.062
date-0.061
han-0.061
wered-0.060
rim-0.060
Token,
Token ID11
Feature activation-5.59e-01

Pos logit contributions
date+0.002
Friend+0.002
iet+0.002
ind+0.002
 hurricane+0.002
Neg logit contributions
 fault-0.002
 pret-0.002
 enough-0.002
 puck-0.002
 loses-0.002
Token shiny
Token ID22441
Feature activation-5.29e-02

Pos logit contributions
 fault+0.034
 pret+0.033
 enough+0.032
 pounds+0.031
 puck+0.031
Neg logit contributions
date-0.034
 hurricane-0.031
iet-0.030
 Amen-0.030
Friend-0.030
Token a
Token ID257
Feature activation-2.83e-01

Pos logit contributions
 hurricane+0.022
date+0.021
 tornado+0.020
 television+0.020
 year+0.020
Neg logit contributions
 fault-0.018
 pret-0.015
paper-0.015
 loses-0.015
 pounds-0.014
Token shirt
Token ID10147
Feature activation+3.61e-01

Pos logit contributions
 fault+0.017
 pret+0.017
 puck+0.017
paper+0.015
 pounds+0.015
Neg logit contributions
han-0.016
date-0.016
 Amen-0.016
ind-0.015
 hurricane-0.015
Token and
Token ID290
Feature activation-4.76e-02

Pos logit contributions
 hurricane+0.022
date+0.022
 tornado+0.021
 television+0.020
 year+0.020
Neg logit contributions
 fault-0.022
 pret-0.020
paper-0.019
 loses-0.019
 puck-0.019
Token pants
Token ID12581
Feature activation+1.38e+00

Pos logit contributions
 fault+0.003
 pret+0.002
paper+0.002
 loses+0.002
eter+0.002
Neg logit contributions
 hurricane-0.003
date-0.003
 television-0.003
 tornado-0.003
 year-0.003
Token combo
Token ID14831
Feature activation-1.87e+00

Pos logit contributions
 hurricane+0.074
date+0.073
 tornado+0.069
 television+0.068
 year+0.068
Neg logit contributions
 fault-0.093
 pret-0.085
paper-0.080
 loses-0.080
 pounds-0.079
Token -
Token ID532
Feature activation-1.83e+00

Pos logit contributions
 fault+0.121
 pret+0.111
paper+0.107
 enough+0.106
 puck+0.105
Neg logit contributions
 hurricane-0.108
date-0.100
 tornado-0.100
 Amen-0.100
ind-0.097
Token and
Token ID290
Feature activation-1.60e+00

Pos logit contributions
 fault+0.108
 pret+0.100
paper+0.094
 puck+0.092
 loses+0.091
Neg logit contributions
 hurricane-0.115
date-0.114
 tornado-0.109
 Amen-0.107
 year-0.106
Token brushed
Token ID31782
Feature activation-7.13e-01

Pos logit contributions
 fault+0.092
 pret+0.087
paper+0.081
 puck+0.080
 loses+0.078
Neg logit contributions
 hurricane-0.102
date-0.100
 tornado-0.096
ind-0.094
 Amen-0.094
Token his
Token ID465
Feature activation-1.72e+00

Pos logit contributions
 fault+0.046
 pret+0.042
paper+0.040
 loses+0.039
 puck+0.039
Neg logit contributions
 hurricane-0.040
date-0.040
 tornado-0.038
 television-0.038
 year-0.037
Token hair
Token ID4190
Feature activation-1.47e-03

Pos logit contributions
 fault+0.096
 pret+0.086
paper+0.081
 loses+0.079
 puck+0.079
Neg logit contributions
 hurricane-0.115
date-0.113
 tornado-0.109
 television-0.106
 year-0.106
Token into
Token ID656
Feature activation-9.02e-01

Pos logit contributions
 fault+0.000
 pret+0.000
 loses+0.000
paper+0.000
 puck+0.000
Neg logit contributions
 hurricane-0.000
date-0.000
 year-0.000
 television-0.000
 tornado-0.000