Token | ith |
Token ID | 342 |
Feature activation | +6.29e-01 |
Pos logit contributions | |
quartered | +0.489 |
uphem | +0.475 |
hern | +0.464 |
uum | +0.457 |
leys | +0.457 |
Neg logit contributions | |
Armstrong | -0.501 |
ppo | -0.466 |
pos | -0.453 |
Cot | -0.444 |
闘 | -0.438 |
Token | l |
Token ID | 75 |
Feature activation | -9.24e-02 |
Pos logit contributions | |
Armstrong | +1.198 |
ppo | +1.107 |
pos | +1.078 |
Cot | +1.050 |
Recon | +1.013 |
Neg logit contributions | |
quartered | -1.343 |
uphem | -1.296 |
hern | -1.262 |
uum | -1.246 |
leys | -1.244 |
Token | ani |
Token ID | 3216 |
Feature activation | +2.49e-01 |
Pos logit contributions | |
quartered | +0.201 |
uphem | +0.194 |
hern | +0.190 |
leys | +0.188 |
uum | +0.187 |
Neg logit contributions | |
Armstrong | -0.173 |
ppo | -0.159 |
pos | -0.154 |
Cot | -0.151 |
闘 | -0.147 |
Token | , |
Token ID | 11 |
Feature activation | +2.06e-01 |
Pos logit contributions | |
Armstrong | +0.446 |
ppo | +0.409 |
pos | +0.396 |
Cot | +0.386 |
闘 | +0.376 |
Neg logit contributions | |
quartered | -0.560 |
uphem | -0.543 |
hern | -0.532 |
leys | -0.525 |
uum | -0.525 |
Token | who |
Token ID | 508 |
Feature activation | +6.16e-01 |
Pos logit contributions | |
Armstrong | +0.390 |
ppo | +0.356 |
pos | +0.349 |
Cot | +0.341 |
Recon | +0.329 |
Neg logit contributions | |
quartered | -0.444 |
uphem | -0.430 |
hern | -0.421 |
uum | -0.414 |
leys | -0.414 |
Token | weighed |
Token ID | 20261 |
Feature activation | +2.76e+00 |
Pos logit contributions | |
Armstrong | +1.130 |
ppo | +1.031 |
pos | +1.014 |
Cot | +0.982 |
Recon | +0.947 |
Neg logit contributions | |
quartered | -1.366 |
uphem | -1.327 |
hern | -1.296 |
uum | -1.275 |
leys | -1.274 |
Token | around |
Token ID | 1088 |
Feature activation | -4.51e-01 |
Pos logit contributions | |
Armstrong | +4.687 |
pos | +4.264 |
ppo | +4.024 |
Cot | +3.989 |
eff | +3.877 |
Neg logit contributions | |
quartered | -6.353 |
uphem | -6.241 |
hern | -5.967 |
leys | -5.950 |
uum | -5.922 |
Token | 128 |
Token ID | 13108 |
Feature activation | -3.44e-02 |
Pos logit contributions | |
quartered | +1.035 |
uphem | +1.005 |
hern | +0.988 |
leys | +0.974 |
uum | +0.973 |
Neg logit contributions | |
Armstrong | -0.777 |
ppo | -0.715 |
pos | -0.686 |
Cot | -0.671 |
闘 | -0.657 |
Token | kg |
Token ID | 14211 |
Feature activation | +8.50e-01 |
Pos logit contributions | |
quartered | +0.086 |
uphem | +0.083 |
hern | +0.082 |
leys | +0.081 |
uum | +0.081 |
Neg logit contributions | |
Armstrong | -0.052 |
ppo | -0.048 |
pos | -0.045 |
Cot | -0.044 |
Recon | -0.043 |
Token | , |
Token ID | 11 |
Feature activation | -6.54e-02 |
Pos logit contributions | |
Armstrong | +1.474 |
ppo | +1.343 |
pos | +1.308 |
Cot | +1.267 |
闘 | +1.222 |
Neg logit contributions | |
quartered | -1.959 |
uphem | -1.906 |
hern | -1.861 |
uum | -1.837 |
leys | -1.833 |
Token | slipped |
Token ID | 18859 |
Feature activation | -6.12e-02 |
Pos logit contributions | |
quartered | +0.139 |
uphem | +0.135 |
hern | +0.132 |
leys | +0.130 |
uum | +0.130 |
Neg logit contributions | |
Armstrong | -0.125 |
ppo | -0.115 |
pos | -0.112 |
Cot | -0.109 |
闘 | -0.106 |
Token | can |
Token ID | 460 |
Feature activation | -3.50e-01 |
Pos logit contributions | |
quartered | +0.434 |
uphem | +0.422 |
hern | +0.413 |
leys | +0.408 |
uum | +0.408 |
Neg logit contributions | |
Armstrong | -0.309 |
ppo | -0.282 |
pos | -0.272 |
Cot | -0.265 |
闘 | -0.257 |
Token | now |
Token ID | 783 |
Feature activation | -9.54e-02 |
Pos logit contributions | |
quartered | +0.761 |
uphem | +0.739 |
hern | +0.723 |
uum | +0.713 |
leys | +0.713 |
Neg logit contributions | |
Armstrong | -0.649 |
ppo | -0.598 |
pos | -0.578 |
Cot | -0.567 |
闘 | -0.553 |
Token | vote |
Token ID | 3015 |
Feature activation | +1.51e-01 |
Pos logit contributions | |
quartered | +0.200 |
uphem | +0.193 |
hern | +0.188 |
uum | +0.186 |
leys | +0.186 |
Neg logit contributions | |
Armstrong | -0.187 |
ppo | -0.173 |
pos | -0.168 |
Cot | -0.164 |
闘 | -0.159 |
Token | in |
Token ID | 287 |
Feature activation | -7.67e-02 |
Pos logit contributions | |
Armstrong | +0.274 |
ppo | +0.250 |
pos | +0.244 |
Cot | +0.237 |
Recon | +0.231 |
Neg logit contributions | |
quartered | -0.338 |
uphem | -0.328 |
hern | -0.321 |
leys | -0.316 |
uum | -0.316 |
Token | B |
Token ID | 347 |
Feature activation | -3.90e-01 |
Pos logit contributions | |
quartered | +0.167 |
uphem | +0.162 |
hern | +0.159 |
leys | +0.157 |
uum | +0.157 |
Neg logit contributions | |
Armstrong | -0.142 |
ppo | -0.131 |
pos | -0.127 |
Cot | -0.124 |
闘 | -0.121 |
Token | atch |
Token ID | 963 |
Feature activation | +2.37e+00 |
Pos logit contributions | |
quartered | +0.800 |
uphem | +0.780 |
hern | +0.764 |
leys | +0.754 |
uum | +0.752 |
Neg logit contributions | |
Armstrong | -0.769 |
ppo | -0.708 |
pos | -0.689 |
Cot | -0.674 |
闘 | -0.663 |
Token | 23 |
Token ID | 2242 |
Feature activation | +2.62e-01 |
Pos logit contributions | |
Armstrong | +4.328 |
ppo | +3.924 |
pos | +3.852 |
Cot | +3.777 |
Recon | +3.698 |
Neg logit contributions | |
quartered | -5.158 |
uphem | -5.001 |
hern | -4.859 |
uum | -4.759 |
leys | -4.680 |
Token | ! |
Token ID | 0 |
Feature activation | -1.30e-01 |
Pos logit contributions | |
Armstrong | +0.479 |
ppo | +0.438 |
pos | +0.427 |
Cot | +0.416 |
Recon | +0.404 |
Neg logit contributions | |
quartered | -0.579 |
uphem | -0.562 |
hern | -0.550 |
leys | -0.542 |
uum | -0.541 |
Token |
|
Token ID | 198 |
Feature activation | -8.40e-02 |
Pos logit contributions | |
quartered | +0.288 |
uphem | +0.280 |
hern | +0.274 |
leys | +0.270 |
uum | +0.270 |
Neg logit contributions | |
Armstrong | -0.237 |
ppo | -0.218 |
pos | -0.211 |
Cot | -0.206 |
闘 | -0.201 |
Token |
|
Token ID | 198 |
Feature activation | -1.12e-02 |
Pos logit contributions | |
quartered | +0.206 |
uphem | +0.197 |
hern | +0.192 |
leys | +0.190 |
uum | +0.189 |
Neg logit contributions | |
Armstrong | -0.139 |
ppo | -0.127 |
pos | -0.123 |
Cot | -0.118 |
Recon | -0.116 |
Token | Currently |
Token ID | 21327 |
Feature activation | -6.52e-01 |
Pos logit contributions | |
quartered | +0.024 |
uphem | +0.023 |
hern | +0.023 |
leys | +0.023 |
uum | +0.022 |
Neg logit contributions | |
Armstrong | -0.021 |
ppo | -0.020 |
pos | -0.019 |
Cot | -0.019 |
闘 | -0.018 |
Token | able |
Token ID | 540 |
Feature activation | -6.06e-01 |
Pos logit contributions | |
Armstrong | +1.619 |
ppo | +1.414 |
pos | +1.394 |
Cot | +1.331 |
Recon | +1.306 |
Neg logit contributions | |
quartered | -2.738 |
uphem | -2.616 |
leys | -2.592 |
hern | -2.580 |
uum | -2.553 |
Token | version |
Token ID | 2196 |
Feature activation | +8.87e-01 |
Pos logit contributions | |
quartered | +1.128 |
uphem | +1.106 |
hern | +1.086 |
leys | +1.077 |
uum | +1.070 |
Neg logit contributions | |
Armstrong | -1.270 |
ppo | -1.198 |
pos | -1.146 |
Cot | -1.139 |
闘 | -1.120 |
Token | <|endoftext|> |
Token ID | 50256 |
Feature activation | +1.24e-01 |
Pos logit contributions | |
Armstrong | +1.251 |
ppo | +1.072 |
pos | +1.044 |
Cot | +1.015 |
Recon | +1.002 |
Neg logit contributions | |
quartered | -2.371 |
uphem | -2.289 |
hern | -2.266 |
leys | -2.234 |
uum | -2.223 |
Token | Rob |
Token ID | 14350 |
Feature activation | +1.82e-01 |
Pos logit contributions | |
Armstrong | +0.245 |
ppo | +0.227 |
pos | +0.221 |
Cot | +0.216 |
闘 | +0.210 |
Neg logit contributions | |
quartered | -0.256 |
uphem | -0.248 |
hern | -0.242 |
leys | -0.239 |
uum | -0.239 |
Token | C |
Token ID | 327 |
Feature activation | -5.38e-02 |
Pos logit contributions | |
Armstrong | +0.351 |
ppo | +0.324 |
pos | +0.315 |
Cot | +0.307 |
闘 | +0.300 |
Neg logit contributions | |
quartered | -0.385 |
uphem | -0.373 |
hern | -0.364 |
leys | -0.359 |
uum | -0.359 |
Token | estern |
Token ID | 3330 |
Feature activation | +2.36e+00 |
Pos logit contributions | |
quartered | +0.101 |
uphem | +0.098 |
hern | +0.096 |
uum | +0.094 |
leys | +0.094 |
Neg logit contributions | |
Armstrong | -0.115 |
ppo | -0.107 |
pos | -0.105 |
Cot | -0.103 |
闘 | -0.101 |
Token | ino |
Token ID | 2879 |
Feature activation | +6.01e-01 |
Pos logit contributions | |
Armstrong | +4.232 |
pos | +3.931 |
Cot | +3.746 |
ppo | +3.730 |
atter | +3.613 |
Neg logit contributions | |
quartered | -5.234 |
uphem | -4.861 |
hern | -4.849 |
htaking | -4.720 |
surn | -4.716 |
Token | hosts |
Token ID | 11453 |
Feature activation | -4.95e-01 |
Pos logit contributions | |
Armstrong | +1.065 |
ppo | +0.976 |
pos | +0.955 |
Cot | +0.923 |
Recon | +0.896 |
Neg logit contributions | |
quartered | -1.369 |
uphem | -1.328 |
hern | -1.296 |
uum | -1.278 |
leys | -1.274 |
Token | a |
Token ID | 257 |
Feature activation | -2.48e-02 |
Pos logit contributions | |
quartered | +1.072 |
uphem | +1.042 |
hern | +1.029 |
leys | +1.015 |
uum | +1.007 |
Neg logit contributions | |
Armstrong | -0.904 |
ppo | -0.838 |
pos | -0.802 |
闘 | -0.784 |
Cot | -0.782 |
Token | jam |
Token ID | 16853 |
Feature activation | +2.29e-01 |
Pos logit contributions | |
quartered | +0.047 |
uphem | +0.046 |
hern | +0.045 |
leys | +0.044 |
uum | +0.044 |
Neg logit contributions | |
Armstrong | -0.053 |
ppo | -0.049 |
pos | -0.048 |
Cot | -0.047 |
闘 | -0.046 |
Token | packed |
Token ID | 11856 |
Feature activation | -2.23e-01 |
Pos logit contributions | |
Armstrong | +0.382 |
ppo | +0.347 |
pos | +0.334 |
Cot | +0.328 |
闘 | +0.321 |
Neg logit contributions | |
quartered | -0.539 |
uphem | -0.523 |
hern | -0.514 |
leys | -0.508 |
uum | -0.508 |
Token | based |
Token ID | 3106 |
Feature activation | +2.95e-02 |
Pos logit contributions | |
Armstrong | +0.259 |
ppo | +0.241 |
pos | +0.234 |
Cot | +0.229 |
闘 | +0.224 |
Neg logit contributions | |
quartered | -0.245 |
uphem | -0.237 |
hern | -0.232 |
leys | -0.228 |
uum | -0.228 |
Token | reservations |
Token ID | 24722 |
Feature activation | +2.68e-01 |
Pos logit contributions | |
Armstrong | +0.056 |
ppo | +0.052 |
pos | +0.051 |
Cot | +0.049 |
闘 | +0.048 |
Neg logit contributions | |
quartered | -0.063 |
uphem | -0.061 |
hern | -0.060 |
leys | -0.059 |
uum | -0.059 |
Token | . |
Token ID | 13 |
Feature activation | -2.61e-02 |
Pos logit contributions | |
Armstrong | +0.497 |
ppo | +0.458 |
pos | +0.445 |
Cot | +0.434 |
闘 | +0.422 |
Neg logit contributions | |
quartered | -0.585 |
uphem | -0.567 |
hern | -0.555 |
leys | -0.548 |
uum | -0.548 |
Token | So |
Token ID | 1406 |
Feature activation | -3.18e-01 |
Pos logit contributions | |
quartered | +0.057 |
uphem | +0.055 |
hern | +0.054 |
leys | +0.053 |
uum | +0.053 |
Neg logit contributions | |
Armstrong | -0.048 |
ppo | -0.045 |
pos | -0.043 |
Cot | -0.042 |
闘 | -0.041 |
Token | B |
Token ID | 347 |
Feature activation | -3.17e-01 |
Pos logit contributions | |
quartered | +0.699 |
uphem | +0.677 |
hern | +0.663 |
leys | +0.655 |
uum | +0.654 |
Neg logit contributions | |
Armstrong | -0.582 |
ppo | -0.536 |
pos | -0.519 |
Cot | -0.507 |
闘 | -0.496 |
Token | SP |
Token ID | 4303 |
Feature activation | +2.34e+00 |
Pos logit contributions | |
quartered | +0.631 |
uphem | +0.612 |
hern | +0.598 |
leys | +0.590 |
uum | +0.589 |
Neg logit contributions | |
Armstrong | -0.646 |
ppo | -0.599 |
pos | -0.583 |
Cot | -0.571 |
闘 | -0.559 |
Token | - |
Token ID | 12 |
Feature activation | +2.79e-01 |
Pos logit contributions | |
Armstrong | +4.454 |
pos | +4.029 |
ppo | +3.990 |
Cot | +3.901 |
Recon | +3.791 |
Neg logit contributions | |
quartered | -4.893 |
uphem | -4.813 |
hern | -4.663 |
uum | -4.581 |
leys | -4.560 |
Token | r |
Token ID | 81 |
Feature activation | +8.05e-01 |
Pos logit contributions | |
Armstrong | +0.564 |
ppo | +0.523 |
pos | +0.510 |
Cot | +0.497 |
闘 | +0.485 |
Neg logit contributions | |
quartered | -0.561 |
uphem | -0.544 |
hern | -0.530 |
uum | -0.524 |
leys | -0.524 |
Token | uled |
Token ID | 6309 |
Feature activation | -5.20e-02 |
Pos logit contributions | |
Armstrong | +1.732 |
ppo | +1.608 |
pos | +1.587 |
Cot | +1.533 |
Recon | +1.495 |
Neg logit contributions | |
quartered | -1.481 |
uphem | -1.443 |
hern | -1.417 |
uum | -1.407 |
leys | -1.382 |
Token | Uttar |
Token ID | 44744 |
Feature activation | +9.75e-01 |
Pos logit contributions | |
quartered | +0.107 |
uphem | +0.103 |
hern | +0.101 |
uum | +0.099 |
leys | +0.099 |
Neg logit contributions | |
Armstrong | -0.104 |
ppo | -0.096 |
pos | -0.094 |
Cot | -0.091 |
闘 | -0.089 |
Token | Pradesh |
Token ID | 25945 |
Feature activation | +1.04e+00 |
Pos logit contributions | |
Armstrong | +1.587 |
ppo | +1.439 |
pos | +1.400 |
Cot | +1.357 |
Recon | +1.309 |
Neg logit contributions | |
quartered | -2.351 |
uphem | -2.280 |
hern | -2.235 |
leys | -2.208 |
uum | -2.206 |
Token | Friday |
Token ID | 3217 |
Feature activation | +3.10e-01 |
Pos logit contributions | |
quartered | +0.096 |
uphem | +0.094 |
hern | +0.092 |
leys | +0.091 |
uum | +0.090 |
Neg logit contributions | |
Armstrong | -0.080 |
ppo | -0.074 |
pos | -0.071 |
Cot | -0.069 |
闘 | -0.068 |
Token | � |
Token ID | 447 |
Feature activation | -2.07e-01 |
Pos logit contributions | |
Armstrong | +0.545 |
ppo | +0.499 |
pos | +0.483 |
Cot | +0.471 |
闘 | +0.458 |
Neg logit contributions | |
quartered | -0.707 |
uphem | -0.686 |
hern | -0.672 |
leys | -0.664 |
uum | -0.664 |
Token | � |
Token ID | 247 |
Feature activation | -6.46e-02 |
Pos logit contributions | |
quartered | +0.369 |
uphem | +0.355 |
hern | +0.346 |
uum | +0.340 |
leys | +0.340 |
Neg logit contributions | |
Armstrong | -0.466 |
ppo | -0.435 |
pos | -0.425 |
Cot | -0.417 |
闘 | -0.408 |
Token | s |
Token ID | 82 |
Feature activation | -1.30e-01 |
Pos logit contributions | |
quartered | +0.136 |
uphem | +0.132 |
hern | +0.129 |
leys | +0.127 |
uum | +0.127 |
Neg logit contributions | |
Armstrong | -0.125 |
ppo | -0.115 |
pos | -0.112 |
Cot | -0.109 |
闘 | -0.107 |
Token | ceremonial |
Token ID | 39490 |
Feature activation | +8.47e-01 |
Pos logit contributions | |
quartered | +0.257 |
uphem | +0.249 |
hern | +0.243 |
leys | +0.239 |
uum | +0.239 |
Neg logit contributions | |
Armstrong | -0.267 |
ppo | -0.249 |
pos | -0.241 |
Cot | -0.236 |
闘 | -0.231 |
Token | weigh |
Token ID | 10164 |
Feature activation | +2.32e+00 |
Pos logit contributions | |
Armstrong | +1.653 |
ppo | +1.514 |
pos | +1.490 |
Cot | +1.454 |
Recon | +1.407 |
Neg logit contributions | |
quartered | -1.779 |
uphem | -1.724 |
hern | -1.677 |
uum | -1.660 |
leys | -1.659 |
Token | - |
Token ID | 12 |
Feature activation | -9.43e-02 |
Pos logit contributions | |
Armstrong | +3.846 |
pos | +3.421 |
ppo | +3.417 |
Cot | +3.252 |
闘 | +3.168 |
Neg logit contributions | |
quartered | -5.541 |
uphem | -5.372 |
hern | -5.224 |
uum | -5.137 |
leys | -5.125 |
Token | in |
Token ID | 259 |
Feature activation | +7.99e-02 |
Pos logit contributions | |
quartered | +0.200 |
uphem | +0.194 |
hern | +0.192 |
leys | +0.189 |
uum | +0.189 |
Neg logit contributions | |
Armstrong | -0.179 |
ppo | -0.166 |
pos | -0.160 |
Cot | -0.159 |
闘 | -0.154 |
Token | . |
Token ID | 13 |
Feature activation | -5.67e-02 |
Pos logit contributions | |
Armstrong | +0.144 |
ppo | +0.132 |
pos | +0.128 |
Cot | +0.125 |
闘 | +0.122 |
Neg logit contributions | |
quartered | -0.179 |
uphem | -0.173 |
hern | -0.170 |
leys | -0.167 |
uum | -0.167 |
Token |
|
Token ID | 198 |
Feature activation | -2.76e-01 |
Pos logit contributions | |
quartered | +0.126 |
uphem | +0.123 |
hern | +0.120 |
uum | +0.118 |
leys | +0.118 |
Neg logit contributions | |
Armstrong | -0.103 |
ppo | -0.095 |
pos | -0.092 |
Cot | -0.090 |
闘 | -0.087 |
Token |
|
Token ID | 198 |
Feature activation | -1.82e-01 |
Pos logit contributions | |
quartered | +0.676 |
uphem | +0.644 |
hern | +0.627 |
leys | +0.620 |
uum | +0.619 |
Neg logit contributions | |
Armstrong | -0.462 |
ppo | -0.420 |
pos | -0.410 |
Cot | -0.393 |
Recon | -0.383 |
Token | community |
Token ID | 2055 |
Feature activation | +3.18e-01 |
Pos logit contributions | |
quartered | +0.351 |
uphem | +0.340 |
hern | +0.333 |
leys | +0.328 |
uum | +0.328 |
Neg logit contributions | |
Armstrong | -0.333 |
ppo | -0.309 |
pos | -0.299 |
Cot | -0.293 |
闘 | -0.286 |
Token | boards |
Token ID | 11490 |
Feature activation | +7.23e-01 |
Pos logit contributions | |
Armstrong | +0.641 |
ppo | +0.593 |
pos | +0.580 |
Cot | +0.564 |
闘 | +0.551 |
Neg logit contributions | |
quartered | -0.645 |
uphem | -0.624 |
hern | -0.609 |
uum | -0.601 |
leys | -0.600 |
Token | across |
Token ID | 1973 |
Feature activation | +5.04e-01 |
Pos logit contributions | |
Armstrong | +1.368 |
ppo | +1.261 |
pos | +1.235 |
Cot | +1.194 |
Recon | +1.161 |
Neg logit contributions | |
quartered | -1.560 |
uphem | -1.505 |
hern | -1.472 |
leys | -1.452 |
uum | -1.451 |
Token | the |
Token ID | 262 |
Feature activation | -9.81e-02 |
Pos logit contributions | |
Armstrong | +0.964 |
ppo | +0.888 |
pos | +0.865 |
Cot | +0.844 |
闘 | +0.821 |
Neg logit contributions | |
quartered | -1.075 |
uphem | -1.040 |
hern | -1.016 |
leys | -1.002 |
uum | -1.002 |
Token | city |
Token ID | 1748 |
Feature activation | -7.97e-01 |
Pos logit contributions | |
quartered | +0.151 |
uphem | +0.145 |
hern | +0.140 |
uum | +0.137 |
leys | +0.137 |
Neg logit contributions | |
Armstrong | -0.245 |
ppo | -0.231 |
pos | -0.227 |
Cot | -0.222 |
Recon | -0.217 |
Token | weigh |
Token ID | 10164 |
Feature activation | +2.25e+00 |
Pos logit contributions | |
quartered | +1.642 |
uphem | +1.598 |
hern | +1.559 |
leys | +1.552 |
uum | +1.548 |
Neg logit contributions | |
Armstrong | -1.545 |
ppo | -1.433 |
pos | -1.370 |
Cot | -1.360 |
闘 | -1.329 |
Token | in |
Token ID | 287 |
Feature activation | -1.87e-01 |
Pos logit contributions | |
Armstrong | +4.014 |
pos | +3.620 |
ppo | +3.618 |
Cot | +3.447 |
Recon | +3.372 |
Neg logit contributions | |
quartered | -5.095 |
uphem | -4.911 |
hern | -4.762 |
uum | -4.722 |
leys | -4.691 |
Token | on |
Token ID | 319 |
Feature activation | -1.46e-01 |
Pos logit contributions | |
quartered | +0.415 |
uphem | +0.403 |
hern | +0.395 |
leys | +0.390 |
uum | +0.389 |
Neg logit contributions | |
Armstrong | -0.340 |
ppo | -0.313 |
pos | -0.302 |
Cot | -0.295 |
闘 | -0.288 |
Token | proposed |
Token ID | 5150 |
Feature activation | +1.09e-01 |
Pos logit contributions | |
quartered | +0.315 |
uphem | +0.306 |
hern | +0.300 |
leys | +0.296 |
uum | +0.295 |
Neg logit contributions | |
Armstrong | -0.271 |
ppo | -0.251 |
pos | -0.242 |
Cot | -0.237 |
闘 | -0.231 |
Token | housing |
Token ID | 5627 |
Feature activation | -3.32e-01 |
Pos logit contributions | |
Armstrong | +0.223 |
ppo | +0.206 |
pos | +0.201 |
Cot | +0.197 |
闘 | +0.192 |
Neg logit contributions | |
quartered | -0.217 |
uphem | -0.210 |
hern | -0.204 |
uum | -0.201 |
leys | -0.201 |
Token | policies |
Token ID | 4788 |
Feature activation | +1.66e-02 |
Pos logit contributions | |
quartered | +0.660 |
uphem | +0.639 |
hern | +0.624 |
leys | +0.617 |
uum | +0.614 |
Neg logit contributions | |
Armstrong | -0.678 |
ppo | -0.631 |
pos | -0.609 |
Cot | -0.599 |
闘 | -0.586 |
Token | behind |
Token ID | 2157 |
Feature activation | +9.49e-01 |
Pos logit contributions | |
Armstrong | +0.592 |
ppo | +0.552 |
pos | +0.538 |
Cot | +0.528 |
闘 | +0.517 |
Neg logit contributions | |
quartered | -0.493 |
uphem | -0.475 |
hern | -0.463 |
leys | -0.456 |
uum | -0.455 |
Token | " |
Token ID | 1 |
Feature activation | -1.22e-01 |
Pos logit contributions | |
Armstrong | +1.753 |
ppo | +1.606 |
pos | +1.570 |
Cot | +1.524 |
Recon | +1.487 |
Neg logit contributions | |
quartered | -2.082 |
uphem | -2.015 |
hern | -1.969 |
uum | -1.951 |
leys | -1.948 |
Token | again |
Token ID | 757 |
Feature activation | +4.37e-01 |
Pos logit contributions | |
quartered | +0.266 |
uphem | +0.257 |
hern | +0.252 |
uum | +0.249 |
leys | +0.249 |
Neg logit contributions | |
Armstrong | -0.227 |
ppo | -0.208 |
pos | -0.203 |
Cot | -0.197 |
Recon | -0.192 |
Token | , |
Token ID | 11 |
Feature activation | +2.33e-03 |
Pos logit contributions | |
Armstrong | +0.773 |
ppo | +0.707 |
pos | +0.687 |
Cot | +0.668 |
Recon | +0.648 |
Neg logit contributions | |
quartered | -0.996 |
uphem | -0.966 |
hern | -0.941 |
leys | -0.934 |
uum | -0.932 |
Token | but |
Token ID | 475 |
Feature activation | +2.53e-01 |
Pos logit contributions | |
Armstrong | +0.004 |
ppo | +0.004 |
pos | +0.004 |
Cot | +0.004 |
Recon | +0.004 |
Neg logit contributions | |
quartered | -0.005 |
uphem | -0.005 |
hern | -0.005 |
uum | -0.005 |
leys | -0.005 |
Token | rather |
Token ID | 2138 |
Feature activation | +2.22e+00 |
Pos logit contributions | |
Armstrong | +0.460 |
ppo | +0.417 |
pos | +0.410 |
Cot | +0.398 |
Recon | +0.385 |
Neg logit contributions | |
quartered | -0.571 |
uphem | -0.553 |
hern | -0.538 |
uum | -0.535 |
leys | -0.532 |
Token | another |
Token ID | 1194 |
Feature activation | +4.53e-01 |
Pos logit contributions | |
Armstrong | +3.850 |
pos | +3.460 |
ppo | +3.428 |
Cot | +3.282 |
eff | +3.213 |
Neg logit contributions | |
quartered | -5.090 |
uphem | -4.914 |
uum | -4.826 |
hern | -4.766 |
leys | -4.728 |
Token | example |
Token ID | 1672 |
Feature activation | +9.36e-02 |
Pos logit contributions | |
Armstrong | +0.897 |
ppo | +0.820 |
pos | +0.810 |
Cot | +0.788 |
Recon | +0.766 |
Neg logit contributions | |
quartered | -0.943 |
uphem | -0.907 |
hern | -0.884 |
uum | -0.879 |
leys | -0.878 |
Token | of |
Token ID | 286 |
Feature activation | -3.38e-02 |
Pos logit contributions | |
Armstrong | +0.186 |
ppo | +0.174 |
pos | +0.166 |
Cot | +0.164 |
闘 | +0.162 |
Neg logit contributions | |
quartered | -0.190 |
uphem | -0.184 |
hern | -0.180 |
uum | -0.177 |
leys | -0.176 |
Token | the |
Token ID | 262 |
Feature activation | +6.70e-02 |
Pos logit contributions | |
quartered | +0.072 |
uphem | +0.069 |
hern | +0.068 |
leys | +0.067 |
uum | +0.067 |
Neg logit contributions | |
Armstrong | -0.065 |
ppo | -0.060 |
pos | -0.058 |
Cot | -0.057 |
闘 | -0.056 |
Token | Obama |
Token ID | 2486 |
Feature activation | +1.11e-02 |
Pos logit contributions | |
Armstrong | +0.135 |
ppo | +0.124 |
pos | +0.122 |
Cot | +0.119 |
Recon | +0.115 |
Neg logit contributions | |
quartered | -0.137 |
uphem | -0.132 |
hern | -0.129 |
uum | -0.128 |
leys | -0.127 |
Token | 24 |
Token ID | 1731 |
Feature activation | -6.39e-02 |
Pos logit contributions | |
quartered | +0.021 |
uphem | +0.020 |
hern | +0.020 |
leys | +0.019 |
uum | +0.019 |
Neg logit contributions | |
Armstrong | -0.018 |
ppo | -0.016 |
pos | -0.016 |
Cot | -0.015 |
闘 | -0.015 |
Token | ) |
Token ID | 8 |
Feature activation | +5.74e-01 |
Pos logit contributions | |
quartered | +0.140 |
uphem | +0.136 |
hern | +0.134 |
leys | +0.132 |
uum | +0.132 |
Neg logit contributions | |
Armstrong | -0.116 |
ppo | -0.108 |
pos | -0.103 |
Cot | -0.101 |
闘 | -0.099 |
Token |
|
Token ID | 198 |
Feature activation | +2.28e-01 |
Pos logit contributions | |
Armstrong | +1.048 |
ppo | +0.969 |
pos | +0.934 |
Cot | +0.910 |
闘 | +0.887 |
Neg logit contributions | |
quartered | -1.258 |
uphem | -1.227 |
hern | -1.201 |
uum | -1.186 |
leys | -1.185 |
Token |
|
Token ID | 198 |
Feature activation | +2.28e-01 |
Pos logit contributions | |
Armstrong | +0.373 |
ppo | +0.337 |
pos | +0.330 |
Cot | +0.316 |
Recon | +0.310 |
Neg logit contributions | |
quartered | -0.574 |
uphem | -0.544 |
hern | -0.526 |
uum | -0.523 |
leys | -0.523 |
Token | C |
Token ID | 34 |
Feature activation | -5.48e-03 |
Pos logit contributions | |
Armstrong | +0.423 |
ppo | +0.390 |
pos | +0.378 |
Cot | +0.369 |
闘 | +0.360 |
Neg logit contributions | |
quartered | -0.498 |
uphem | -0.482 |
hern | -0.472 |
leys | -0.466 |
uum | -0.465 |
Token | atch |
Token ID | 963 |
Feature activation | +2.20e+00 |
Pos logit contributions | |
quartered | +0.012 |
uphem | +0.011 |
hern | +0.011 |
leys | +0.011 |
uum | +0.011 |
Neg logit contributions | |
Armstrong | -0.011 |
ppo | -0.010 |
pos | -0.009 |
Cot | -0.009 |
Recon | -0.009 |
Token | a |
Token ID | 257 |
Feature activation | +1.68e-01 |
Pos logit contributions | |
Armstrong | +4.031 |
ppo | +3.690 |
pos | +3.634 |
Cot | +3.516 |
Recon | +3.480 |
Neg logit contributions | |
quartered | -4.796 |
uphem | -4.647 |
hern | -4.519 |
leys | -4.466 |
uum | -4.452 |
Token | blue |
Token ID | 4171 |
Feature activation | +4.34e-01 |
Pos logit contributions | |
Armstrong | +0.327 |
ppo | +0.302 |
pos | +0.293 |
Cot | +0.286 |
闘 | +0.279 |
Neg logit contributions | |
quartered | -0.353 |
uphem | -0.342 |
hern | -0.334 |
uum | -0.329 |
leys | -0.329 |
Token | - |
Token ID | 12 |
Feature activation | +2.51e-01 |
Pos logit contributions | |
Armstrong | +0.842 |
ppo | +0.778 |
pos | +0.756 |
Cot | +0.738 |
闘 | +0.720 |
Neg logit contributions | |
quartered | -0.911 |
uphem | -0.883 |
hern | -0.863 |
leys | -0.851 |
uum | -0.851 |
Token | painted |
Token ID | 47351 |
Feature activation | -3.32e-02 |
Pos logit contributions | |
Armstrong | +0.517 |
ppo | +0.479 |
pos | +0.467 |
Cot | +0.457 |
闘 | +0.448 |
Neg logit contributions | |
quartered | -0.497 |
uphem | -0.480 |
hern | -0.469 |
uum | -0.462 |
leys | -0.461 |
Token | box |
Token ID | 3091 |
Feature activation | +4.62e-01 |
Pos logit contributions | |
quartered | +0.071 |
uphem | +0.068 |
hern | +0.067 |
uum | +0.066 |
leys | +0.066 |
Neg logit contributions | |
Armstrong | -0.063 |
ppo | -0.058 |
pos | -0.056 |
Cot | -0.055 |
闘 | -0.054 |
Token | historic |
Token ID | 9566 |
Feature activation | +1.03e-01 |
Pos logit contributions | |
Armstrong | +0.466 |
ppo | +0.431 |
pos | +0.426 |
Cot | +0.416 |
Recon | +0.404 |
Neg logit contributions | |
quartered | -0.399 |
uphem | -0.384 |
hern | -0.373 |
leys | -0.369 |
uum | -0.367 |
Token | Broadway |
Token ID | 20617 |
Feature activation | +5.76e-02 |
Pos logit contributions | |
Armstrong | +0.216 |
ppo | +0.200 |
pos | +0.196 |
Cot | +0.192 |
Recon | +0.186 |
Neg logit contributions | |
quartered | -0.199 |
uphem | -0.192 |
hern | -0.187 |
leys | -0.184 |
uum | -0.184 |
Token | musical |
Token ID | 10530 |
Feature activation | +1.42e-01 |
Pos logit contributions | |
Armstrong | +0.126 |
ppo | +0.117 |
pos | +0.115 |
Cot | +0.112 |
Recon | +0.109 |
Neg logit contributions | |
quartered | -0.108 |
uphem | -0.104 |
hern | -0.101 |
leys | -0.099 |
uum | -0.099 |
Token | has |
Token ID | 468 |
Feature activation | +3.28e-01 |
Pos logit contributions | |
Armstrong | +0.260 |
ppo | +0.237 |
pos | +0.233 |
Cot | +0.225 |
Recon | +0.218 |
Neg logit contributions | |
quartered | -0.314 |
uphem | -0.306 |
hern | -0.300 |
leys | -0.296 |
uum | -0.294 |
Token | gone |
Token ID | 3750 |
Feature activation | +2.41e-01 |
Pos logit contributions | |
Armstrong | +0.722 |
ppo | +0.664 |
pos | +0.663 |
Cot | +0.642 |
Recon | +0.623 |
Neg logit contributions | |
quartered | -0.610 |
uphem | -0.587 |
hern | -0.569 |
leys | -0.564 |
uum | -0.564 |
Token | missing |
Token ID | 4814 |
Feature activation | +2.18e+00 |
Pos logit contributions | |
Armstrong | +0.449 |
ppo | +0.412 |
pos | +0.402 |
Cot | +0.391 |
Recon | +0.379 |
Neg logit contributions | |
quartered | -0.527 |
uphem | -0.510 |
hern | -0.497 |
leys | -0.492 |
uum | -0.492 |
Token | . |
Token ID | 13 |
Feature activation | -4.52e-02 |
Pos logit contributions | |
Armstrong | +3.503 |
ppo | +3.118 |
pos | +3.091 |
Cot | +3.002 |
eff | +2.893 |
Neg logit contributions | |
quartered | -5.250 |
uphem | -5.208 |
leys | -5.017 |
uum | -4.921 |
hern | -4.903 |
Token |
|
Token ID | 198 |
Feature activation | +1.63e-01 |
Pos logit contributions | |
quartered | +0.101 |
uphem | +0.098 |
hern | +0.096 |
uum | +0.095 |
leys | +0.095 |
Neg logit contributions | |
Armstrong | -0.082 |
ppo | -0.075 |
pos | -0.073 |
Cot | -0.071 |
闘 | -0.069 |
Token |
|
Token ID | 198 |
Feature activation | +2.37e-01 |
Pos logit contributions | |
Armstrong | +0.267 |
ppo | +0.240 |
pos | +0.235 |
Cot | +0.226 |
Recon | +0.220 |
Neg logit contributions | |
quartered | -0.408 |
uphem | -0.389 |
hern | -0.377 |
uum | -0.373 |
leys | -0.373 |
Token | Jeff |
Token ID | 19139 |
Feature activation | +2.03e-02 |
Pos logit contributions | |
Armstrong | +0.461 |
ppo | +0.425 |
pos | +0.413 |
Cot | +0.404 |
闘 | +0.393 |
Neg logit contributions | |
quartered | -0.500 |
uphem | -0.482 |
hern | -0.470 |
uum | -0.464 |
leys | -0.464 |
Token | rey |
Token ID | 4364 |
Feature activation | +1.70e-01 |
Pos logit contributions | |
Armstrong | +0.049 |
ppo | +0.046 |
pos | +0.045 |
Cot | +0.044 |
闘 | +0.043 |
Neg logit contributions | |
quartered | -0.033 |
uphem | -0.032 |
hern | -0.031 |
leys | -0.030 |
uum | -0.030 |
Token | mirror |
Token ID | 10162 |
Feature activation | -4.99e-01 |
Pos logit contributions | |
quartered | +0.390 |
uphem | +0.377 |
hern | +0.369 |
leys | +0.363 |
uum | +0.363 |
Neg logit contributions | |
Armstrong | -0.380 |
ppo | -0.353 |
pos | -0.341 |
Cot | -0.335 |
闘 | -0.328 |
Token | ing |
Token ID | 278 |
Feature activation | +5.86e-02 |
Pos logit contributions | |
quartered | +0.695 |
uphem | +0.662 |
hern | +0.640 |
leys | +0.626 |
uum | +0.625 |
Neg logit contributions | |
Armstrong | -1.319 |
ppo | -1.246 |
pos | -1.218 |
Cot | -1.199 |
闘 | -1.180 |
Token | themes |
Token ID | 13460 |
Feature activation | +8.50e-01 |
Pos logit contributions | |
Armstrong | +0.110 |
ppo | +0.101 |
pos | +0.098 |
Cot | +0.096 |
闘 | +0.093 |
Neg logit contributions | |
quartered | -0.127 |
uphem | -0.123 |
hern | -0.120 |
leys | -0.119 |
uum | -0.119 |
Token | of |
Token ID | 286 |
Feature activation | +2.07e-01 |
Pos logit contributions | |
Armstrong | +1.595 |
ppo | +1.447 |
pos | +1.436 |
Cot | +1.387 |
Recon | +1.338 |
Neg logit contributions | |
quartered | -1.857 |
uphem | -1.798 |
hern | -1.758 |
leys | -1.740 |
uum | -1.730 |
Token | more |
Token ID | 517 |
Feature activation | +2.09e-01 |
Pos logit contributions | |
Armstrong | +0.416 |
ppo | +0.385 |
pos | +0.375 |
Cot | +0.366 |
闘 | +0.357 |
Neg logit contributions | |
quartered | -0.419 |
uphem | -0.405 |
hern | -0.396 |
leys | -0.390 |
uum | -0.390 |
Token | weight |
Token ID | 3463 |
Feature activation | +2.18e+00 |
Pos logit contributions | |
Armstrong | +0.403 |
ppo | +0.373 |
pos | +0.362 |
Cot | +0.354 |
闘 | +0.345 |
Neg logit contributions | |
quartered | -0.442 |
uphem | -0.428 |
hern | -0.418 |
leys | -0.412 |
uum | -0.412 |
Token | y |
Token ID | 88 |
Feature activation | -3.61e-01 |
Pos logit contributions | |
Armstrong | +3.734 |
ppo | +3.356 |
pos | +3.326 |
Cot | +3.135 |
eff | +3.035 |
Neg logit contributions | |
quartered | -5.055 |
uphem | -4.955 |
hern | -4.809 |
leys | -4.757 |
uum | -4.732 |
Token | American |
Token ID | 1605 |
Feature activation | +5.69e-01 |
Pos logit contributions | |
quartered | +0.736 |
uphem | +0.711 |
hern | +0.697 |
leys | +0.688 |
uum | +0.687 |
Neg logit contributions | |
Armstrong | -0.719 |
ppo | -0.669 |
pos | -0.645 |
Cot | -0.633 |
闘 | -0.621 |
Token | movies |
Token ID | 6918 |
Feature activation | +3.67e-01 |
Pos logit contributions | |
Armstrong | +1.228 |
ppo | +1.133 |
pos | +1.128 |
Cot | +1.090 |
Recon | +1.061 |
Neg logit contributions | |
quartered | -1.075 |
uphem | -1.038 |
hern | -1.007 |
uum | -0.994 |
leys | -0.992 |
Token | like |
Token ID | 588 |
Feature activation | +3.72e-01 |
Pos logit contributions | |
Armstrong | +0.643 |
ppo | +0.586 |
pos | +0.573 |
Cot | +0.553 |
闘 | +0.535 |
Neg logit contributions | |
quartered | -0.844 |
uphem | -0.819 |
hern | -0.799 |
leys | -0.793 |
uum | -0.792 |
Token | Meet |
Token ID | 21167 |
Feature activation | +7.93e-01 |
Pos logit contributions | |
Armstrong | +0.696 |
ppo | +0.639 |
pos | +0.622 |
Cot | +0.607 |
Recon | +0.589 |
Neg logit contributions | |
quartered | -0.809 |
uphem | -0.785 |
hern | -0.767 |
leys | -0.757 |
uum | -0.756 |
Token | during |
Token ID | 1141 |
Feature activation | -2.98e-02 |
Pos logit contributions | |
quartered | +0.572 |
uphem | +0.553 |
hern | +0.543 |
leys | +0.536 |
uum | +0.535 |
Neg logit contributions | |
Armstrong | -0.457 |
ppo | -0.422 |
pos | -0.404 |
Cot | -0.397 |
闘 | -0.389 |
Token | their |
Token ID | 511 |
Feature activation | +1.94e-01 |
Pos logit contributions | |
quartered | +0.063 |
uphem | +0.061 |
hern | +0.060 |
uum | +0.060 |
leys | +0.059 |
Neg logit contributions | |
Armstrong | -0.056 |
ppo | -0.052 |
pos | -0.049 |
Cot | -0.049 |
闘 | -0.048 |
Token | performance |
Token ID | 2854 |
Feature activation | -1.43e-01 |
Pos logit contributions | |
Armstrong | +0.382 |
ppo | +0.354 |
pos | +0.343 |
Cot | +0.336 |
闘 | +0.328 |
Neg logit contributions | |
quartered | -0.400 |
uphem | -0.387 |
hern | -0.378 |
uum | -0.373 |
leys | -0.373 |
Token | , |
Token ID | 11 |
Feature activation | +3.94e-01 |
Pos logit contributions | |
quartered | +0.316 |
uphem | +0.306 |
hern | +0.300 |
leys | +0.296 |
uum | +0.296 |
Neg logit contributions | |
Armstrong | -0.260 |
ppo | -0.240 |
pos | -0.231 |
Cot | -0.227 |
闘 | -0.221 |
Token | but |
Token ID | 475 |
Feature activation | +6.46e-01 |
Pos logit contributions | |
Armstrong | +0.687 |
ppo | +0.620 |
pos | +0.614 |
Cot | +0.593 |
Recon | +0.571 |
Neg logit contributions | |
quartered | -0.915 |
uphem | -0.886 |
hern | -0.865 |
uum | -0.856 |
leys | -0.856 |
Token | rather |
Token ID | 2138 |
Feature activation | +2.16e+00 |
Pos logit contributions | |
Armstrong | +1.041 |
ppo | +0.943 |
pos | +0.922 |
Cot | +0.888 |
Recon | +0.856 |
Neg logit contributions | |
quartered | -1.572 |
uphem | -1.530 |
hern | -1.497 |
uum | -1.483 |
leys | -1.479 |
Token | a |
Token ID | 257 |
Feature activation | +1.73e-01 |
Pos logit contributions | |
Armstrong | +3.857 |
pos | +3.554 |
ppo | +3.476 |
Cot | +3.342 |
Recon | +3.260 |
Neg logit contributions | |
quartered | -4.774 |
uphem | -4.678 |
hern | -4.547 |
uum | -4.520 |
leys | -4.470 |
Token | different |
Token ID | 1180 |
Feature activation | -3.56e-01 |
Pos logit contributions | |
Armstrong | +0.343 |
ppo | +0.316 |
pos | +0.312 |
Cot | +0.302 |
Recon | +0.294 |
Neg logit contributions | |
quartered | -0.358 |
uphem | -0.347 |
hern | -0.338 |
uum | -0.335 |
leys | -0.334 |
Token | stance |
Token ID | 12046 |
Feature activation | +9.29e-02 |
Pos logit contributions | |
quartered | +0.735 |
uphem | +0.710 |
hern | +0.695 |
uum | +0.686 |
leys | +0.685 |
Neg logit contributions | |
Armstrong | -0.701 |
ppo | -0.650 |
pos | -0.628 |
Cot | -0.617 |
闘 | -0.604 |
Token | on |
Token ID | 319 |
Feature activation | -2.56e-01 |
Pos logit contributions | |
Armstrong | +0.169 |
ppo | +0.155 |
pos | +0.150 |
Cot | +0.147 |
闘 | +0.143 |
Neg logit contributions | |
quartered | -0.206 |
uphem | -0.200 |
hern | -0.196 |
leys | -0.193 |
uum | -0.193 |
Token | how |
Token ID | 703 |
Feature activation | -3.59e-01 |
Pos logit contributions | |
quartered | +0.556 |
uphem | +0.540 |
hern | +0.528 |
leys | +0.521 |
uum | +0.521 |
Neg logit contributions | |
Armstrong | -0.477 |
ppo | -0.440 |
pos | -0.425 |
Cot | -0.416 |
闘 | -0.406 |
Token | Silver |
Token ID | 26766 |
Feature activation | +3.08e-01 |
Pos logit contributions | |
Armstrong | +0.089 |
ppo | +0.082 |
pos | +0.080 |
Cot | +0.078 |
闘 | +0.076 |
Neg logit contributions | |
quartered | -0.099 |
uphem | -0.096 |
hern | -0.093 |
leys | -0.092 |
uum | -0.092 |
Token | Bullet |
Token ID | 18003 |
Feature activation | +6.26e-01 |
Pos logit contributions | |
Armstrong | +0.298 |
ppo | +0.252 |
pos | +0.238 |
Cot | +0.228 |
闘 | +0.216 |
Neg logit contributions | |
quartered | -0.940 |
uphem | -0.925 |
hern | -0.908 |
uum | -0.904 |
leys | -0.899 |
Token | � |
Token ID | 447 |
Feature activation | -4.80e-01 |
Pos logit contributions | |
Armstrong | +1.151 |
ppo | +1.058 |
pos | +1.029 |
Cot | +1.002 |
闘 | +0.973 |
Neg logit contributions | |
quartered | -1.377 |
uphem | -1.335 |
hern | -1.307 |
leys | -1.291 |
uum | -1.289 |
Token | � |
Token ID | 251 |
Feature activation | +1.80e-01 |
Pos logit contributions | |
quartered | +0.863 |
uphem | +0.833 |
hern | +0.807 |
uum | +0.801 |
leys | +0.797 |
Neg logit contributions | |
Armstrong | -1.073 |
ppo | -1.002 |
pos | -0.976 |
Cot | -0.962 |
闘 | -0.939 |
Token | but |
Token ID | 475 |
Feature activation | +5.95e-01 |
Pos logit contributions | |
Armstrong | +0.330 |
ppo | +0.303 |
pos | +0.295 |
Cot | +0.287 |
闘 | +0.279 |
Neg logit contributions | |
quartered | -0.396 |
uphem | -0.384 |
hern | -0.376 |
leys | -0.372 |
uum | -0.371 |
Token | rather |
Token ID | 2138 |
Feature activation | +2.14e+00 |
Pos logit contributions | |
Armstrong | +0.935 |
ppo | +0.840 |
pos | +0.819 |
Cot | +0.790 |
Recon | +0.764 |
Neg logit contributions | |
quartered | -1.472 |
uphem | -1.436 |
hern | -1.401 |
uum | -1.388 |
leys | -1.387 |
Token | a |
Token ID | 257 |
Feature activation | -5.29e-04 |
Pos logit contributions | |
Armstrong | +3.739 |
pos | +3.370 |
ppo | +3.321 |
Cot | +3.195 |
Chain | +3.154 |
Neg logit contributions | |
quartered | -4.837 |
uphem | -4.755 |
hern | -4.589 |
leys | -4.552 |
uum | -4.527 |
Token | combination |
Token ID | 6087 |
Feature activation | -1.14e-01 |
Pos logit contributions | |
quartered | +0.001 |
uphem | +0.001 |
hern | +0.001 |
leys | +0.001 |
uum | +0.001 |
Neg logit contributions | |
Armstrong | -0.001 |
ppo | -0.001 |
pos | -0.001 |
Cot | -0.001 |
Recon | -0.001 |
Token | of |
Token ID | 286 |
Feature activation | -1.50e-01 |
Pos logit contributions | |
quartered | +0.224 |
uphem | +0.217 |
hern | +0.211 |
uum | +0.211 |
leys | +0.210 |
Neg logit contributions | |
Armstrong | -0.234 |
ppo | -0.220 |
pos | -0.210 |
Cot | -0.207 |
闘 | -0.203 |
Token | D |
Token ID | 360 |
Feature activation | -2.77e-01 |
Pos logit contributions | |
quartered | +0.334 |
uphem | +0.324 |
hern | +0.317 |
leys | +0.313 |
uum | +0.313 |
Neg logit contributions | |
Armstrong | -0.273 |
ppo | -0.251 |
pos | -0.243 |
Cot | -0.237 |
闘 | -0.231 |
Token | DoS |
Token ID | 46498 |
Feature activation | -1.73e-01 |
Pos logit contributions | |
quartered | +0.585 |
uphem | +0.570 |
hern | +0.558 |
uum | +0.551 |
leys | +0.550 |
Neg logit contributions | |
Armstrong | -0.532 |
ppo | -0.491 |
pos | -0.476 |
Cot | -0.467 |
闘 | -0.457 |
Token | — |
Token ID | 960 |
Feature activation | -1.54e-01 |
Pos logit contributions | |
quartered | +1.369 |
uphem | +1.321 |
hern | +1.302 |
leys | +1.279 |
uum | +1.278 |
Neg logit contributions | |
Armstrong | -1.123 |
ppo | -1.037 |
pos | -0.992 |
Cot | -0.979 |
闘 | -0.966 |
Token | these |
Token ID | 27218 |
Feature activation | -1.55e-01 |
Pos logit contributions | |
quartered | +0.306 |
uphem | +0.296 |
hern | +0.289 |
leys | +0.285 |
uum | +0.285 |
Neg logit contributions | |
Armstrong | -0.317 |
ppo | -0.294 |
pos | -0.286 |
Cot | -0.280 |
闘 | -0.274 |
Token | are |
Token ID | 389 |
Feature activation | +4.56e-01 |
Pos logit contributions | |
quartered | +0.324 |
uphem | +0.312 |
hern | +0.306 |
uum | +0.302 |
leys | +0.302 |
Neg logit contributions | |
Armstrong | -0.301 |
ppo | -0.281 |
pos | -0.269 |
Cot | -0.264 |
闘 | -0.261 |
Token | the |
Token ID | 262 |
Feature activation | +1.50e-01 |
Pos logit contributions | |
Armstrong | +0.854 |
ppo | +0.775 |
pos | +0.763 |
Cot | +0.743 |
Recon | +0.720 |
Neg logit contributions | |
quartered | -0.994 |
uphem | -0.968 |
hern | -0.942 |
leys | -0.934 |
uum | -0.931 |
Token | common |
Token ID | 2219 |
Feature activation | -5.35e-01 |
Pos logit contributions | |
Armstrong | +0.297 |
ppo | +0.275 |
pos | +0.268 |
Cot | +0.261 |
Recon | +0.254 |
Neg logit contributions | |
quartered | -0.309 |
uphem | -0.300 |
hern | -0.292 |
leys | -0.288 |
uum | -0.288 |
Token | threads |
Token ID | 14390 |
Feature activation | +2.12e+00 |
Pos logit contributions | |
quartered | +1.078 |
uphem | +1.036 |
hern | +1.020 |
leys | +1.009 |
uum | +1.003 |
Neg logit contributions | |
Armstrong | -1.072 |
ppo | -1.006 |
pos | -0.960 |
Cot | -0.946 |
闘 | -0.942 |
Token | of |
Token ID | 286 |
Feature activation | +3.78e-01 |
Pos logit contributions | |
Armstrong | +3.611 |
pos | +3.241 |
ppo | +3.148 |
Cot | +3.059 |
Chain | +3.005 |
Neg logit contributions | |
quartered | -4.873 |
uphem | -4.834 |
hern | -4.628 |
leys | -4.614 |
uum | -4.594 |
Token | my |
Token ID | 616 |
Feature activation | +2.96e-01 |
Pos logit contributions | |
Armstrong | +0.682 |
ppo | +0.624 |
pos | +0.608 |
Cot | +0.592 |
Recon | +0.573 |
Neg logit contributions | |
quartered | -0.848 |
uphem | -0.824 |
hern | -0.805 |
leys | -0.795 |
uum | -0.795 |
Token | family |
Token ID | 1641 |
Feature activation | +2.74e-01 |
Pos logit contributions | |
Armstrong | +0.594 |
ppo | +0.549 |
pos | +0.537 |
Cot | +0.523 |
Recon | +0.509 |
Neg logit contributions | |
quartered | -0.601 |
uphem | -0.583 |
hern | -0.568 |
leys | -0.559 |
uum | -0.559 |
Token | history |
Token ID | 2106 |
Feature activation | -2.04e-01 |
Pos logit contributions | |
Armstrong | +0.505 |
ppo | +0.465 |
pos | +0.450 |
Cot | +0.440 |
闘 | +0.428 |
Neg logit contributions | |
quartered | -0.602 |
uphem | -0.584 |
hern | -0.571 |
leys | -0.564 |
uum | -0.564 |
Token | and |
Token ID | 290 |
Feature activation | -9.07e-03 |
Pos logit contributions | |
quartered | +0.453 |
uphem | +0.436 |
hern | +0.429 |
leys | +0.423 |
uum | +0.423 |
Neg logit contributions | |
Armstrong | -0.371 |
ppo | -0.342 |
pos | -0.328 |
Cot | -0.323 |
闘 | -0.316 |
Token | greatest |
Token ID | 6000 |
Feature activation | -7.92e-02 |
Pos logit contributions | |
quartered | +0.985 |
uphem | +0.951 |
hern | +0.928 |
leys | +0.914 |
uum | +0.913 |
Neg logit contributions | |
Armstrong | -1.055 |
ppo | -0.980 |
pos | -0.953 |
Cot | -0.934 |
闘 | -0.913 |
Token | city |
Token ID | 1748 |
Feature activation | -5.19e-01 |
Pos logit contributions | |
quartered | +0.158 |
uphem | +0.153 |
hern | +0.149 |
uum | +0.147 |
leys | +0.147 |
Neg logit contributions | |
Armstrong | -0.163 |
ppo | -0.150 |
pos | -0.147 |
Cot | -0.144 |
闘 | -0.140 |
Token | felt |
Token ID | 2936 |
Feature activation | +3.47e-01 |
Pos logit contributions | |
quartered | +1.107 |
uphem | +1.073 |
hern | +1.049 |
leys | +1.035 |
uum | +1.035 |
Neg logit contributions | |
Armstrong | -0.986 |
ppo | -0.911 |
pos | -0.882 |
Cot | -0.863 |
闘 | -0.842 |
Token | compelled |
Token ID | 20232 |
Feature activation | -5.40e-01 |
Pos logit contributions | |
Armstrong | +0.723 |
ppo | +0.663 |
pos | +0.654 |
Cot | +0.638 |
闘 | +0.619 |
Neg logit contributions | |
quartered | -0.690 |
uphem | -0.664 |
hern | -0.647 |
uum | -0.638 |
leys | -0.637 |
Token | to |
Token ID | 284 |
Feature activation | -6.65e-02 |
Pos logit contributions | |
quartered | +1.037 |
uphem | +1.013 |
hern | +0.979 |
leys | +0.969 |
uum | +0.962 |
Neg logit contributions | |
Armstrong | -1.119 |
ppo | -1.044 |
pos | -1.010 |
Cot | -0.996 |
Recon | -0.983 |
Token | weigh |
Token ID | 10164 |
Feature activation | +2.12e+00 |
Pos logit contributions | |
quartered | +0.139 |
uphem | +0.134 |
hern | +0.131 |
uum | +0.129 |
leys | +0.129 |
Neg logit contributions | |
Armstrong | -0.131 |
ppo | -0.121 |
pos | -0.118 |
Cot | -0.115 |
闘 | -0.112 |
Token | in |
Token ID | 287 |
Feature activation | +5.12e-02 |
Pos logit contributions | |
Armstrong | +3.603 |
ppo | +3.209 |
pos | +3.203 |
Cot | +3.068 |
闘 | +2.972 |
Neg logit contributions | |
quartered | -5.001 |
uphem | -4.768 |
hern | -4.691 |
uum | -4.649 |
leys | -4.603 |
Token | last |
Token ID | 938 |
Feature activation | +6.94e-01 |
Pos logit contributions | |
Armstrong | +0.091 |
ppo | +0.083 |
pos | +0.081 |
Cot | +0.079 |
闘 | +0.076 |
Neg logit contributions | |
quartered | -0.116 |
uphem | -0.113 |
hern | -0.110 |
uum | -0.109 |
leys | -0.109 |
Token | week |
Token ID | 1285 |
Feature activation | +3.99e-01 |
Pos logit contributions | |
Armstrong | +1.128 |
ppo | +1.021 |
pos | +0.994 |
Cot | +0.963 |
闘 | +0.929 |
Neg logit contributions | |
quartered | -1.680 |
uphem | -1.630 |
hern | -1.595 |
uum | -1.577 |
leys | -1.577 |
Token | . |
Token ID | 13 |
Feature activation | -2.71e-01 |
Pos logit contributions | |
Armstrong | +0.699 |
ppo | +0.637 |
pos | +0.623 |
Cot | +0.603 |
闘 | +0.583 |
Neg logit contributions | |
quartered | -0.917 |
uphem | -0.890 |
hern | -0.870 |
uum | -0.857 |
leys | -0.855 |
Token |
|
Token ID | 198 |
Feature activation | -4.37e-01 |
Pos logit contributions | |
quartered | +0.599 |
uphem | +0.582 |
hern | +0.570 |
leys | +0.563 |
uum | +0.562 |
Neg logit contributions | |
Armstrong | -0.491 |
ppo | -0.453 |
pos | -0.439 |
Cot | -0.427 |
闘 | -0.417 |
Token | most |
Token ID | 749 |
Feature activation | -4.09e-02 |
Pos logit contributions | |
quartered | +0.164 |
uphem | +0.158 |
hern | +0.153 |
uum | +0.151 |
leys | +0.151 |
Neg logit contributions | |
Armstrong | -0.204 |
ppo | -0.190 |
pos | -0.187 |
Cot | -0.182 |
Recon | -0.178 |
Token | used |
Token ID | 973 |
Feature activation | +3.80e-01 |
Pos logit contributions | |
quartered | +0.073 |
uphem | +0.070 |
hern | +0.068 |
leys | +0.067 |
uum | +0.067 |
Neg logit contributions | |
Armstrong | -0.093 |
ppo | -0.086 |
pos | -0.085 |
Cot | -0.083 |
Recon | -0.081 |
Token | to |
Token ID | 284 |
Feature activation | +7.37e-02 |
Pos logit contributions | |
Armstrong | +0.692 |
ppo | +0.629 |
pos | +0.627 |
Cot | +0.597 |
Recon | +0.584 |
Neg logit contributions | |
quartered | -0.854 |
uphem | -0.823 |
hern | -0.802 |
leys | -0.796 |
uum | -0.793 |
Token | , |
Token ID | 11 |
Feature activation | +2.41e-01 |
Pos logit contributions | |
Armstrong | +0.138 |
ppo | +0.127 |
pos | +0.124 |
Cot | +0.121 |
闘 | +0.117 |
Neg logit contributions | |
quartered | -0.160 |
uphem | -0.155 |
hern | -0.151 |
leys | -0.149 |
uum | -0.149 |
Token | but |
Token ID | 475 |
Feature activation | +3.00e-01 |
Pos logit contributions | |
Armstrong | +0.428 |
ppo | +0.390 |
pos | +0.383 |
Cot | +0.371 |
Recon | +0.359 |
Neg logit contributions | |
quartered | -0.546 |
uphem | -0.529 |
hern | -0.517 |
uum | -0.512 |
leys | -0.511 |
Token | rather |
Token ID | 2138 |
Feature activation | +2.09e+00 |
Pos logit contributions | |
Armstrong | +0.518 |
ppo | +0.471 |
pos | +0.462 |
Cot | +0.446 |
Recon | +0.432 |
Neg logit contributions | |
quartered | -0.697 |
uphem | -0.675 |
hern | -0.661 |
uum | -0.653 |
leys | -0.653 |
Token | adds |
Token ID | 6673 |
Feature activation | +1.12e-01 |
Pos logit contributions | |
Armstrong | +3.749 |
pos | +3.471 |
ppo | +3.387 |
Cot | +3.243 |
eff | +3.185 |
Neg logit contributions | |
quartered | -4.678 |
uphem | -4.529 |
hern | -4.420 |
uum | -4.356 |
leys | -4.337 |
Token | nuance |
Token ID | 47128 |
Feature activation | +1.97e-01 |
Pos logit contributions | |
Armstrong | +0.214 |
ppo | +0.197 |
pos | +0.191 |
Cot | +0.187 |
闘 | +0.183 |
Neg logit contributions | |
quartered | -0.238 |
uphem | -0.231 |
hern | -0.225 |
leys | -0.222 |
uum | -0.222 |
Token | and |
Token ID | 290 |
Feature activation | -1.09e-01 |
Pos logit contributions | |
Armstrong | +0.364 |
ppo | +0.335 |
pos | +0.325 |
Cot | +0.317 |
闘 | +0.308 |
Neg logit contributions | |
quartered | -0.432 |
uphem | -0.419 |
hern | -0.410 |
uum | -0.405 |
leys | -0.405 |
Token | depth |
Token ID | 6795 |
Feature activation | +8.44e-01 |
Pos logit contributions | |
quartered | +0.222 |
uphem | +0.215 |
hern | +0.211 |
leys | +0.208 |
uum | +0.207 |
Neg logit contributions | |
Armstrong | -0.215 |
ppo | -0.200 |
pos | -0.193 |
Cot | -0.189 |
闘 | -0.185 |
Token | . |
Token ID | 13 |
Feature activation | +1.15e-02 |
Pos logit contributions | |
Armstrong | +1.559 |
ppo | +1.435 |
pos | +1.407 |
Cot | +1.358 |
Recon | +1.315 |
Neg logit contributions | |
quartered | -1.851 |
uphem | -1.789 |
hern | -1.743 |
uum | -1.736 |
leys | -1.732 |
Token | batches |
Token ID | 37830 |
Feature activation | +1.24e+00 |
Pos logit contributions | |
quartered | +0.050 |
uphem | +0.048 |
hern | +0.047 |
leys | +0.047 |
uum | +0.047 |
Neg logit contributions | |
Armstrong | -0.042 |
ppo | -0.039 |
pos | -0.038 |
Cot | -0.037 |
闘 | -0.036 |
Token | : |
Token ID | 25 |
Feature activation | +1.02e-01 |
Pos logit contributions | |
Armstrong | +2.217 |
ppo | +2.012 |
pos | +1.978 |
Cot | +1.918 |
Recon | +1.863 |
Neg logit contributions | |
quartered | -2.799 |
uphem | -2.737 |
hern | -2.677 |
leys | -2.619 |
uum | -2.613 |
Token |
|
Token ID | 198 |
Feature activation | +8.73e-02 |
Pos logit contributions | |
Armstrong | +0.181 |
ppo | +0.166 |
pos | +0.161 |
Cot | +0.157 |
闘 | +0.152 |
Neg logit contributions | |
quartered | -0.233 |
uphem | -0.226 |
hern | -0.222 |
leys | -0.219 |
uum | -0.219 |
Token |
|
Token ID | 198 |
Feature activation | +9.29e-02 |
Pos logit contributions | |
Armstrong | +0.142 |
ppo | +0.129 |
pos | +0.126 |
Cot | +0.120 |
Recon | +0.118 |
Neg logit contributions | |
quartered | -0.220 |
uphem | -0.209 |
hern | -0.202 |
leys | -0.201 |
uum | -0.200 |
Token | B |
Token ID | 33 |
Feature activation | -1.61e-01 |
Pos logit contributions | |
Armstrong | +0.172 |
ppo | +0.158 |
pos | +0.153 |
Cot | +0.150 |
闘 | +0.146 |
Neg logit contributions | |
quartered | -0.203 |
uphem | -0.197 |
hern | -0.193 |
leys | -0.190 |
uum | -0.190 |
Token | atch |
Token ID | 963 |
Feature activation | +2.07e+00 |
Pos logit contributions | |
quartered | +0.496 |
uphem | +0.487 |
hern | +0.480 |
leys | +0.477 |
uum | +0.477 |
Neg logit contributions | |
Armstrong | -0.153 |
ppo | -0.129 |
pos | -0.122 |
Cot | -0.115 |
闘 | -0.110 |
Token | 23 |
Token ID | 2242 |
Feature activation | +3.11e-01 |
Pos logit contributions | |
Armstrong | +3.566 |
ppo | +3.204 |
pos | +3.164 |
Cot | +3.081 |
Recon | +3.010 |
Neg logit contributions | |
quartered | -4.753 |
uphem | -4.623 |
hern | -4.499 |
uum | -4.398 |
leys | -4.374 |
Token |
|
Token ID | 198 |
Feature activation | +7.38e-02 |
Pos logit contributions | |
Armstrong | +0.531 |
ppo | +0.485 |
pos | +0.470 |
Cot | +0.457 |
闘 | +0.443 |
Neg logit contributions | |
quartered | -0.727 |
uphem | -0.706 |
hern | -0.692 |
leys | -0.683 |
uum | -0.682 |
Token |
|
Token ID | 198 |
Feature activation | +1.03e-01 |
Pos logit contributions | |
Armstrong | +0.121 |
ppo | +0.109 |
pos | +0.107 |
Cot | +0.102 |
Recon | +0.100 |
Neg logit contributions | |
quartered | -0.185 |
uphem | -0.175 |
hern | -0.170 |
leys | -0.169 |
uum | -0.169 |
Token | B |
Token ID | 33 |
Feature activation | -1.04e-01 |
Pos logit contributions | |
Armstrong | +0.198 |
ppo | +0.183 |
pos | +0.178 |
Cot | +0.174 |
闘 | +0.170 |
Neg logit contributions | |
quartered | -0.216 |
uphem | -0.209 |
hern | -0.204 |
leys | -0.202 |
uum | -0.202 |
Token | atch |
Token ID | 963 |
Feature activation | +1.83e+00 |
Pos logit contributions | |
quartered | +0.318 |
uphem | +0.313 |
hern | +0.308 |
uum | +0.307 |
leys | +0.306 |
Neg logit contributions | |
Armstrong | -0.098 |
ppo | -0.082 |
pos | -0.079 |
Cot | -0.074 |
闘 | -0.071 |
Token | of |
Token ID | 286 |
Feature activation | -4.06e-01 |
Pos logit contributions | |
quartered | +0.498 |
uphem | +0.486 |
hern | +0.471 |
uum | +0.465 |
leys | +0.464 |
Neg logit contributions | |
Armstrong | -0.415 |
ppo | -0.375 |
pos | -0.372 |
Cot | -0.359 |
Recon | -0.349 |
Token | Charleston |
Token ID | 26070 |
Feature activation | +4.51e-02 |
Pos logit contributions | |
quartered | +0.828 |
uphem | +0.799 |
hern | +0.784 |
leys | +0.771 |
uum | +0.769 |
Neg logit contributions | |
Armstrong | -0.803 |
ppo | -0.753 |
pos | -0.723 |
Cot | -0.708 |
闘 | -0.702 |
Token | . |
Token ID | 13 |
Feature activation | -4.47e-02 |
Pos logit contributions | |
Armstrong | +0.082 |
ppo | +0.076 |
pos | +0.073 |
Cot | +0.072 |
闘 | +0.070 |
Neg logit contributions | |
quartered | -0.100 |
uphem | -0.097 |
hern | -0.095 |
uum | -0.094 |
leys | -0.093 |
Token |
|
Token ID | 198 |
Feature activation | -1.07e-01 |
Pos logit contributions | |
quartered | +0.097 |
uphem | +0.094 |
hern | +0.092 |
leys | +0.091 |
uum | +0.091 |
Neg logit contributions | |
Armstrong | -0.082 |
ppo | -0.076 |
pos | -0.074 |
Cot | -0.071 |
闘 | -0.070 |
Token |
|
Token ID | 198 |
Feature activation | -1.20e-03 |
Pos logit contributions | |
quartered | +0.264 |
uphem | +0.252 |
hern | +0.245 |
leys | +0.242 |
uum | +0.242 |
Neg logit contributions | |
Armstrong | -0.178 |
ppo | -0.161 |
pos | -0.157 |
Cot | -0.151 |
Recon | -0.147 |
Token | Rather |
Token ID | 27202 |
Feature activation | +2.04e+00 |
Pos logit contributions | |
quartered | +0.003 |
uphem | +0.002 |
hern | +0.002 |
leys | +0.002 |
uum | +0.002 |
Neg logit contributions | |
Armstrong | -0.002 |
ppo | -0.002 |
pos | -0.002 |
Cot | -0.002 |
闘 | -0.002 |
Token | than |
Token ID | 621 |
Feature activation | +1.30e+00 |
Pos logit contributions | |
Armstrong | +2.664 |
ppo | +2.340 |
pos | +2.272 |
Cot | +2.175 |
Recon | +2.075 |
Neg logit contributions | |
quartered | -5.529 |
uphem | -5.406 |
hern | -5.307 |
uum | -5.272 |
leys | -5.269 |
Token | releg |
Token ID | 24542 |
Feature activation | -1.52e+00 |
Pos logit contributions | |
Armstrong | +2.600 |
pos | +2.394 |
ppo | +2.364 |
Cot | +2.288 |
Recon | +2.232 |
Neg logit contributions | |
quartered | -2.673 |
uphem | -2.573 |
uum | -2.518 |
hern | -2.511 |
leys | -2.499 |
Token | ating |
Token ID | 803 |
Feature activation | -4.14e-01 |
Pos logit contributions | |
quartered | +3.356 |
uphem | +3.290 |
hern | +3.228 |
uum | +3.214 |
leys | +3.168 |
Neg logit contributions | |
Armstrong | -2.703 |
ppo | -2.493 |
pos | -2.418 |
Cot | -2.383 |
闘 | -2.354 |
Token | slavery |
Token ID | 13503 |
Feature activation | -1.06e-01 |
Pos logit contributions | |
quartered | +0.911 |
uphem | +0.884 |
hern | +0.871 |
leys | +0.852 |
uum | +0.850 |
Neg logit contributions | |
Armstrong | -0.748 |
ppo | -0.704 |
pos | -0.667 |
Cot | -0.653 |
闘 | -0.651 |
Token | to |
Token ID | 284 |
Feature activation | +1.99e-01 |
Pos logit contributions | |
quartered | +0.221 |
uphem | +0.214 |
hern | +0.210 |
uum | +0.206 |
leys | +0.205 |
Neg logit contributions | |
Armstrong | -0.205 |
ppo | -0.191 |
pos | -0.184 |
Cot | -0.181 |
闘 | -0.177 |
Token | races |
Token ID | 9558 |
Feature activation | +6.35e-01 |
Pos logit contributions | |
Armstrong | +0.647 |
ppo | +0.595 |
pos | +0.583 |
Cot | +0.567 |
Recon | +0.551 |
Neg logit contributions | |
quartered | -0.692 |
uphem | -0.671 |
hern | -0.655 |
uum | -0.646 |
leys | -0.645 |
Token | because |
Token ID | 780 |
Feature activation | -6.50e-02 |
Pos logit contributions | |
Armstrong | +1.148 |
ppo | +1.052 |
pos | +1.024 |
Cot | +0.996 |
Recon | +0.965 |
Neg logit contributions | |
quartered | -1.419 |
uphem | -1.378 |
hern | -1.347 |
uum | -1.331 |
leys | -1.330 |
Token | it |
Token ID | 340 |
Feature activation | +1.30e-01 |
Pos logit contributions | |
quartered | +0.141 |
uphem | +0.137 |
hern | +0.134 |
leys | +0.132 |
uum | +0.132 |
Neg logit contributions | |
Armstrong | -0.120 |
ppo | -0.112 |
pos | -0.108 |
Cot | -0.105 |
闘 | -0.104 |
Token | not |
Token ID | 407 |
Feature activation | +5.60e-02 |
Pos logit contributions | |
Armstrong | +0.228 |
ppo | +0.210 |
pos | +0.201 |
Cot | +0.198 |
闘 | +0.193 |
Neg logit contributions | |
quartered | -0.294 |
uphem | -0.286 |
hern | -0.280 |
leys | -0.277 |
uum | -0.277 |
Token | only |
Token ID | 691 |
Feature activation | -1.42e-01 |
Pos logit contributions | |
Armstrong | +0.115 |
ppo | +0.107 |
pos | +0.104 |
Cot | +0.102 |
闘 | +0.099 |
Neg logit contributions | |
quartered | -0.111 |
uphem | -0.107 |
hern | -0.105 |
leys | -0.103 |
uum | -0.103 |
Token | ske |
Token ID | 6146 |
Feature activation | +2.01e+00 |
Pos logit contributions | |
quartered | +0.318 |
uphem | +0.309 |
hern | +0.303 |
leys | +0.299 |
uum | +0.299 |
Neg logit contributions | |
Armstrong | -0.254 |
ppo | -0.234 |
pos | -0.225 |
Cot | -0.221 |
闘 | -0.216 |
Token | ws |
Token ID | 18504 |
Feature activation | +6.26e-01 |
Pos logit contributions | |
Armstrong | +1.962 |
ppo | +1.714 |
pos | +1.575 |
Cot | +1.420 |
eff | +1.339 |
Neg logit contributions | |
quartered | -6.139 |
uphem | -5.927 |
hern | -5.846 |
uum | -5.761 |
leys | -5.730 |
Token | voters |
Token ID | 4446 |
Feature activation | +1.83e-01 |
Pos logit contributions | |
Armstrong | +1.134 |
ppo | +1.035 |
pos | +1.011 |
Cot | +0.982 |
Recon | +0.950 |
Neg logit contributions | |
quartered | -1.399 |
uphem | -1.359 |
hern | -1.329 |
leys | -1.311 |
uum | -1.311 |
Token | � |
Token ID | 447 |
Feature activation | -2.04e-01 |
Pos logit contributions | |
Armstrong | +0.336 |
ppo | +0.310 |
pos | +0.300 |
Cot | +0.293 |
闘 | +0.285 |
Neg logit contributions | |
quartered | -0.401 |
uphem | -0.389 |
hern | -0.381 |
leys | -0.376 |
uum | -0.376 |
Token | � |
Token ID | 247 |
Feature activation | -2.78e-01 |
Pos logit contributions | |
quartered | +0.393 |
uphem | +0.378 |
hern | +0.369 |
leys | +0.365 |
uum | +0.362 |
Neg logit contributions | |
Armstrong | -0.433 |
ppo | -0.404 |
pos | -0.395 |
Cot | -0.385 |
闘 | -0.375 |
Token | choices |
Token ID | 7747 |
Feature activation | +5.47e-01 |
Pos logit contributions | |
quartered | +0.585 |
uphem | +0.566 |
hern | +0.554 |
leys | +0.546 |
uum | +0.546 |
Neg logit contributions | |
Armstrong | -0.539 |
ppo | -0.498 |
pos | -0.483 |
Cot | -0.472 |
闘 | -0.461 |
Token | , |
Token ID | 11 |
Feature activation | -1.55e-01 |
Pos logit contributions | |
quartered | +0.241 |
uphem | +0.234 |
hern | +0.229 |
leys | +0.226 |
uum | +0.226 |
Neg logit contributions | |
Armstrong | -0.192 |
ppo | -0.176 |
pos | -0.170 |
Cot | -0.166 |
闘 | -0.162 |
Token | while |
Token ID | 981 |
Feature activation | -8.84e-01 |
Pos logit contributions | |
quartered | +0.347 |
uphem | +0.336 |
hern | +0.327 |
uum | +0.324 |
leys | +0.323 |
Neg logit contributions | |
Armstrong | -0.283 |
ppo | -0.258 |
pos | -0.253 |
Cot | -0.247 |
Recon | -0.238 |
Token | the |
Token ID | 262 |
Feature activation | +3.52e-02 |
Pos logit contributions | |
quartered | +1.920 |
uphem | +1.869 |
hern | +1.828 |
leys | +1.811 |
uum | +1.798 |
Neg logit contributions | |
Armstrong | -1.624 |
ppo | -1.506 |
pos | -1.442 |
Cot | -1.407 |
闘 | -1.398 |
Token | Bur |
Token ID | 5481 |
Feature activation | +3.51e-01 |
Pos logit contributions | |
Armstrong | +0.071 |
ppo | +0.066 |
pos | +0.064 |
Cot | +0.063 |
Recon | +0.061 |
Neg logit contributions | |
quartered | -0.071 |
uphem | -0.069 |
hern | -0.067 |
uum | -0.066 |
leys | -0.066 |
Token | m |
Token ID | 76 |
Feature activation | +9.01e-01 |
Pos logit contributions | |
Armstrong | +0.686 |
ppo | +0.641 |
pos | +0.619 |
Cot | +0.604 |
闘 | +0.600 |
Neg logit contributions | |
quartered | -0.728 |
uphem | -0.704 |
hern | -0.695 |
leys | -0.688 |
uum | -0.679 |
Token | ese |
Token ID | 2771 |
Feature activation | +2.00e+00 |
Pos logit contributions | |
Armstrong | +1.928 |
ppo | +1.794 |
pos | +1.747 |
Cot | +1.711 |
闘 | +1.675 |
Neg logit contributions | |
quartered | -1.712 |
uphem | -1.653 |
hern | -1.611 |
leys | -1.587 |
uum | -1.586 |
Token | authorities |
Token ID | 4773 |
Feature activation | -2.31e-01 |
Pos logit contributions | |
Armstrong | +4.208 |
pos | +3.916 |
ppo | +3.884 |
Cot | +3.816 |
Recon | +3.767 |
Neg logit contributions | |
quartered | -3.762 |
uphem | -3.610 |
hern | -3.590 |
uum | -3.517 |
aukee | -3.466 |
Token | said |
Token ID | 531 |
Feature activation | +1.94e-01 |
Pos logit contributions | |
quartered | +0.414 |
uphem | +0.399 |
hern | +0.388 |
leys | +0.382 |
uum | +0.382 |
Neg logit contributions | |
Armstrong | -0.517 |
ppo | -0.483 |
pos | -0.471 |
Cot | -0.462 |
闘 | -0.453 |
Token | there |
Token ID | 612 |
Feature activation | +1.78e-01 |
Pos logit contributions | |
Armstrong | +0.353 |
ppo | +0.323 |
pos | +0.315 |
Cot | +0.307 |
Recon | +0.296 |
Neg logit contributions | |
quartered | -0.434 |
uphem | -0.421 |
hern | -0.412 |
uum | -0.406 |
leys | -0.406 |
Token | had |
Token ID | 550 |
Feature activation | +1.26e-01 |
Pos logit contributions | |
Armstrong | +0.366 |
ppo | +0.340 |
pos | +0.330 |
Cot | +0.323 |
闘 | +0.316 |
Neg logit contributions | |
quartered | -0.352 |
uphem | -0.341 |
hern | -0.333 |
leys | -0.328 |
uum | -0.328 |
Token | been |
Token ID | 587 |
Feature activation | -2.26e-01 |
Pos logit contributions | |
Armstrong | +0.292 |
ppo | +0.274 |
pos | +0.268 |
Cot | +0.262 |
闘 | +0.257 |
Neg logit contributions | |
quartered | -0.217 |
uphem | -0.209 |
hern | -0.203 |
leys | -0.200 |
uum | -0.200 |
Token | during |
Token ID | 1141 |
Feature activation | +2.77e-01 |
Pos logit contributions | |
quartered | +0.440 |
uphem | +0.427 |
hern | +0.419 |
leys | +0.413 |
uum | +0.413 |
Neg logit contributions | |
Armstrong | -0.356 |
ppo | -0.327 |
pos | -0.316 |
Cot | -0.309 |
闘 | -0.301 |
Token | this |
Token ID | 428 |
Feature activation | +6.48e-02 |
Pos logit contributions | |
Armstrong | +0.494 |
ppo | +0.453 |
pos | +0.439 |
Cot | +0.428 |
闘 | +0.416 |
Neg logit contributions | |
quartered | -0.626 |
uphem | -0.608 |
hern | -0.595 |
leys | -0.587 |
uum | -0.587 |
Token | time |
Token ID | 640 |
Feature activation | +1.31e-01 |
Pos logit contributions | |
Armstrong | +0.116 |
ppo | +0.106 |
pos | +0.103 |
Cot | +0.100 |
闘 | +0.098 |
Neg logit contributions | |
quartered | -0.146 |
uphem | -0.141 |
hern | -0.139 |
uum | -0.137 |
leys | -0.137 |
Token | period |
Token ID | 2278 |
Feature activation | +3.02e-01 |
Pos logit contributions | |
Armstrong | +0.247 |
ppo | +0.228 |
pos | +0.221 |
Cot | +0.216 |
闘 | +0.211 |
Neg logit contributions | |
quartered | -0.282 |
uphem | -0.273 |
hern | -0.267 |
uum | -0.264 |
leys | -0.263 |
Token | but |
Token ID | 475 |
Feature activation | +6.58e-01 |
Pos logit contributions | |
Armstrong | +0.544 |
ppo | +0.500 |
pos | +0.484 |
Cot | +0.472 |
闘 | +0.460 |
Neg logit contributions | |
quartered | -0.674 |
uphem | -0.654 |
hern | -0.640 |
leys | -0.632 |
uum | -0.632 |
Token | rather |
Token ID | 2138 |
Feature activation | +2.00e+00 |
Pos logit contributions | |
Armstrong | +1.199 |
ppo | +1.080 |
pos | +1.072 |
Cot | +1.039 |
Recon | +0.997 |
Neg logit contributions | |
quartered | -1.474 |
uphem | -1.442 |
hern | -1.386 |
uum | -1.377 |
leys | -1.376 |
Token | " |
Token ID | 366 |
Feature activation | -4.65e-02 |
Pos logit contributions | |
Armstrong | +3.602 |
pos | +3.219 |
ppo | +3.139 |
Cot | +3.119 |
eff | +2.997 |
Neg logit contributions | |
quartered | -4.495 |
uphem | -4.433 |
leys | -4.189 |
uum | -4.185 |
hern | -4.174 |
Token | on |
Token ID | 261 |
Feature activation | -9.74e-03 |
Pos logit contributions | |
quartered | +0.092 |
uphem | +0.089 |
hern | +0.086 |
leys | +0.085 |
uum | +0.085 |
Neg logit contributions | |
Armstrong | -0.096 |
ppo | -0.089 |
pos | -0.087 |
Cot | -0.085 |
闘 | -0.083 |
Token | the |
Token ID | 262 |
Feature activation | -3.37e-01 |
Pos logit contributions | |
quartered | +0.021 |
uphem | +0.020 |
hern | +0.020 |
leys | +0.019 |
uum | +0.019 |
Neg logit contributions | |
Armstrong | -0.019 |
ppo | -0.017 |
pos | -0.017 |
Cot | -0.016 |
闘 | -0.016 |
Token | other |
Token ID | 584 |
Feature activation | -1.04e-01 |
Pos logit contributions | |
quartered | +0.671 |
uphem | +0.649 |
hern | +0.634 |
uum | +0.623 |
leys | +0.623 |
Neg logit contributions | |
Armstrong | -0.694 |
ppo | -0.642 |
pos | -0.629 |
Cot | -0.614 |
Recon | -0.597 |
Token | side |
Token ID | 1735 |
Feature activation | +1.97e-01 |
Pos logit contributions | |
quartered | +0.201 |
uphem | +0.194 |
hern | +0.190 |
uum | +0.187 |
leys | +0.187 |
Neg logit contributions | |
Armstrong | -0.218 |
ppo | -0.203 |
pos | -0.198 |
Cot | -0.194 |
闘 | -0.189 |
Token | t |
Token ID | 83 |
Feature activation | +9.75e-02 |
Pos logit contributions | |
quartered | +0.123 |
uphem | +0.120 |
hern | +0.117 |
leys | +0.115 |
uum | +0.115 |
Neg logit contributions | |
Armstrong | -0.099 |
ppo | -0.091 |
pos | -0.088 |
Cot | -0.086 |
闘 | -0.083 |
Token | like |
Token ID | 588 |
Feature activation | +4.49e-01 |
Pos logit contributions | |
Armstrong | +0.179 |
ppo | +0.163 |
pos | +0.160 |
Cot | +0.154 |
Recon | +0.151 |
Neg logit contributions | |
quartered | -0.217 |
uphem | -0.210 |
hern | -0.205 |
uum | -0.203 |
leys | -0.203 |
Token | the |
Token ID | 262 |
Feature activation | -8.86e-02 |
Pos logit contributions | |
Armstrong | +0.812 |
ppo | +0.739 |
pos | +0.721 |
Cot | +0.704 |
Recon | +0.681 |
Neg logit contributions | |
quartered | -1.003 |
uphem | -0.976 |
hern | -0.955 |
uum | -0.943 |
leys | -0.939 |
Token | way |
Token ID | 835 |
Feature activation | -4.48e-01 |
Pos logit contributions | |
quartered | +0.187 |
uphem | +0.181 |
hern | +0.177 |
leys | +0.175 |
uum | +0.175 |
Neg logit contributions | |
Armstrong | -0.171 |
ppo | -0.158 |
pos | -0.153 |
Cot | -0.150 |
闘 | -0.146 |
Token | they |
Token ID | 484 |
Feature activation | -5.22e-01 |
Pos logit contributions | |
quartered | +1.002 |
uphem | +0.969 |
hern | +0.949 |
leys | +0.941 |
uum | +0.938 |
Neg logit contributions | |
Armstrong | -0.801 |
ppo | -0.740 |
pos | -0.713 |
Cot | -0.696 |
闘 | -0.684 |
Token | treat |
Token ID | 2190 |
Feature activation | -2.40e+00 |
Pos logit contributions | |
quartered | +1.068 |
uphem | +1.034 |
hern | +1.007 |
uum | +0.996 |
leys | +0.995 |
Neg logit contributions | |
Armstrong | -1.039 |
ppo | -0.961 |
pos | -0.933 |
Cot | -0.915 |
闘 | -0.893 |
Token | people |
Token ID | 661 |
Feature activation | -6.22e-01 |
Pos logit contributions | |
quartered | +4.965 |
uphem | +4.727 |
uum | +4.682 |
hern | +4.652 |
leys | +4.605 |
Neg logit contributions | |
Armstrong | -4.344 |
ppo | -4.155 |
闘 | -4.009 |
pos | -3.905 |
Cot | -3.826 |
Token | � |
Token ID | 447 |
Feature activation | -2.40e-01 |
Pos logit contributions | |
quartered | +1.382 |
uphem | +1.339 |
hern | +1.312 |
uum | +1.295 |
leys | +1.295 |
Neg logit contributions | |
Armstrong | -1.126 |
ppo | -1.036 |
pos | -1.002 |
Cot | -0.978 |
闘 | -0.956 |
Token | � |
Token ID | 247 |
Feature activation | -6.74e-02 |
Pos logit contributions | |
quartered | +0.485 |
uphem | +0.466 |
hern | +0.455 |
leys | +0.448 |
uum | +0.445 |
Neg logit contributions | |
Armstrong | -0.488 |
ppo | -0.450 |
pos | -0.441 |
Cot | -0.428 |
闘 | -0.420 |
Token | s |
Token ID | 82 |
Feature activation | -1.36e-01 |
Pos logit contributions | |
quartered | +0.142 |
uphem | +0.138 |
hern | +0.135 |
leys | +0.132 |
uum | +0.132 |
Neg logit contributions | |
Armstrong | -0.130 |
ppo | -0.120 |
pos | -0.117 |
Cot | -0.114 |
Recon | -0.110 |
Token | privacy |
Token ID | 6782 |
Feature activation | -2.76e-01 |
Pos logit contributions | |
quartered | +0.275 |
uphem | +0.266 |
hern | +0.259 |
leys | +0.255 |
uum | +0.255 |
Neg logit contributions | |
Armstrong | -0.277 |
ppo | -0.256 |
pos | -0.250 |
Cot | -0.245 |
闘 | -0.238 |
Token | . |
Token ID | 13 |
Feature activation | -5.79e-02 |
Pos logit contributions | |
quartered | +0.395 |
uphem | +0.383 |
hern | +0.376 |
uum | +0.371 |
leys | +0.370 |
Neg logit contributions | |
Armstrong | -0.316 |
ppo | -0.291 |
pos | -0.280 |
Cot | -0.274 |
闘 | -0.267 |
Token |
|
Token ID | 198 |
Feature activation | -2.30e-01 |
Pos logit contributions | |
quartered | +0.127 |
uphem | +0.124 |
hern | +0.121 |
leys | +0.120 |
uum | +0.120 |
Neg logit contributions | |
Armstrong | -0.105 |
ppo | -0.097 |
pos | -0.094 |
Cot | -0.091 |
闘 | -0.089 |
Token |
|
Token ID | 198 |
Feature activation | -7.29e-02 |
Pos logit contributions | |
quartered | +0.560 |
uphem | +0.536 |
hern | +0.521 |
leys | +0.516 |
uum | +0.514 |
Neg logit contributions | |
Armstrong | -0.386 |
ppo | -0.351 |
pos | -0.343 |
Cot | -0.329 |
Recon | -0.320 |
Token | " |
Token ID | 1 |
Feature activation | +1.10e-02 |
Pos logit contributions | |
quartered | +0.153 |
uphem | +0.148 |
hern | +0.144 |
leys | +0.142 |
uum | +0.142 |
Neg logit contributions | |
Armstrong | -0.142 |
ppo | -0.131 |
pos | -0.127 |
Cot | -0.125 |
闘 | -0.121 |
Token | The |
Token ID | 464 |
Feature activation | +1.57e-01 |
Pos logit contributions | |
Armstrong | +0.021 |
ppo | +0.019 |
pos | +0.019 |
Cot | +0.018 |
闘 | +0.018 |
Neg logit contributions | |
quartered | -0.024 |
uphem | -0.023 |
hern | -0.022 |
uum | -0.022 |
leys | -0.022 |
Token | reaction |
Token ID | 6317 |
Feature activation | -2.13e+00 |
Pos logit contributions | |
Armstrong | +0.308 |
ppo | +0.284 |
pos | +0.277 |
Cot | +0.271 |
Recon | +0.263 |
Neg logit contributions | |
quartered | -0.327 |
uphem | -0.317 |
hern | -0.309 |
uum | -0.304 |
leys | -0.304 |
Token | has |
Token ID | 468 |
Feature activation | -2.38e-01 |
Pos logit contributions | |
quartered | +4.225 |
hern | +4.031 |
uphem | +4.019 |
leys | +4.003 |
uum | +3.986 |
Neg logit contributions | |
Armstrong | -4.082 |
ppo | -3.886 |
闘 | -3.723 |
pos | -3.605 |
Cot | -3.593 |
Token | been |
Token ID | 587 |
Feature activation | -4.79e-01 |
Pos logit contributions | |
quartered | +0.404 |
uphem | +0.387 |
hern | +0.378 |
leys | +0.371 |
uum | +0.371 |
Neg logit contributions | |
Armstrong | -0.558 |
ppo | -0.523 |
pos | -0.510 |
Cot | -0.501 |
闘 | -0.492 |
Token | crazy |
Token ID | 7165 |
Feature activation | -5.44e-01 |
Pos logit contributions | |
quartered | +0.935 |
uphem | +0.902 |
hern | +0.885 |
leys | +0.872 |
uum | +0.871 |
Neg logit contributions | |
Armstrong | -0.989 |
ppo | -0.923 |
pos | -0.892 |
Cot | -0.878 |
闘 | -0.865 |
Token | . |
Token ID | 13 |
Feature activation | -3.59e-02 |
Pos logit contributions | |
quartered | +1.177 |
uphem | +1.137 |
hern | +1.120 |
leys | +1.104 |
uum | +1.103 |
Neg logit contributions | |
Armstrong | -1.009 |
ppo | -0.936 |
pos | -0.898 |
Cot | -0.883 |
闘 | -0.869 |
Token | I |
Token ID | 314 |
Feature activation | +1.05e-02 |
Pos logit contributions | |
quartered | +0.078 |
uphem | +0.076 |
hern | +0.074 |
uum | +0.072 |
leys | +0.072 |
Neg logit contributions | |
Armstrong | -0.069 |
ppo | -0.063 |
pos | -0.061 |
Cot | -0.060 |
Recon | -0.058 |
Token | , |
Token ID | 11 |
Feature activation | -8.66e-02 |
Pos logit contributions | |
quartered | +0.152 |
uphem | +0.148 |
hern | +0.145 |
leys | +0.143 |
uum | +0.143 |
Neg logit contributions | |
Armstrong | -0.125 |
ppo | -0.114 |
pos | -0.111 |
Cot | -0.108 |
闘 | -0.105 |
Token | and |
Token ID | 290 |
Feature activation | -1.32e-01 |
Pos logit contributions | |
quartered | +0.185 |
uphem | +0.180 |
hern | +0.175 |
uum | +0.173 |
leys | +0.173 |
Neg logit contributions | |
Armstrong | -0.165 |
ppo | -0.152 |
pos | -0.148 |
Cot | -0.145 |
闘 | -0.140 |
Token | is |
Token ID | 318 |
Feature activation | -1.38e-01 |
Pos logit contributions | |
quartered | +0.282 |
uphem | +0.273 |
hern | +0.267 |
leys | +0.264 |
uum | +0.263 |
Neg logit contributions | |
Armstrong | -0.253 |
ppo | -0.234 |
pos | -0.227 |
Cot | -0.222 |
闘 | -0.216 |
Token | a |
Token ID | 257 |
Feature activation | -1.52e-01 |
Pos logit contributions | |
quartered | +0.277 |
uphem | +0.268 |
hern | +0.261 |
leys | +0.258 |
uum | +0.257 |
Neg logit contributions | |
Armstrong | -0.279 |
ppo | -0.259 |
pos | -0.253 |
Cot | -0.247 |
闘 | -0.240 |
Token | potential |
Token ID | 2785 |
Feature activation | -4.67e-02 |
Pos logit contributions | |
quartered | +0.308 |
uphem | +0.295 |
hern | +0.287 |
leys | +0.285 |
uum | +0.285 |
Neg logit contributions | |
Armstrong | -0.311 |
ppo | -0.286 |
pos | -0.283 |
Cot | -0.275 |
Recon | -0.267 |
Token | pand |
Token ID | 19798 |
Feature activation | -2.12e+00 |
Pos logit contributions | |
quartered | +0.089 |
uphem | +0.086 |
hern | +0.084 |
leys | +0.083 |
uum | +0.083 |
Neg logit contributions | |
Armstrong | -0.099 |
ppo | -0.092 |
pos | -0.090 |
Cot | -0.088 |
闘 | -0.086 |
Token | emic |
Token ID | 5314 |
Feature activation | -5.60e-01 |
Pos logit contributions | |
quartered | +1.250 |
uphem | +1.208 |
hern | +1.113 |
uum | +1.107 |
leys | +1.023 |
Neg logit contributions | |
Armstrong | -7.184 |
ppo | -6.838 |
pos | -6.746 |
Cot | -6.694 |
闘 | -6.622 |
Token | threat |
Token ID | 2372 |
Feature activation | -6.63e-01 |
Pos logit contributions | |
quartered | +1.231 |
uphem | +1.197 |
hern | +1.173 |
uum | +1.157 |
leys | +1.154 |
Neg logit contributions | |
Armstrong | -1.024 |
ppo | -0.943 |
pos | -0.911 |
Cot | -0.890 |
闘 | -0.871 |
Token | . |
Token ID | 13 |
Feature activation | +1.83e-03 |
Pos logit contributions | |
quartered | +1.449 |
uphem | +1.406 |
hern | +1.379 |
uum | +1.359 |
leys | +1.358 |
Neg logit contributions | |
Armstrong | -1.220 |
ppo | -1.127 |
pos | -1.088 |
Cot | -1.060 |
闘 | -1.040 |
Token | Phase |
Token ID | 18983 |
Feature activation | +4.11e-01 |
Pos logit contributions | |
Armstrong | +0.003 |
ppo | +0.003 |
pos | +0.003 |
Cot | +0.003 |
闘 | +0.003 |
Neg logit contributions | |
quartered | -0.004 |
uphem | -0.004 |
hern | -0.004 |
leys | -0.004 |
uum | -0.004 |
Token | 3 |
Token ID | 513 |
Feature activation | -4.61e-01 |
Pos logit contributions | |
Armstrong | +0.743 |
ppo | +0.685 |
pos | +0.662 |
Cot | +0.645 |
闘 | +0.630 |
Neg logit contributions | |
quartered | -0.914 |
uphem | -0.887 |
hern | -0.869 |
uum | -0.857 |
leys | -0.857 |
Token | two |
Token ID | 734 |
Feature activation | +3.26e-01 |
Pos logit contributions | |
quartered | +0.272 |
uphem | +0.264 |
hern | +0.256 |
uum | +0.254 |
leys | +0.253 |
Neg logit contributions | |
Armstrong | -0.206 |
ppo | -0.186 |
pos | -0.184 |
Cot | -0.179 |
Recon | -0.171 |
Token | countries |
Token ID | 2678 |
Feature activation | -4.71e-01 |
Pos logit contributions | |
Armstrong | +0.781 |
ppo | +0.723 |
pos | +0.722 |
Cot | +0.710 |
Recon | +0.688 |
Neg logit contributions | |
quartered | -0.542 |
uphem | -0.519 |
hern | -0.503 |
leys | -0.492 |
uum | -0.490 |
Token | . |
Token ID | 13 |
Feature activation | +1.22e-01 |
Pos logit contributions | |
quartered | +1.041 |
uphem | +1.011 |
hern | +0.994 |
uum | +0.980 |
leys | +0.979 |
Neg logit contributions | |
Armstrong | -0.851 |
ppo | -0.783 |
pos | -0.755 |
Cot | -0.738 |
闘 | -0.724 |
Token | Strong |
Token ID | 13535 |
Feature activation | +7.91e-01 |
Pos logit contributions | |
Armstrong | +0.219 |
ppo | +0.200 |
pos | +0.194 |
Cot | +0.190 |
闘 | +0.184 |
Neg logit contributions | |
quartered | -0.274 |
uphem | -0.266 |
hern | -0.260 |
leys | -0.257 |
uum | -0.256 |
Token | signal |
Token ID | 6737 |
Feature activation | +4.53e-01 |
Pos logit contributions | |
Armstrong | +1.571 |
ppo | +1.444 |
pos | +1.433 |
Cot | +1.383 |
Recon | +1.350 |
Neg logit contributions | |
quartered | -1.624 |
uphem | -1.571 |
hern | -1.529 |
uum | -1.510 |
leys | -1.504 |
Token | pand |
Token ID | 19798 |
Feature activation | -2.08e+00 |
Pos logit contributions | |
Armstrong | +0.826 |
ppo | +0.756 |
pos | +0.738 |
Cot | +0.717 |
Recon | +0.696 |
Neg logit contributions | |
quartered | -1.006 |
uphem | -0.977 |
hern | -0.956 |
uum | -0.943 |
leys | -0.942 |
Token | emic |
Token ID | 5314 |
Feature activation | -7.04e-01 |
Pos logit contributions | |
quartered | +1.974 |
uphem | +1.885 |
hern | +1.853 |
uum | +1.849 |
leys | +1.807 |
Neg logit contributions | |
Armstrong | -6.244 |
ppo | -5.923 |
pos | -5.792 |
Cot | -5.741 |
闘 | -5.705 |
Token | imminent |
Token ID | 21161 |
Feature activation | +6.14e-02 |
Pos logit contributions | |
quartered | +1.498 |
uphem | +1.453 |
hern | +1.426 |
uum | +1.406 |
leys | +1.405 |
Neg logit contributions | |
Armstrong | -1.331 |
ppo | -1.231 |
pos | -1.185 |
Cot | -1.164 |
闘 | -1.140 |
Token | . |
Token ID | 13 |
Feature activation | +2.51e-01 |
Pos logit contributions | |
Armstrong | +0.110 |
ppo | +0.101 |
pos | +0.098 |
Cot | +0.096 |
Recon | +0.093 |
Neg logit contributions | |
quartered | -0.139 |
uphem | -0.135 |
hern | -0.131 |
leys | -0.130 |
uum | -0.130 |
Token | Phase |
Token ID | 18983 |
Feature activation | +4.73e-01 |
Pos logit contributions | |
Armstrong | +0.465 |
ppo | +0.424 |
pos | +0.414 |
Cot | +0.406 |
Recon | +0.393 |
Neg logit contributions | |
quartered | -0.557 |
uphem | -0.537 |
hern | -0.525 |
uum | -0.518 |
leys | -0.518 |
Token | 6 |
Token ID | 718 |
Feature activation | -2.50e-01 |
Pos logit contributions | |
Armstrong | +0.868 |
ppo | +0.799 |
pos | +0.774 |
Cot | +0.755 |
闘 | +0.736 |
Neg logit contributions | |
quartered | -1.042 |
uphem | -1.012 |
hern | -0.990 |
leys | -0.977 |
uum | -0.977 |
Token | <|endoftext|> |
Token ID | 50256 |
Feature activation | +3.79e-01 |
Pos logit contributions | |
Armstrong | +0.055 |
ppo | +0.051 |
pos | +0.049 |
Cot | +0.048 |
闘 | +0.047 |
Neg logit contributions | |
quartered | -0.070 |
uphem | -0.068 |
hern | -0.068 |
uum | -0.066 |
leys | -0.066 |
Token | Image |
Token ID | 5159 |
Feature activation | -3.28e-01 |
Pos logit contributions | |
Armstrong | +0.747 |
ppo | +0.694 |
pos | +0.673 |
Cot | +0.656 |
闘 | +0.640 |
Neg logit contributions | |
quartered | -0.784 |
uphem | -0.756 |
hern | -0.738 |
leys | -0.729 |
uum | -0.728 |
Token | copyright |
Token ID | 6634 |
Feature activation | -8.76e-01 |
Pos logit contributions | |
quartered | +0.501 |
uphem | +0.480 |
hern | +0.465 |
uum | +0.456 |
leys | +0.455 |
Neg logit contributions | |
Armstrong | -0.821 |
ppo | -0.776 |
pos | -0.756 |
Cot | -0.744 |
闘 | -0.734 |
Token | ESA |
Token ID | 42212 |
Feature activation | +8.20e-02 |
Pos logit contributions | |
quartered | +2.027 |
uphem | +1.969 |
hern | +1.932 |
leys | +1.918 |
uum | +1.908 |
Neg logit contributions | |
Armstrong | -1.454 |
ppo | -1.358 |
pos | -1.290 |
Cot | -1.246 |
闘 | -1.246 |
Token | Image |
Token ID | 7412 |
Feature activation | -1.41e-01 |
Pos logit contributions | |
Armstrong | +0.138 |
ppo | +0.127 |
pos | +0.123 |
Cot | +0.119 |
闘 | +0.116 |
Neg logit contributions | |
quartered | -0.192 |
uphem | -0.187 |
hern | -0.183 |
uum | -0.181 |
leys | -0.181 |
Token | caption |
Token ID | 8305 |
Feature activation | -2.07e+00 |
Pos logit contributions | |
quartered | +0.260 |
uphem | +0.248 |
hern | +0.237 |
leys | +0.236 |
uum | +0.234 |
Neg logit contributions | |
Armstrong | -0.317 |
ppo | -0.295 |
pos | -0.293 |
Cot | -0.281 |
Chain | -0.275 |
Token | Go |
Token ID | 1514 |
Feature activation | +1.03e-03 |
Pos logit contributions | |
quartered | +4.259 |
uphem | +4.120 |
hern | +4.061 |
leys | +4.045 |
uum | +4.032 |
Neg logit contributions | |
Armstrong | -3.609 |
ppo | -3.502 |
pos | -3.406 |
闘 | -3.298 |
Cot | -3.189 |
Token | ce |
Token ID | 344 |
Feature activation | -6.80e-01 |
Pos logit contributions | |
Armstrong | +0.002 |
ppo | +0.002 |
pos | +0.002 |
Cot | +0.002 |
Recon | +0.002 |
Neg logit contributions | |
quartered | -0.002 |
uphem | -0.002 |
hern | -0.002 |
uum | -0.002 |
leys | -0.002 |
Token | uses |
Token ID | 3544 |
Feature activation | -1.01e-01 |
Pos logit contributions | |
quartered | +1.458 |
uphem | +1.409 |
hern | +1.380 |
leys | +1.365 |
uum | +1.361 |
Neg logit contributions | |
Armstrong | -1.281 |
ppo | -1.182 |
pos | -1.148 |
Cot | -1.121 |
闘 | -1.099 |
Token | an |
Token ID | 281 |
Feature activation | +2.03e-02 |
Pos logit contributions | |
quartered | +0.213 |
uphem | +0.206 |
hern | +0.202 |
leys | +0.199 |
uum | +0.198 |
Neg logit contributions | |
Armstrong | -0.196 |
ppo | -0.181 |
pos | -0.175 |
Cot | -0.172 |
闘 | -0.168 |
Token | electric |
Token ID | 5186 |
Feature activation | -2.71e-03 |
Pos logit contributions | |
Armstrong | +0.040 |
ppo | +0.037 |
pos | +0.036 |
Cot | +0.035 |
闘 | +0.034 |
Neg logit contributions | |
quartered | -0.042 |
uphem | -0.041 |
hern | -0.040 |
leys | -0.040 |
uum | -0.040 |
Token | unsupported |
Token ID | 24222 |
Feature activation | -4.95e-01 |
Pos logit contributions | |
Armstrong | +0.444 |
ppo | +0.419 |
pos | +0.413 |
Cot | +0.405 |
Recon | +0.398 |
Neg logit contributions | |
quartered | -0.191 |
uphem | -0.180 |
hern | -0.173 |
leys | -0.169 |
uum | -0.168 |
Token | on |
Token ID | 319 |
Feature activation | +6.20e-03 |
Pos logit contributions | |
quartered | +1.167 |
uphem | +1.129 |
hern | +1.099 |
leys | +1.087 |
uum | +1.086 |
Neg logit contributions | |
Armstrong | -0.852 |
ppo | -0.774 |
pos | -0.751 |
Cot | -0.726 |
闘 | -0.707 |
Token | your |
Token ID | 534 |
Feature activation | -3.04e-01 |
Pos logit contributions | |
Armstrong | +0.011 |
ppo | +0.010 |
pos | +0.010 |
Cot | +0.010 |
Recon | +0.010 |
Neg logit contributions | |
quartered | -0.014 |
uphem | -0.014 |
hern | -0.013 |
leys | -0.013 |
uum | -0.013 |
Token | device |
Token ID | 3335 |
Feature activation | +7.19e-02 |
Pos logit contributions | |
quartered | +0.701 |
uphem | +0.679 |
hern | +0.666 |
leys | +0.658 |
uum | +0.651 |
Neg logit contributions | |
Armstrong | -0.537 |
ppo | -0.490 |
pos | -0.481 |
Cot | -0.462 |
Recon | -0.447 |
Token | Media |
Token ID | 6343 |
Feature activation | -4.02e-01 |
Pos logit contributions | |
Armstrong | +0.169 |
ppo | +0.156 |
pos | +0.153 |
Cot | +0.151 |
Recon | +0.148 |
Neg logit contributions | |
quartered | -0.127 |
uphem | -0.120 |
hern | -0.116 |
leys | -0.114 |
uum | -0.113 |
Token | caption |
Token ID | 8305 |
Feature activation | -2.04e+00 |
Pos logit contributions | |
quartered | +0.711 |
uphem | +0.676 |
hern | +0.657 |
leys | +0.644 |
uum | +0.643 |
Neg logit contributions | |
Armstrong | -0.931 |
ppo | -0.868 |
pos | -0.857 |
Cot | -0.832 |
Recon | -0.813 |
Token | A |
Token ID | 317 |
Feature activation | -2.17e-01 |
Pos logit contributions | |
quartered | +4.088 |
uphem | +4.037 |
leys | +3.938 |
hern | +3.923 |
uum | +3.894 |
Neg logit contributions | |
Armstrong | -3.687 |
ppo | -3.559 |
pos | -3.463 |
闘 | -3.356 |
Cot | -3.246 |
Token | drone |
Token ID | 12170 |
Feature activation | -5.60e-02 |
Pos logit contributions | |
quartered | +0.428 |
uphem | +0.413 |
hern | +0.403 |
leys | +0.397 |
uum | +0.397 |
Neg logit contributions | |
Armstrong | -0.450 |
ppo | -0.418 |
pos | -0.407 |
Cot | -0.398 |
闘 | -0.389 |
Token | - |
Token ID | 12 |
Feature activation | -8.03e-03 |
Pos logit contributions | |
quartered | +0.122 |
uphem | +0.118 |
hern | +0.116 |
leys | +0.114 |
uum | +0.114 |
Neg logit contributions | |
Armstrong | -0.104 |
ppo | -0.096 |
pos | -0.093 |
Cot | -0.091 |
闘 | -0.088 |
Token | camera |
Token ID | 25695 |
Feature activation | -2.55e-02 |
Pos logit contributions | |
quartered | +0.016 |
uphem | +0.016 |
hern | +0.015 |
leys | +0.015 |
uum | +0.015 |
Neg logit contributions | |
Armstrong | -0.016 |
ppo | -0.015 |
pos | -0.015 |
Cot | -0.014 |
闘 | -0.014 |
Token | was |
Token ID | 373 |
Feature activation | +5.21e-02 |
Pos logit contributions | |
quartered | +0.053 |
uphem | +0.051 |
hern | +0.050 |
leys | +0.049 |
uum | +0.049 |
Neg logit contributions | |
Armstrong | -0.050 |
ppo | -0.046 |
pos | -0.045 |
Cot | -0.044 |
闘 | -0.043 |
Token | friend |
Token ID | 1545 |
Feature activation | +8.20e-02 |
Pos logit contributions | |
quartered | +0.390 |
uphem | +0.377 |
hern | +0.368 |
leys | +0.363 |
uum | +0.363 |
Neg logit contributions | |
Armstrong | -0.395 |
ppo | -0.366 |
pos | -0.356 |
Cot | -0.348 |
闘 | -0.340 |
Token | and |
Token ID | 290 |
Feature activation | -4.37e-01 |
Pos logit contributions | |
Armstrong | +0.155 |
ppo | +0.143 |
pos | +0.139 |
Cot | +0.135 |
闘 | +0.132 |
Neg logit contributions | |
quartered | -0.176 |
uphem | -0.171 |
hern | -0.167 |
leys | -0.165 |
uum | -0.165 |
Token | creative |
Token ID | 7325 |
Feature activation | -2.18e-01 |
Pos logit contributions | |
quartered | +0.878 |
uphem | +0.851 |
hern | +0.833 |
leys | +0.821 |
uum | +0.820 |
Neg logit contributions | |
Armstrong | -0.880 |
ppo | -0.824 |
pos | -0.791 |
Cot | -0.776 |
闘 | -0.767 |
Token | partner |
Token ID | 5212 |
Feature activation | -3.30e-01 |
Pos logit contributions | |
quartered | +0.433 |
uphem | +0.418 |
hern | +0.409 |
leys | +0.403 |
uum | +0.403 |
Neg logit contributions | |
Armstrong | -0.446 |
ppo | -0.415 |
pos | -0.403 |
Cot | -0.395 |
闘 | -0.386 |
Token | Alexandra |
Token ID | 44075 |
Feature activation | -1.41e+00 |
Pos logit contributions | |
quartered | +0.704 |
uphem | +0.682 |
hern | +0.668 |
leys | +0.660 |
uum | +0.660 |
Neg logit contributions | |
Armstrong | -0.624 |
ppo | -0.580 |
pos | -0.560 |
Cot | -0.545 |
闘 | -0.536 |
Token | Kot |
Token ID | 21702 |
Feature activation | -2.04e+00 |
Pos logit contributions | |
quartered | +2.927 |
uphem | +2.878 |
hern | +2.861 |
uum | +2.831 |
leys | +2.811 |
Neg logit contributions | |
Armstrong | -2.530 |
ppo | -2.422 |
pos | -2.322 |
闘 | -2.260 |
Cot | -2.198 |
Token | che |
Token ID | 2395 |
Feature activation | +2.47e-01 |
Pos logit contributions | |
quartered | +3.858 |
uphem | +3.741 |
hern | +3.700 |
uum | +3.628 |
leys | +3.608 |
Neg logit contributions | |
Armstrong | -4.253 |
ppo | -3.962 |
pos | -3.848 |
闘 | -3.772 |
Cot | -3.771 |
Token | ff |
Token ID | 487 |
Feature activation | +5.17e-01 |
Pos logit contributions | |
Armstrong | +0.494 |
ppo | +0.458 |
pos | +0.445 |
Cot | +0.435 |
闘 | +0.425 |
Neg logit contributions | |
quartered | -0.504 |
uphem | -0.488 |
hern | -0.477 |
leys | -0.470 |
uum | -0.470 |
Token | ( |
Token ID | 357 |
Feature activation | +1.96e-01 |
Pos logit contributions | |
Armstrong | +0.923 |
ppo | +0.843 |
pos | +0.819 |
Cot | +0.798 |
闘 | +0.773 |
Neg logit contributions | |
quartered | -1.167 |
uphem | -1.135 |
hern | -1.108 |
leys | -1.094 |
uum | -1.093 |
Token | the |
Token ID | 1169 |
Feature activation | -2.73e-01 |
Pos logit contributions | |
Armstrong | +0.387 |
ppo | +0.357 |
pos | +0.348 |
Cot | +0.340 |
Recon | +0.331 |
Neg logit contributions | |
quartered | -0.405 |
uphem | -0.393 |
hern | -0.382 |
leys | -0.377 |
uum | -0.377 |
Token | daughter |
Token ID | 4957 |
Feature activation | -2.66e-01 |
Pos logit contributions | |
quartered | +0.543 |
uphem | +0.525 |
hern | +0.513 |
leys | +0.506 |
uum | +0.506 |
Neg logit contributions | |
Armstrong | -0.556 |
ppo | -0.518 |
pos | -0.501 |
Cot | -0.491 |
闘 | -0.482 |
Token | the |
Token ID | 262 |
Feature activation | -7.96e-02 |
Pos logit contributions | |
quartered | +0.850 |
uphem | +0.824 |
hern | +0.806 |
uum | +0.801 |
leys | +0.794 |
Neg logit contributions | |
Armstrong | -0.714 |
ppo | -0.666 |
pos | -0.631 |
Cot | -0.624 |
闘 | -0.621 |
Token | way |
Token ID | 835 |
Feature activation | -2.68e-01 |
Pos logit contributions | |
quartered | +0.172 |
uphem | +0.167 |
hern | +0.163 |
leys | +0.161 |
uum | +0.161 |
Neg logit contributions | |
Armstrong | -0.149 |
ppo | -0.138 |
pos | -0.133 |
Cot | -0.130 |
闘 | -0.127 |
Token | I |
Token ID | 314 |
Feature activation | -4.07e-01 |
Pos logit contributions | |
quartered | +0.585 |
uphem | +0.567 |
hern | +0.556 |
uum | +0.549 |
leys | +0.548 |
Neg logit contributions | |
Armstrong | -0.494 |
ppo | -0.458 |
pos | -0.438 |
Cot | -0.431 |
闘 | -0.424 |
Token | view |
Token ID | 1570 |
Feature activation | -9.06e-01 |
Pos logit contributions | |
quartered | +0.836 |
uphem | +0.812 |
hern | +0.792 |
uum | +0.782 |
leys | +0.781 |
Neg logit contributions | |
Armstrong | -0.803 |
ppo | -0.744 |
pos | -0.717 |
Cot | -0.707 |
闘 | -0.694 |
Token | and |
Token ID | 290 |
Feature activation | -3.25e-01 |
Pos logit contributions | |
quartered | +1.959 |
uphem | +1.891 |
hern | +1.858 |
uum | +1.847 |
leys | +1.841 |
Neg logit contributions | |
Armstrong | -1.652 |
ppo | -1.554 |
pos | -1.460 |
闘 | -1.451 |
Cot | -1.446 |
Token | treat |
Token ID | 2190 |
Feature activation | -2.04e+00 |
Pos logit contributions | |
quartered | +0.632 |
uphem | +0.612 |
hern | +0.596 |
leys | +0.588 |
uum | +0.587 |
Neg logit contributions | |
Armstrong | -0.677 |
ppo | -0.631 |
pos | -0.608 |
Cot | -0.601 |
闘 | -0.590 |
Token | others |
Token ID | 1854 |
Feature activation | -3.97e-01 |
Pos logit contributions | |
quartered | +4.430 |
uphem | +4.291 |
uum | +4.223 |
hern | +4.198 |
leys | +4.167 |
Neg logit contributions | |
Armstrong | -3.447 |
ppo | -3.313 |
闘 | -3.194 |
Cot | -3.029 |
pos | -2.994 |
Token | , |
Token ID | 11 |
Feature activation | +1.56e-01 |
Pos logit contributions | |
quartered | +0.858 |
uphem | +0.831 |
hern | +0.815 |
uum | +0.805 |
leys | +0.801 |
Neg logit contributions | |
Armstrong | -0.739 |
ppo | -0.683 |
pos | -0.649 |
Cot | -0.644 |
闘 | -0.635 |
Token | and |
Token ID | 290 |
Feature activation | -1.83e-02 |
Pos logit contributions | |
Armstrong | +0.269 |
ppo | +0.247 |
pos | +0.237 |
Cot | +0.232 |
闘 | +0.226 |
Neg logit contributions | |
quartered | -0.361 |
uphem | -0.351 |
hern | -0.344 |
leys | -0.340 |
uum | -0.339 |
Token | how |
Token ID | 703 |
Feature activation | -4.29e-01 |
Pos logit contributions | |
quartered | +0.039 |
uphem | +0.038 |
hern | +0.037 |
leys | +0.037 |
uum | +0.036 |
Neg logit contributions | |
Armstrong | -0.035 |
ppo | -0.032 |
pos | -0.031 |
Cot | -0.030 |
闘 | -0.030 |
Token | they |
Token ID | 484 |
Feature activation | -2.01e-01 |
Pos logit contributions | |
quartered | +0.941 |
uphem | +0.913 |
hern | +0.893 |
leys | +0.883 |
uum | +0.882 |
Neg logit contributions | |
Armstrong | -0.784 |
ppo | -0.729 |
pos | -0.695 |
Cot | -0.683 |
闘 | -0.674 |
Token | <|endoftext|> |
Token ID | 50256 |
Feature activation | +1.55e-01 |
Pos logit contributions | |
Armstrong | +0.104 |
ppo | +0.095 |
pos | +0.092 |
Cot | +0.090 |
Recon | +0.087 |
Neg logit contributions | |
quartered | -0.132 |
uphem | -0.128 |
hern | -0.126 |
leys | -0.124 |
uum | -0.124 |
Token | Image |
Token ID | 5159 |
Feature activation | -4.00e-01 |
Pos logit contributions | |
Armstrong | +0.306 |
ppo | +0.284 |
pos | +0.276 |
Cot | +0.269 |
闘 | +0.263 |
Neg logit contributions | |
quartered | -0.321 |
uphem | -0.311 |
hern | -0.303 |
leys | -0.299 |
uum | -0.299 |
Token | copyright |
Token ID | 6634 |
Feature activation | -1.10e+00 |
Pos logit contributions | |
quartered | +0.612 |
uphem | +0.587 |
hern | +0.568 |
uum | +0.557 |
leys | +0.556 |
Neg logit contributions | |
Armstrong | -1.000 |
ppo | -0.944 |
pos | -0.920 |
Cot | -0.907 |
闘 | -0.894 |
Token | AP |
Token ID | 3486 |
Feature activation | -2.67e-01 |
Pos logit contributions | |
quartered | +2.548 |
uphem | +2.473 |
hern | +2.427 |
leys | +2.411 |
uum | +2.395 |
Neg logit contributions | |
Armstrong | -1.819 |
ppo | -1.706 |
pos | -1.616 |
闘 | -1.565 |
Cot | -1.558 |
Token | Image |
Token ID | 7412 |
Feature activation | -3.00e-01 |
Pos logit contributions | |
quartered | +0.589 |
uphem | +0.576 |
hern | +0.566 |
uum | +0.557 |
leys | +0.554 |
Neg logit contributions | |
Armstrong | -0.475 |
ppo | -0.441 |
pos | -0.427 |
Cot | -0.414 |
闘 | -0.411 |
Token | caption |
Token ID | 8305 |
Feature activation | -2.03e+00 |
Pos logit contributions | |
quartered | +0.559 |
uphem | +0.526 |
hern | +0.503 |
leys | +0.500 |
uum | +0.497 |
Neg logit contributions | |
Armstrong | -0.673 |
ppo | -0.628 |
pos | -0.622 |
Cot | -0.601 |
Recon | -0.585 |
Token | Peace |
Token ID | 12689 |
Feature activation | +1.53e-01 |
Pos logit contributions | |
quartered | +4.100 |
uphem | +4.033 |
leys | +3.927 |
hern | +3.905 |
uum | +3.888 |
Neg logit contributions | |
Armstrong | -3.671 |
ppo | -3.526 |
pos | -3.447 |
闘 | -3.349 |
Cot | -3.229 |
Token | talks |
Token ID | 6130 |
Feature activation | +2.79e-01 |
Pos logit contributions | |
Armstrong | +0.297 |
ppo | +0.274 |
pos | +0.266 |
Cot | +0.260 |
闘 | +0.254 |
Neg logit contributions | |
quartered | -0.323 |
uphem | -0.313 |
hern | -0.306 |
leys | -0.302 |
uum | -0.301 |
Token | between |
Token ID | 1022 |
Feature activation | -7.91e-02 |
Pos logit contributions | |
Armstrong | +0.540 |
ppo | +0.499 |
pos | +0.486 |
Cot | +0.472 |
Recon | +0.460 |
Neg logit contributions | |
quartered | -0.588 |
uphem | -0.571 |
hern | -0.558 |
leys | -0.551 |
uum | -0.550 |
Token | the |
Token ID | 262 |
Feature activation | -1.89e-02 |
Pos logit contributions | |
quartered | +0.162 |
uphem | +0.157 |
hern | +0.153 |
leys | +0.151 |
uum | +0.151 |
Neg logit contributions | |
Armstrong | -0.157 |
ppo | -0.145 |
pos | -0.141 |
Cot | -0.138 |
闘 | -0.135 |
Token | government |
Token ID | 1230 |
Feature activation | -3.16e-01 |
Pos logit contributions | |
quartered | +0.039 |
uphem | +0.038 |
hern | +0.037 |
leys | +0.036 |
uum | +0.036 |
Neg logit contributions | |
Armstrong | -0.037 |
ppo | -0.034 |
pos | -0.034 |
Cot | -0.033 |
闘 | -0.032 |
Token | could |
Token ID | 714 |
Feature activation | -2.52e-01 |
Pos logit contributions | |
quartered | +0.101 |
uphem | +0.098 |
hern | +0.095 |
leys | +0.094 |
uum | +0.094 |
Neg logit contributions | |
Armstrong | -0.108 |
ppo | -0.101 |
pos | -0.098 |
Cot | -0.096 |
闘 | -0.094 |
Token | not |
Token ID | 407 |
Feature activation | -3.10e-01 |
Pos logit contributions | |
quartered | +0.524 |
uphem | +0.509 |
hern | +0.497 |
leys | +0.490 |
uum | +0.490 |
Neg logit contributions | |
Armstrong | -0.490 |
ppo | -0.453 |
pos | -0.438 |
Cot | -0.430 |
闘 | -0.421 |
Token | have |
Token ID | 423 |
Feature activation | -3.79e-01 |
Pos logit contributions | |
quartered | +0.663 |
uphem | +0.646 |
hern | +0.633 |
leys | +0.623 |
uum | +0.621 |
Neg logit contributions | |
Armstrong | -0.581 |
ppo | -0.536 |
pos | -0.515 |
Cot | -0.507 |
闘 | -0.498 |
Token | anticipated |
Token ID | 14486 |
Feature activation | +4.06e-01 |
Pos logit contributions | |
quartered | +0.716 |
uphem | +0.694 |
hern | +0.677 |
leys | +0.665 |
uum | +0.664 |
Neg logit contributions | |
Armstrong | -0.810 |
ppo | -0.755 |
pos | -0.731 |
Cot | -0.719 |
闘 | -0.707 |
Token | the |
Token ID | 262 |
Feature activation | -1.34e-01 |
Pos logit contributions | |
Armstrong | +0.713 |
ppo | +0.648 |
pos | +0.633 |
Cot | +0.617 |
闘 | +0.594 |
Neg logit contributions | |
quartered | -0.931 |
uphem | -0.905 |
hern | -0.884 |
uum | -0.875 |
leys | -0.873 |
Token | reaction |
Token ID | 6317 |
Feature activation | -2.02e+00 |
Pos logit contributions | |
quartered | +0.272 |
uphem | +0.263 |
hern | +0.257 |
leys | +0.254 |
uum | +0.253 |
Neg logit contributions | |
Armstrong | -0.270 |
ppo | -0.250 |
pos | -0.243 |
Cot | -0.238 |
闘 | -0.232 |
Token | to |
Token ID | 284 |
Feature activation | -3.53e-01 |
Pos logit contributions | |
quartered | +4.044 |
uphem | +3.940 |
hern | +3.859 |
leys | +3.824 |
uum | +3.813 |
Neg logit contributions | |
Armstrong | -3.843 |
ppo | -3.713 |
pos | -3.431 |
闘 | -3.427 |
Cot | -3.393 |
Token | his |
Token ID | 465 |
Feature activation | -1.31e-01 |
Pos logit contributions | |
quartered | +0.789 |
uphem | +0.765 |
hern | +0.751 |
leys | +0.741 |
uum | +0.739 |
Neg logit contributions | |
Armstrong | -0.627 |
ppo | -0.582 |
pos | -0.558 |
Cot | -0.543 |
闘 | -0.535 |
Token | latest |
Token ID | 3452 |
Feature activation | +1.16e-01 |
Pos logit contributions | |
quartered | +0.269 |
uphem | +0.260 |
hern | +0.254 |
leys | +0.250 |
uum | +0.250 |
Neg logit contributions | |
Armstrong | -0.260 |
ppo | -0.241 |
pos | -0.234 |
Cot | -0.229 |
闘 | -0.224 |
Token | novel |
Token ID | 5337 |
Feature activation | -1.58e-01 |
Pos logit contributions | |
Armstrong | +0.248 |
ppo | +0.230 |
pos | +0.225 |
Cot | +0.221 |
闘 | +0.215 |
Neg logit contributions | |
quartered | -0.220 |
uphem | -0.212 |
hern | -0.206 |
leys | -0.203 |
uum | -0.203 |
Token | for |
Token ID | 329 |
Feature activation | -7.87e-02 |
Pos logit contributions | |
quartered | +0.352 |
uphem | +0.341 |
hern | +0.334 |
leys | +0.330 |
uum | +0.330 |
Neg logit contributions | |
Armstrong | -0.284 |
ppo | -0.262 |
pos | -0.252 |
Cot | -0.246 |
闘 | -0.240 |
Token | � |
Token ID | 247 |
Feature activation | -1.97e-01 |
Pos logit contributions | |
Armstrong | +0.404 |
ppo | +0.373 |
pos | +0.366 |
Cot | +0.353 |
闘 | +0.347 |
Neg logit contributions | |
quartered | -0.408 |
uphem | -0.391 |
hern | -0.385 |
leys | -0.377 |
uum | -0.371 |
Token | t |
Token ID | 83 |
Feature activation | +6.89e-02 |
Pos logit contributions | |
quartered | +0.443 |
uphem | +0.428 |
hern | +0.417 |
leys | +0.412 |
uum | +0.410 |
Neg logit contributions | |
Armstrong | -0.352 |
ppo | -0.325 |
pos | -0.316 |
Cot | -0.306 |
闘 | -0.296 |
Token | know |
Token ID | 760 |
Feature activation | -8.79e-02 |
Pos logit contributions | |
Armstrong | +0.133 |
ppo | +0.123 |
pos | +0.120 |
Cot | +0.117 |
Recon | +0.114 |
Neg logit contributions | |
quartered | -0.146 |
uphem | -0.140 |
hern | -0.137 |
uum | -0.136 |
leys | -0.135 |
Token | how |
Token ID | 703 |
Feature activation | -5.02e-01 |
Pos logit contributions | |
quartered | +0.189 |
uphem | +0.183 |
hern | +0.179 |
uum | +0.177 |
leys | +0.177 |
Neg logit contributions | |
Armstrong | -0.167 |
ppo | -0.153 |
pos | -0.149 |
Cot | -0.146 |
Recon | -0.141 |
Token | to |
Token ID | 284 |
Feature activation | -4.59e-01 |
Pos logit contributions | |
quartered | +0.984 |
uphem | +0.965 |
hern | +0.936 |
leys | +0.919 |
uum | +0.909 |
Neg logit contributions | |
Armstrong | -1.023 |
ppo | -0.970 |
pos | -0.924 |
Cot | -0.906 |
闘 | -0.896 |
Token | behave |
Token ID | 17438 |
Feature activation | -2.02e+00 |
Pos logit contributions | |
quartered | +0.945 |
uphem | +0.916 |
hern | +0.894 |
leys | +0.881 |
uum | +0.881 |
Neg logit contributions | |
Armstrong | -0.908 |
ppo | -0.841 |
pos | -0.816 |
Cot | -0.799 |
闘 | -0.780 |
Token | in |
Token ID | 287 |
Feature activation | -1.08e-01 |
Pos logit contributions | |
quartered | +4.225 |
uphem | +4.105 |
hern | +4.054 |
leys | +3.959 |
uum | +3.949 |
Neg logit contributions | |
Armstrong | -3.762 |
ppo | -3.501 |
pos | -3.304 |
Cot | -3.293 |
闘 | -3.262 |
Token | the |
Token ID | 262 |
Feature activation | -4.51e-02 |
Pos logit contributions | |
quartered | +0.216 |
uphem | +0.209 |
hern | +0.204 |
leys | +0.201 |
uum | +0.201 |
Neg logit contributions | |
Armstrong | -0.221 |
ppo | -0.205 |
pos | -0.200 |
Cot | -0.195 |
闘 | -0.190 |
Token | 21 |
Token ID | 2310 |
Feature activation | +5.51e-01 |
Pos logit contributions | |
quartered | +0.088 |
uphem | +0.085 |
hern | +0.082 |
uum | +0.081 |
leys | +0.081 |
Neg logit contributions | |
Armstrong | -0.095 |
ppo | -0.088 |
pos | -0.086 |
Cot | -0.084 |
Recon | -0.082 |
Token | st |
Token ID | 301 |
Feature activation | -1.38e-01 |
Pos logit contributions | |
Armstrong | +1.028 |
ppo | +0.946 |
pos | +0.919 |
Cot | +0.896 |
闘 | +0.873 |
Neg logit contributions | |
quartered | -1.198 |
uphem | -1.161 |
hern | -1.136 |
leys | -1.120 |
uum | -1.120 |
Token | century |
Token ID | 4289 |
Feature activation | +1.37e-01 |
Pos logit contributions | |
quartered | +0.234 |
uphem | +0.225 |
hern | +0.222 |
leys | +0.218 |
uum | +0.215 |
Neg logit contributions | |
Armstrong | -0.321 |
ppo | -0.302 |
pos | -0.291 |
Cot | -0.286 |
闘 | -0.282 |
Token | � |
Token ID | 447 |
Feature activation | +4.66e-02 |
Pos logit contributions | |
Armstrong | +0.336 |
ppo | +0.309 |
pos | +0.299 |
Cot | +0.293 |
闘 | +0.284 |
Neg logit contributions | |
quartered | -0.401 |
uphem | -0.388 |
hern | -0.380 |
leys | -0.375 |
uum | -0.375 |
Token | � |
Token ID | 251 |
Feature activation | +2.83e-01 |
Pos logit contributions | |
Armstrong | +0.108 |
ppo | +0.101 |
pos | +0.098 |
Cot | +0.097 |
闘 | +0.095 |
Neg logit contributions | |
quartered | -0.080 |
uphem | -0.077 |
hern | -0.075 |
leys | -0.073 |
uum | -0.073 |
Token |
|
Token ID | 198 |
Feature activation | +3.01e-01 |
Pos logit contributions | |
Armstrong | +0.512 |
ppo | +0.469 |
pos | +0.454 |
Cot | +0.446 |
Recon | +0.432 |
Neg logit contributions | |
quartered | -0.634 |
uphem | -0.613 |
hern | -0.601 |
uum | -0.592 |
leys | -0.592 |
Token |
|
Token ID | 198 |
Feature activation | +3.25e-01 |
Pos logit contributions | |
Armstrong | +0.480 |
ppo | +0.431 |
pos | +0.424 |
Cot | +0.401 |
Recon | +0.394 |
Neg logit contributions | |
quartered | -0.779 |
uphem | -0.734 |
hern | -0.711 |
leys | -0.705 |
uum | -0.704 |
Token | In |
Token ID | 818 |
Feature activation | +4.76e-01 |
Pos logit contributions | |
Armstrong | +0.606 |
ppo | +0.560 |
pos | +0.542 |
Cot | +0.530 |
Recon | +0.515 |
Neg logit contributions | |
quartered | -0.710 |
uphem | -0.682 |
hern | -0.667 |
leys | -0.658 |
uum | -0.657 |
Token | reaction |
Token ID | 6317 |
Feature activation | -2.00e+00 |
Pos logit contributions | |
Armstrong | +0.845 |
ppo | +0.775 |
pos | +0.752 |
Cot | +0.733 |
Recon | +0.709 |
Neg logit contributions | |
quartered | -1.077 |
uphem | -1.046 |
hern | -1.024 |
leys | -1.012 |
uum | -1.009 |
Token | , |
Token ID | 11 |
Feature activation | -4.77e-01 |
Pos logit contributions | |
quartered | +3.743 |
uphem | +3.625 |
leys | +3.489 |
hern | +3.477 |
uum | +3.468 |
Neg logit contributions | |
Armstrong | -4.015 |
ppo | -3.847 |
pos | -3.657 |
闘 | -3.646 |
Cot | -3.527 |
Token | Josh |
Token ID | 8518 |
Feature activation | -1.99e-02 |
Pos logit contributions | |
quartered | +1.058 |
uphem | +1.026 |
hern | +1.004 |
leys | +0.993 |
uum | +0.991 |
Neg logit contributions | |
Armstrong | -0.863 |
ppo | -0.796 |
pos | -0.768 |
Cot | -0.748 |
闘 | -0.735 |
Token | , |
Token ID | 11 |
Feature activation | -2.22e-01 |
Pos logit contributions | |
quartered | +0.042 |
uphem | +0.041 |
hern | +0.040 |
leys | +0.039 |
uum | +0.039 |
Neg logit contributions | |
Armstrong | -0.038 |
ppo | -0.036 |
pos | -0.035 |
Cot | -0.034 |
闘 | -0.033 |
Token | Michelle |
Token ID | 16738 |
Feature activation | +8.13e-01 |
Pos logit contributions | |
quartered | +0.487 |
uphem | +0.472 |
hern | +0.462 |
uum | +0.456 |
leys | +0.456 |
Neg logit contributions | |
Armstrong | -0.409 |
ppo | -0.375 |
pos | -0.365 |
Cot | -0.357 |
Recon | -0.346 |
Token | , |
Token ID | 11 |
Feature activation | -1.58e-01 |
Pos logit contributions | |
Armstrong | +1.471 |
ppo | +1.354 |
pos | +1.309 |
Cot | +1.277 |
闘 | +1.245 |
Neg logit contributions | |
quartered | -1.810 |
uphem | -1.758 |
hern | -1.720 |
uum | -1.699 |
leys | -1.699 |
Token | a |
Token ID | 257 |
Feature activation | -1.98e-01 |
Pos logit contributions | |
quartered | +0.873 |
uphem | +0.846 |
hern | +0.827 |
leys | +0.816 |
uum | +0.816 |
Neg logit contributions | |
Armstrong | -0.766 |
ppo | -0.707 |
pos | -0.686 |
Cot | -0.670 |
闘 | -0.652 |
Token | different |
Token ID | 1180 |
Feature activation | -2.62e-01 |
Pos logit contributions | |
quartered | +0.402 |
uphem | +0.388 |
hern | +0.378 |
uum | +0.373 |
leys | +0.373 |
Neg logit contributions | |
Armstrong | -0.398 |
ppo | -0.368 |
pos | -0.361 |
Cot | -0.352 |
Recon | -0.342 |
Token | region |
Token ID | 3814 |
Feature activation | -3.20e-02 |
Pos logit contributions | |
quartered | +0.489 |
uphem | +0.472 |
hern | +0.459 |
leys | +0.451 |
uum | +0.451 |
Neg logit contributions | |
Armstrong | -0.569 |
ppo | -0.530 |
pos | -0.518 |
Cot | -0.507 |
闘 | -0.494 |
Token | . |
Token ID | 13 |
Feature activation | +8.92e-02 |
Pos logit contributions | |
quartered | +0.072 |
uphem | +0.070 |
hern | +0.068 |
leys | +0.067 |
uum | +0.067 |
Neg logit contributions | |
Armstrong | -0.057 |
ppo | -0.053 |
pos | -0.051 |
Cot | -0.050 |
闘 | -0.048 |
Token | Global |
Token ID | 8060 |
Feature activation | -9.61e-02 |
Pos logit contributions | |
Armstrong | +0.163 |
ppo | +0.149 |
pos | +0.145 |
Cot | +0.142 |
闘 | +0.137 |
Neg logit contributions | |
quartered | -0.199 |
uphem | -0.192 |
hern | -0.188 |
leys | -0.186 |
uum | -0.185 |
Token | pand |
Token ID | 19798 |
Feature activation | -1.99e+00 |
Pos logit contributions | |
quartered | +0.173 |
uphem | +0.167 |
hern | +0.162 |
leys | +0.159 |
uum | +0.159 |
Neg logit contributions | |
Armstrong | -0.216 |
ppo | -0.201 |
pos | -0.197 |
Cot | -0.193 |
闘 | -0.188 |
Token | emic |
Token ID | 5314 |
Feature activation | -6.61e-01 |
Pos logit contributions | |
quartered | +1.277 |
uphem | +1.186 |
uum | +1.122 |
hern | +1.115 |
leys | +1.071 |
Neg logit contributions | |
Armstrong | -6.666 |
ppo | -6.340 |
pos | -6.231 |
Cot | -6.187 |
闘 | -6.112 |
Token | under |
Token ID | 739 |
Feature activation | -2.52e-01 |
Pos logit contributions | |
quartered | +1.414 |
uphem | +1.370 |
hern | +1.341 |
uum | +1.323 |
leys | +1.322 |
Neg logit contributions | |
Armstrong | -1.254 |
ppo | -1.158 |
pos | -1.121 |
Cot | -1.097 |
闘 | -1.070 |
Token | way |
Token ID | 835 |
Feature activation | -1.99e-01 |
Pos logit contributions | |
quartered | +0.524 |
uphem | +0.508 |
hern | +0.495 |
leys | +0.489 |
uum | +0.487 |
Neg logit contributions | |
Armstrong | -0.496 |
ppo | -0.456 |
pos | -0.447 |
Cot | -0.434 |
Recon | -0.424 |
Token | . |
Token ID | 13 |
Feature activation | +6.66e-02 |
Pos logit contributions | |
quartered | +0.443 |
uphem | +0.429 |
hern | +0.420 |
leys | +0.414 |
uum | +0.414 |
Neg logit contributions | |
Armstrong | -0.363 |
ppo | -0.333 |
pos | -0.323 |
Cot | -0.316 |
闘 | -0.306 |
Token | Post |
Token ID | 2947 |
Feature activation | -4.34e-01 |
Pos logit contributions | |
Armstrong | +0.123 |
ppo | +0.113 |
pos | +0.110 |
Cot | +0.108 |
Recon | +0.104 |
Neg logit contributions | |
quartered | -0.147 |
uphem | -0.142 |
hern | -0.139 |
leys | -0.137 |
uum | -0.137 |
Token | unsupported |
Token ID | 24222 |
Feature activation | -3.43e-01 |
Pos logit contributions | |
Armstrong | +0.645 |
ppo | +0.609 |
pos | +0.599 |
Cot | +0.588 |
Recon | +0.577 |
Neg logit contributions | |
quartered | -0.283 |
uphem | -0.267 |
hern | -0.257 |
leys | -0.251 |
uum | -0.250 |
Token | on |
Token ID | 319 |
Feature activation | +4.32e-02 |
Pos logit contributions | |
quartered | +0.818 |
uphem | +0.789 |
hern | +0.767 |
leys | +0.758 |
uum | +0.758 |
Neg logit contributions | |
Armstrong | -0.584 |
ppo | -0.530 |
pos | -0.514 |
Cot | -0.496 |
闘 | -0.484 |
Token | your |
Token ID | 534 |
Feature activation | -2.38e-01 |
Pos logit contributions | |
Armstrong | +0.078 |
ppo | +0.072 |
pos | +0.070 |
Cot | +0.068 |
Recon | +0.066 |
Neg logit contributions | |
quartered | -0.097 |
uphem | -0.095 |
hern | -0.090 |
uum | -0.090 |
leys | -0.090 |
Token | device |
Token ID | 3335 |
Feature activation | -5.62e-02 |
Pos logit contributions | |
quartered | +0.548 |
uphem | +0.531 |
hern | +0.521 |
leys | +0.516 |
uum | +0.510 |
Neg logit contributions | |
Armstrong | -0.418 |
ppo | -0.382 |
pos | -0.375 |
Cot | -0.360 |
Recon | -0.349 |
Token | Media |
Token ID | 6343 |
Feature activation | -3.42e-01 |
Pos logit contributions | |
quartered | +0.099 |
uphem | +0.094 |
hern | +0.091 |
leys | +0.089 |
uum | +0.089 |
Neg logit contributions | |
Armstrong | -0.132 |
ppo | -0.122 |
pos | -0.120 |
Cot | -0.118 |
Recon | -0.116 |
Token | caption |
Token ID | 8305 |
Feature activation | -1.99e+00 |
Pos logit contributions | |
quartered | +0.604 |
uphem | +0.575 |
hern | +0.558 |
leys | +0.547 |
uum | +0.547 |
Neg logit contributions | |
Armstrong | -0.795 |
ppo | -0.740 |
pos | -0.730 |
Cot | -0.710 |
Recon | -0.694 |
Token | Nigel |
Token ID | 28772 |
Feature activation | +2.71e-01 |
Pos logit contributions | |
quartered | +3.990 |
uphem | +3.936 |
leys | +3.836 |
hern | +3.823 |
uum | +3.792 |
Neg logit contributions | |
Armstrong | -3.594 |
ppo | -3.463 |
pos | -3.370 |
闘 | -3.258 |
Cot | -3.166 |
Token | Farage |
Token ID | 33904 |
Feature activation | +1.27e-01 |
Pos logit contributions | |
Armstrong | +0.533 |
ppo | +0.494 |
pos | +0.480 |
Cot | +0.469 |
闘 | +0.458 |
Neg logit contributions | |
quartered | -0.559 |
uphem | -0.541 |
hern | -0.528 |
leys | -0.521 |
uum | -0.521 |
Token | : |
Token ID | 25 |
Feature activation | -1.54e-01 |
Pos logit contributions | |
Armstrong | +0.239 |
ppo | +0.218 |
pos | +0.215 |
Cot | +0.208 |
Recon | +0.201 |
Neg logit contributions | |
quartered | -0.275 |
uphem | -0.266 |
hern | -0.260 |
uum | -0.256 |
leys | -0.256 |
Token | " |
Token ID | 366 |
Feature activation | -3.22e-02 |
Pos logit contributions | |
quartered | +0.344 |
uphem | +0.335 |
hern | +0.327 |
leys | +0.324 |
uum | +0.323 |
Neg logit contributions | |
Armstrong | -0.273 |
ppo | -0.254 |
pos | -0.244 |
Cot | -0.238 |
闘 | -0.234 |
Token | Lord |
Token ID | 22438 |
Feature activation | +3.74e-01 |
Pos logit contributions | |
quartered | +0.069 |
uphem | +0.066 |
hern | +0.065 |
leys | +0.064 |
uum | +0.064 |
Neg logit contributions | |
Armstrong | -0.062 |
ppo | -0.057 |
pos | -0.055 |
Cot | -0.054 |
闘 | -0.053 |
Token | outbreak |
Token ID | 17645 |
Feature activation | -6.12e-01 |
Pos logit contributions | |
quartered | +0.387 |
uphem | +0.372 |
hern | +0.361 |
leys | +0.355 |
uum | +0.355 |
Neg logit contributions | |
Armstrong | -0.533 |
ppo | -0.500 |
pos | -0.488 |
Cot | -0.479 |
闘 | -0.469 |
Token | a |
Token ID | 257 |
Feature activation | -1.07e-01 |
Pos logit contributions | |
quartered | +1.362 |
uphem | +1.323 |
hern | +1.296 |
uum | +1.278 |
leys | +1.277 |
Neg logit contributions | |
Armstrong | -1.105 |
ppo | -1.017 |
pos | -0.982 |
Cot | -0.959 |
闘 | -0.936 |
Token | full |
Token ID | 1336 |
Feature activation | -7.52e-01 |
Pos logit contributions | |
quartered | +0.218 |
uphem | +0.209 |
hern | +0.203 |
leys | +0.202 |
uum | +0.202 |
Neg logit contributions | |
Armstrong | -0.217 |
ppo | -0.200 |
pos | -0.197 |
Cot | -0.192 |
Recon | -0.186 |
Token | - |
Token ID | 12 |
Feature activation | +1.53e-01 |
Pos logit contributions | |
quartered | +1.630 |
uphem | +1.585 |
hern | +1.572 |
uum | +1.527 |
leys | +1.524 |
Neg logit contributions | |
Armstrong | -1.383 |
ppo | -1.291 |
pos | -1.218 |
Cot | -1.204 |
闘 | -1.187 |
Token | blown |
Token ID | 31290 |
Feature activation | +1.11e-01 |
Pos logit contributions | |
Armstrong | +0.330 |
ppo | +0.308 |
pos | +0.299 |
Cot | +0.294 |
闘 | +0.289 |
Neg logit contributions | |
quartered | -0.285 |
uphem | -0.276 |
hern | -0.271 |
leys | -0.266 |
uum | -0.265 |
Token | pand |
Token ID | 19798 |
Feature activation | -1.97e+00 |
Pos logit contributions | |
Armstrong | +0.274 |
ppo | +0.257 |
pos | +0.253 |
Cot | +0.248 |
Recon | +0.242 |
Neg logit contributions | |
quartered | -0.176 |
uphem | -0.168 |
hern | -0.162 |
leys | -0.160 |
uum | -0.160 |
Token | emic |
Token ID | 5314 |
Feature activation | -7.42e-01 |
Pos logit contributions | |
quartered | +1.275 |
uphem | +1.208 |
hern | +1.149 |
uum | +1.120 |
leys | +1.072 |
Neg logit contributions | |
Armstrong | -6.590 |
ppo | -6.282 |
pos | -6.176 |
Cot | -6.118 |
闘 | -6.053 |
Token | . |
Token ID | 13 |
Feature activation | -2.53e-01 |
Pos logit contributions | |
quartered | +1.619 |
uphem | +1.577 |
hern | +1.547 |
uum | +1.525 |
leys | +1.515 |
Neg logit contributions | |
Armstrong | -1.363 |
ppo | -1.259 |
pos | -1.207 |
Cot | -1.187 |
闘 | -1.164 |
Token | SW |
Token ID | 12672 |
Feature activation | +1.52e-01 |
Pos logit contributions | |
quartered | +0.557 |
uphem | +0.543 |
hern | +0.531 |
leys | +0.524 |
uum | +0.524 |
Neg logit contributions | |
Armstrong | -0.457 |
ppo | -0.423 |
pos | -0.409 |
Cot | -0.397 |
闘 | -0.389 |
Token | INE |
Token ID | 8881 |
Feature activation | +1.73e-01 |
Pos logit contributions | |
Armstrong | +0.295 |
ppo | +0.272 |
pos | +0.265 |
Cot | +0.259 |
闘 | +0.251 |
Neg logit contributions | |
quartered | -0.318 |
uphem | -0.308 |
hern | -0.301 |
uum | -0.297 |
leys | -0.297 |
Token | FL |
Token ID | 9977 |
Feature activation | -3.91e-02 |
Pos logit contributions | |
Armstrong | +0.330 |
ppo | +0.305 |
pos | +0.298 |
Cot | +0.289 |
Recon | +0.281 |
Neg logit contributions | |
quartered | -0.371 |
uphem | -0.359 |
hern | -0.351 |
leys | -0.347 |
uum | -0.347 |
Token |
|
Token ID | 198 |
Feature activation | +6.04e-02 |
Pos logit contributions | |
quartered | +0.134 |
uphem | +0.128 |
hern | +0.124 |
uum | +0.123 |
leys | +0.123 |
Neg logit contributions | |
Armstrong | -0.090 |
ppo | -0.082 |
pos | -0.080 |
Cot | -0.077 |
Recon | -0.075 |
Token | Image |
Token ID | 5159 |
Feature activation | -9.12e-01 |
Pos logit contributions | |
Armstrong | +0.119 |
ppo | +0.110 |
pos | +0.107 |
Cot | +0.105 |
闘 | +0.102 |
Neg logit contributions | |
quartered | -0.125 |
uphem | -0.121 |
hern | -0.118 |
leys | -0.116 |
uum | -0.116 |
Token | copyright |
Token ID | 6634 |
Feature activation | -9.88e-01 |
Pos logit contributions | |
quartered | +1.260 |
uphem | +1.213 |
hern | +1.159 |
leys | +1.134 |
uum | +1.130 |
Neg logit contributions | |
Armstrong | -2.374 |
ppo | -2.273 |
pos | -2.199 |
闘 | -2.170 |
Cot | -2.165 |
Token | AFP |
Token ID | 8916 |
Feature activation | +6.28e-01 |
Pos logit contributions | |
quartered | +2.378 |
uphem | +2.322 |
hern | +2.279 |
leys | +2.260 |
uum | +2.245 |
Neg logit contributions | |
Armstrong | -1.535 |
ppo | -1.414 |
pos | -1.346 |
闘 | -1.298 |
Cot | -1.296 |
Token | Image |
Token ID | 7412 |
Feature activation | -8.94e-02 |
Pos logit contributions | |
Armstrong | +1.092 |
ppo | +1.000 |
pos | +0.969 |
Cot | +0.947 |
闘 | +0.926 |
Neg logit contributions | |
quartered | -1.435 |
uphem | -1.395 |
hern | -1.371 |
uum | -1.352 |
leys | -1.351 |
Token | caption |
Token ID | 8305 |
Feature activation | -1.96e+00 |
Pos logit contributions | |
quartered | +0.168 |
uphem | +0.158 |
hern | +0.151 |
leys | +0.150 |
uum | +0.149 |
Neg logit contributions | |
Armstrong | -0.199 |
ppo | -0.187 |
pos | -0.184 |
Cot | -0.178 |
Recon | -0.174 |
Token | The |
Token ID | 383 |
Feature activation | -1.64e-01 |
Pos logit contributions | |
quartered | +3.963 |
uphem | +3.905 |
leys | +3.812 |
hern | +3.804 |
uum | +3.762 |
Neg logit contributions | |
Armstrong | -3.587 |
ppo | -3.348 |
pos | -3.339 |
闘 | -3.217 |
Cot | -3.122 |
Token | Kurds |
Token ID | 23880 |
Feature activation | +2.26e-01 |
Pos logit contributions | |
quartered | +0.367 |
uphem | +0.357 |
hern | +0.349 |
uum | +0.344 |
leys | +0.344 |
Neg logit contributions | |
Armstrong | -0.296 |
ppo | -0.272 |
pos | -0.264 |
Cot | -0.257 |
Recon | -0.249 |
Token | ' |
Token ID | 6 |
Feature activation | +7.35e-01 |
Pos logit contributions | |
Armstrong | +0.440 |
ppo | +0.408 |
pos | +0.397 |
Cot | +0.388 |
Recon | +0.376 |
Neg logit contributions | |
quartered | -0.473 |
uphem | -0.460 |
hern | -0.449 |
uum | -0.442 |
leys | -0.441 |
Token | push |
Token ID | 4574 |
Feature activation | -7.71e-02 |
Pos logit contributions | |
Armstrong | +1.463 |
ppo | +1.354 |
pos | +1.330 |
Cot | +1.301 |
Recon | +1.265 |
Neg logit contributions | |
quartered | -1.517 |
uphem | -1.476 |
hern | -1.427 |
uum | -1.416 |
leys | -1.405 |
Token | has |
Token ID | 468 |
Feature activation | +1.23e-01 |
Pos logit contributions | |
quartered | +0.168 |
uphem | +0.163 |
hern | +0.160 |
leys | +0.158 |
uum | +0.157 |
Neg logit contributions | |
Armstrong | -0.143 |
ppo | -0.132 |
pos | -0.128 |
Cot | -0.125 |
闘 | -0.122 |
Token | a |
Token ID | 257 |
Feature activation | +1.48e-01 |
Pos logit contributions | |
Armstrong | +0.809 |
ppo | +0.740 |
pos | +0.724 |
Cot | +0.703 |
Recon | +0.680 |
Neg logit contributions | |
quartered | -0.952 |
uphem | -0.921 |
hern | -0.898 |
uum | -0.891 |
leys | -0.887 |
Token | blow |
Token ID | 6611 |
Feature activation | -3.53e-01 |
Pos logit contributions | |
Armstrong | +0.312 |
ppo | +0.289 |
pos | +0.284 |
Cot | +0.277 |
Recon | +0.270 |
Neg logit contributions | |
quartered | -0.288 |
uphem | -0.278 |
hern | -0.270 |
uum | -0.267 |
leys | -0.267 |
Token | to |
Token ID | 284 |
Feature activation | -9.21e-02 |
Pos logit contributions | |
quartered | +0.732 |
uphem | +0.709 |
hern | +0.691 |
leys | +0.680 |
uum | +0.679 |
Neg logit contributions | |
Armstrong | -0.693 |
ppo | -0.643 |
pos | -0.624 |
Cot | -0.611 |
闘 | -0.598 |
Token | the |
Token ID | 262 |
Feature activation | -8.87e-02 |
Pos logit contributions | |
quartered | +0.202 |
uphem | +0.197 |
hern | +0.193 |
leys | +0.190 |
uum | +0.190 |
Neg logit contributions | |
Armstrong | -0.167 |
ppo | -0.154 |
pos | -0.149 |
Cot | -0.145 |
闘 | -0.143 |
Token | |
Token ID | 22020 |
Feature activation | +4.45e-01 |
Pos logit contributions | |
quartered | +0.182 |
uphem | +0.176 |
hern | +0.171 |
uum | +0.169 |
leys | +0.169 |
Neg logit contributions | |
Armstrong | -0.177 |
ppo | -0.164 |
pos | -0.160 |
Cot | -0.156 |
闘 | -0.152 |
Token | administ |
Token ID | 39081 |
Feature activation | -1.96e+00 |
Pos logit contributions | |
Armstrong | +0.896 |
ppo | +0.834 |
pos | +0.812 |
Cot | +0.789 |
Recon | +0.769 |
Neg logit contributions | |
quartered | -0.897 |
uphem | -0.872 |
hern | -0.843 |
leys | -0.836 |
uum | -0.834 |
Token | ration |
Token ID | 1358 |
Feature activation | -6.30e-01 |
Pos logit contributions | |
quartered | +2.494 |
uphem | +2.395 |
hern | +2.323 |
uum | +2.323 |
leys | +2.292 |
Neg logit contributions | |
Armstrong | -5.245 |
ppo | -4.976 |
闘 | -4.792 |
Cot | -4.791 |
pos | -4.781 |
Token | � |
Token ID | 447 |
Feature activation | -3.81e-01 |
Pos logit contributions | |
quartered | +1.381 |
uphem | +1.344 |
hern | +1.316 |
leys | +1.294 |
uum | +1.293 |
Neg logit contributions | |
Armstrong | -1.147 |
ppo | -1.066 |
pos | -1.018 |
Cot | -1.001 |
闘 | -0.981 |
Token | � |
Token ID | 247 |
Feature activation | -2.45e-01 |
Pos logit contributions | |
quartered | +0.789 |
uphem | +0.756 |
hern | +0.737 |
leys | +0.727 |
uum | +0.718 |
Neg logit contributions | |
Armstrong | -0.761 |
ppo | -0.702 |
pos | -0.689 |
Cot | -0.662 |
闘 | -0.654 |
Token | s |
Token ID | 82 |
Feature activation | -1.58e-01 |
Pos logit contributions | |
quartered | +0.503 |
uphem | +0.487 |
hern | +0.475 |
leys | +0.468 |
uum | +0.468 |
Neg logit contributions | |
Armstrong | -0.486 |
ppo | -0.450 |
pos | -0.438 |
Cot | -0.427 |
闘 | -0.417 |
Token | credibility |
Token ID | 16532 |
Feature activation | -8.28e-02 |
Pos logit contributions | |
quartered | +0.324 |
uphem | +0.314 |
hern | +0.306 |
leys | +0.302 |
uum | +0.302 |
Neg logit contributions | |
Armstrong | -0.314 |
ppo | -0.290 |
pos | -0.282 |
Cot | -0.276 |
闘 | -0.269 |
Token | media |
Token ID | 2056 |
Feature activation | -2.38e-02 |
Pos logit contributions | |
Armstrong | +1.831 |
ppo | +1.694 |
pos | +1.682 |
Cot | +1.626 |
Recon | +1.603 |
Neg logit contributions | |
quartered | -1.556 |
uphem | -1.503 |
hern | -1.479 |
uum | -1.458 |
leys | -1.447 |
Token | reported |
Token ID | 2098 |
Feature activation | -3.37e-01 |
Pos logit contributions | |
quartered | +0.044 |
uphem | +0.042 |
hern | +0.041 |
uum | +0.041 |
leys | +0.040 |
Neg logit contributions | |
Armstrong | -0.053 |
ppo | -0.049 |
pos | -0.048 |
Cot | -0.047 |
Recon | -0.046 |
Token | . |
Token ID | 13 |
Feature activation | -2.05e-02 |
Pos logit contributions | |
quartered | +0.734 |
uphem | +0.710 |
hern | +0.696 |
leys | +0.687 |
uum | +0.686 |
Neg logit contributions | |
Armstrong | -0.621 |
ppo | -0.579 |
pos | -0.555 |
Cot | -0.544 |
闘 | -0.533 |
Token | <|endoftext|> |
Token ID | 50256 |
Feature activation | +3.07e-01 |
Pos logit contributions | |
quartered | +0.045 |
uphem | +0.043 |
hern | +0.042 |
leys | +0.042 |
uum | +0.042 |
Neg logit contributions | |
Armstrong | -0.038 |
ppo | -0.035 |
pos | -0.034 |
Cot | -0.033 |
闘 | -0.032 |
Token | Image |
Token ID | 5159 |
Feature activation | -4.41e-01 |
Pos logit contributions | |
Armstrong | +0.607 |
ppo | +0.563 |
pos | +0.546 |
Cot | +0.533 |
闘 | +0.520 |
Neg logit contributions | |
quartered | -0.634 |
uphem | -0.612 |
hern | -0.598 |
leys | -0.590 |
uum | -0.590 |
Token | caption |
Token ID | 8305 |
Feature activation | -1.96e+00 |
Pos logit contributions | |
quartered | +0.678 |
uphem | +0.650 |
hern | +0.629 |
uum | +0.617 |
leys | +0.617 |
Neg logit contributions | |
Armstrong | -1.098 |
ppo | -1.037 |
pos | -1.010 |
Cot | -0.995 |
闘 | -0.982 |
Token | Some |
Token ID | 2773 |
Feature activation | -8.17e-01 |
Pos logit contributions | |
quartered | +4.054 |
uphem | +3.953 |
leys | +3.855 |
hern | +3.844 |
uum | +3.803 |
Neg logit contributions | |
Armstrong | -3.513 |
ppo | -3.366 |
pos | -3.249 |
闘 | -3.169 |
Cot | -3.081 |
Token | talked |
Token ID | 6619 |
Feature activation | +8.18e-01 |
Pos logit contributions | |
quartered | +1.677 |
uphem | +1.616 |
hern | +1.568 |
leys | +1.563 |
uum | +1.553 |
Neg logit contributions | |
Armstrong | -1.606 |
ppo | -1.493 |
pos | -1.436 |
Cot | -1.413 |
闘 | -1.398 |
Token | of |
Token ID | 286 |
Feature activation | +9.11e-02 |
Pos logit contributions | |
Armstrong | +1.497 |
ppo | +1.365 |
pos | +1.338 |
Cot | +1.298 |
Recon | +1.255 |
Neg logit contributions | |
quartered | -1.810 |
uphem | -1.755 |
hern | -1.721 |
uum | -1.697 |
leys | -1.696 |
Token | " |
Token ID | 366 |
Feature activation | -4.79e-03 |
Pos logit contributions | |
Armstrong | +0.178 |
ppo | +0.166 |
pos | +0.160 |
Cot | +0.157 |
闘 | +0.153 |
Neg logit contributions | |
quartered | -0.189 |
uphem | -0.183 |
hern | -0.179 |
leys | -0.176 |
uum | -0.176 |
Token | b |
Token ID | 65 |
Feature activation | -3.32e-01 |
Pos logit contributions | |
quartered | +0.009 |
uphem | +0.009 |
hern | +0.009 |
leys | +0.009 |
uum | +0.009 |
Neg logit contributions | |
Armstrong | -0.010 |
ppo | -0.009 |
pos | -0.009 |
Cot | -0.009 |
闘 | -0.009 |
Token | <|endoftext|> |
Token ID | 50256 |
Feature activation | +1.78e-01 |
Pos logit contributions | |
Armstrong | +0.370 |
ppo | +0.340 |
pos | +0.329 |
Cot | +0.320 |
闘 | +0.312 |
Neg logit contributions | |
quartered | -0.474 |
uphem | -0.460 |
hern | -0.451 |
uum | -0.445 |
leys | -0.445 |
Token | Image |
Token ID | 5159 |
Feature activation | -4.59e-01 |
Pos logit contributions | |
Armstrong | +0.352 |
ppo | +0.326 |
pos | +0.317 |
Cot | +0.309 |
闘 | +0.302 |
Neg logit contributions | |
quartered | -0.368 |
uphem | -0.355 |
hern | -0.347 |
leys | -0.342 |
uum | -0.342 |
Token | copyright |
Token ID | 6634 |
Feature activation | -1.02e+00 |
Pos logit contributions | |
quartered | +0.697 |
uphem | +0.669 |
hern | +0.647 |
uum | +0.635 |
leys | +0.634 |
Neg logit contributions | |
Armstrong | -1.148 |
ppo | -1.085 |
pos | -1.056 |
Cot | -1.041 |
闘 | -1.028 |
Token | publicity |
Token ID | 19403 |
Feature activation | +3.37e-01 |
Pos logit contributions | |
quartered | +2.350 |
uphem | +2.282 |
hern | +2.239 |
leys | +2.224 |
uum | +2.211 |
Neg logit contributions | |
Armstrong | -1.681 |
ppo | -1.573 |
pos | -1.493 |
闘 | -1.444 |
Cot | -1.440 |
Token | Image |
Token ID | 7412 |
Feature activation | +9.96e-02 |
Pos logit contributions | |
Armstrong | +0.605 |
ppo | +0.555 |
pos | +0.538 |
Cot | +0.525 |
闘 | +0.509 |
Neg logit contributions | |
quartered | -0.757 |
uphem | -0.735 |
hern | -0.718 |
leys | -0.709 |
uum | -0.709 |
Token | caption |
Token ID | 8305 |
Feature activation | -1.95e+00 |
Pos logit contributions | |
Armstrong | +0.218 |
pos | +0.203 |
ppo | +0.202 |
Cot | +0.194 |
eff | +0.189 |
Neg logit contributions | |
quartered | -0.192 |
uphem | -0.180 |
hern | -0.173 |
leys | -0.172 |
uum | -0.171 |
Token | The |
Token ID | 383 |
Feature activation | -3.40e-01 |
Pos logit contributions | |
quartered | +4.029 |
uphem | +3.925 |
leys | +3.832 |
hern | +3.823 |
uum | +3.796 |
Neg logit contributions | |
Armstrong | -3.476 |
ppo | -3.357 |
pos | -3.222 |
闘 | -3.149 |
Cot | -3.044 |
Token | Greeks |
Token ID | 25059 |
Feature activation | +1.43e-01 |
Pos logit contributions | |
quartered | +0.679 |
uphem | +0.656 |
hern | +0.641 |
leys | +0.632 |
uum | +0.631 |
Neg logit contributions | |
Armstrong | -0.691 |
ppo | -0.643 |
pos | -0.624 |
Cot | -0.610 |
闘 | -0.599 |
Token | were |
Token ID | 547 |
Feature activation | -1.74e-01 |
Pos logit contributions | |
Armstrong | +0.287 |
ppo | +0.264 |
pos | +0.259 |
Cot | +0.252 |
Recon | +0.245 |
Neg logit contributions | |
quartered | -0.294 |
uphem | -0.285 |
hern | -0.278 |
uum | -0.275 |
leys | -0.274 |
Token | totally |
Token ID | 6635 |
Feature activation | -9.07e-01 |
Pos logit contributions | |
quartered | +0.333 |
uphem | +0.322 |
hern | +0.313 |
leys | +0.309 |
uum | +0.309 |
Neg logit contributions | |
Armstrong | -0.371 |
ppo | -0.345 |
pos | -0.336 |
Cot | -0.329 |
闘 | -0.321 |
Token | " |
Token ID | 366 |
Feature activation | -2.65e-01 |
Pos logit contributions | |
quartered | +1.708 |
uphem | +1.648 |
hern | +1.619 |
leys | +1.584 |
uum | +1.583 |
Neg logit contributions | |
Armstrong | -1.933 |
ppo | -1.812 |
pos | -1.743 |
Cot | -1.716 |
闘 | -1.696 |
Token | at |
Token ID | 379 |
Feature activation | -5.58e-02 |
Pos logit contributions | |
Armstrong | +1.367 |
ppo | +1.236 |
pos | +1.223 |
Cot | +1.176 |
Recon | +1.142 |
Neg logit contributions | |
quartered | -1.832 |
uphem | -1.779 |
hern | -1.748 |
leys | -1.731 |
uum | -1.728 |
Token | the |
Token ID | 262 |
Feature activation | +2.57e-01 |
Pos logit contributions | |
quartered | +0.125 |
uphem | +0.122 |
hern | +0.119 |
uum | +0.118 |
leys | +0.118 |
Neg logit contributions | |
Armstrong | -0.099 |
ppo | -0.092 |
pos | -0.088 |
Cot | -0.086 |
闘 | -0.084 |
Token | Labour |
Token ID | 7179 |
Feature activation | +1.83e-01 |
Pos logit contributions | |
Armstrong | +0.481 |
ppo | +0.441 |
pos | +0.431 |
Cot | +0.420 |
Recon | +0.408 |
Neg logit contributions | |
quartered | -0.559 |
uphem | -0.542 |
hern | -0.530 |
leys | -0.524 |
uum | -0.523 |
Token | Conference |
Token ID | 8785 |
Feature activation | +7.49e-01 |
Pos logit contributions | |
Armstrong | +0.411 |
ppo | +0.382 |
pos | +0.376 |
Cot | +0.368 |
Recon | +0.359 |
Neg logit contributions | |
quartered | -0.332 |
uphem | -0.318 |
hern | -0.309 |
uum | -0.305 |
leys | -0.304 |
Token | . |
Token ID | 13 |
Feature activation | +2.07e-01 |
Pos logit contributions | |
Armstrong | +1.251 |
ppo | +1.125 |
pos | +1.113 |
Cot | +1.075 |
Recon | +1.043 |
Neg logit contributions | |
quartered | -1.764 |
uphem | -1.712 |
hern | -1.684 |
leys | -1.663 |
uum | -1.659 |
Token | Refer |
Token ID | 33973 |
Feature activation | -1.95e+00 |
Pos logit contributions | |
Armstrong | +0.390 |
ppo | +0.356 |
pos | +0.348 |
Cot | +0.340 |
Recon | +0.330 |
Neg logit contributions | |
quartered | -0.449 |
uphem | -0.434 |
hern | -0.424 |
leys | -0.419 |
uum | -0.418 |
Token | ring |
Token ID | 1806 |
Feature activation | +9.00e-02 |
Pos logit contributions | |
quartered | +2.723 |
uphem | +2.582 |
uum | +2.421 |
leys | +2.419 |
hern | +2.417 |
Neg logit contributions | |
Armstrong | -5.043 |
ppo | -4.840 |
pos | -4.664 |
闘 | -4.619 |
Cot | -4.619 |
Token | to |
Token ID | 284 |
Feature activation | -1.28e-01 |
Pos logit contributions | |
Armstrong | +0.181 |
ppo | +0.171 |
pos | +0.163 |
Cot | +0.161 |
闘 | +0.159 |
Neg logit contributions | |
quartered | -0.181 |
uphem | -0.176 |
hern | -0.171 |
leys | -0.169 |
uum | -0.167 |
Token | Cathy |
Token ID | 40636 |
Feature activation | -2.70e-01 |
Pos logit contributions | |
quartered | +0.276 |
uphem | +0.268 |
hern | +0.262 |
leys | +0.259 |
uum | +0.258 |
Neg logit contributions | |
Armstrong | -0.240 |
ppo | -0.222 |
pos | -0.214 |
Cot | -0.210 |
闘 | -0.205 |
Token | , |
Token ID | 11 |
Feature activation | -1.66e-01 |
Pos logit contributions | |
quartered | +0.575 |
uphem | +0.557 |
hern | +0.545 |
uum | +0.538 |
leys | +0.537 |
Neg logit contributions | |
Armstrong | -0.513 |
ppo | -0.476 |
pos | -0.461 |
Cot | -0.449 |
闘 | -0.439 |
Token | a |
Token ID | 257 |
Feature activation | +1.21e-01 |
Pos logit contributions | |
quartered | +0.360 |
uphem | +0.348 |
hern | +0.341 |
leys | +0.337 |
uum | +0.337 |
Neg logit contributions | |
Armstrong | -0.310 |
ppo | -0.285 |
pos | -0.278 |
Cot | -0.271 |
闘 | -0.263 |
Token | s |
Token ID | 82 |
Feature activation | +1.24e-01 |
Pos logit contributions | |
quartered | +0.639 |
uphem | +0.620 |
hern | +0.605 |
leys | +0.597 |
uum | +0.596 |
Neg logit contributions | |
Armstrong | -0.586 |
ppo | -0.541 |
pos | -0.527 |
Cot | -0.514 |
闘 | -0.501 |
Token | bonus |
Token ID | 7202 |
Feature activation | +3.37e-01 |
Pos logit contributions | |
Armstrong | +0.234 |
ppo | +0.216 |
pos | +0.209 |
Cot | +0.204 |
闘 | +0.199 |
Neg logit contributions | |
quartered | -0.268 |
uphem | -0.259 |
hern | -0.254 |
uum | -0.250 |
leys | -0.250 |
Token | , |
Token ID | 11 |
Feature activation | +8.07e-02 |
Pos logit contributions | |
Armstrong | +0.620 |
ppo | +0.570 |
pos | +0.553 |
Cot | +0.539 |
闘 | +0.525 |
Neg logit contributions | |
quartered | -0.741 |
uphem | -0.719 |
hern | -0.704 |
uum | -0.694 |
leys | -0.694 |
Token | but |
Token ID | 475 |
Feature activation | +1.36e-01 |
Pos logit contributions | |
Armstrong | +0.151 |
ppo | +0.139 |
pos | +0.135 |
Cot | +0.132 |
闘 | +0.128 |
Neg logit contributions | |
quartered | -0.175 |
uphem | -0.169 |
hern | -0.165 |
leys | -0.163 |
uum | -0.163 |
Token | now |
Token ID | 783 |
Feature activation | -7.51e-02 |
Pos logit contributions | |
Armstrong | +0.250 |
ppo | +0.229 |
pos | +0.223 |
Cot | +0.217 |
闘 | +0.211 |
Neg logit contributions | |
quartered | -0.298 |
uphem | -0.290 |
hern | -0.283 |
leys | -0.279 |
uum | -0.279 |
Token | rather |
Token ID | 2138 |
Feature activation | +1.68e+00 |
Pos logit contributions | |
quartered | +0.158 |
uphem | +0.153 |
hern | +0.150 |
leys | +0.148 |
uum | +0.147 |
Neg logit contributions | |
Armstrong | -0.145 |
ppo | -0.134 |
pos | -0.130 |
Cot | -0.127 |
闘 | -0.124 |
Token | than |
Token ID | 621 |
Feature activation | +8.63e-01 |
Pos logit contributions | |
Armstrong | +2.442 |
ppo | +2.147 |
pos | +2.138 |
Cot | +2.043 |
Chain | +1.944 |
Neg logit contributions | |
quartered | -4.326 |
uphem | -4.223 |
hern | -4.112 |
uum | -4.088 |
leys | -4.085 |
Token | having |
Token ID | 1719 |
Feature activation | +2.58e-01 |
Pos logit contributions | |
Armstrong | +1.601 |
ppo | +1.455 |
pos | +1.446 |
Cot | +1.403 |
Recon | +1.358 |
Neg logit contributions | |
quartered | -1.894 |
uphem | -1.847 |
hern | -1.786 |
uum | -1.773 |
leys | -1.770 |
Token | the |
Token ID | 262 |
Feature activation | -1.85e-01 |
Pos logit contributions | |
Armstrong | +0.482 |
ppo | +0.444 |
pos | +0.430 |
Cot | +0.420 |
闘 | +0.409 |
Neg logit contributions | |
quartered | -0.562 |
uphem | -0.545 |
hern | -0.533 |
leys | -0.526 |
uum | -0.526 |
Token | sign |
Token ID | 1051 |
Feature activation | +3.79e-02 |
Pos logit contributions | |
quartered | +0.405 |
uphem | +0.391 |
hern | +0.384 |
uum | +0.380 |
leys | +0.378 |
Neg logit contributions | |
Armstrong | -0.340 |
ppo | -0.315 |
pos | -0.300 |
Cot | -0.295 |
闘 | -0.290 |
Token | of |
Token ID | 286 |
Feature activation | +1.73e-02 |
Pos logit contributions | |
Armstrong | +0.072 |
ppo | +0.066 |
pos | +0.064 |
Cot | +0.063 |
闘 | +0.061 |
Neg logit contributions | |
quartered | -0.081 |
uphem | -0.079 |
hern | -0.077 |
uum | -0.076 |
leys | -0.076 |
Token | ains |
Token ID | 1299 |
Feature activation | +8.30e-01 |
Pos logit contributions | |
Armstrong | +0.736 |
ppo | +0.686 |
pos | +0.668 |
Cot | +0.655 |
闘 | +0.641 |
Neg logit contributions | |
quartered | -0.642 |
uphem | -0.620 |
hern | -0.605 |
uum | -0.596 |
leys | -0.596 |
Token | and |
Token ID | 290 |
Feature activation | +7.69e-02 |
Pos logit contributions | |
Armstrong | +1.484 |
ppo | +1.361 |
pos | +1.320 |
Cot | +1.285 |
Recon | +1.242 |
Neg logit contributions | |
quartered | -1.868 |
uphem | -1.815 |
hern | -1.773 |
uum | -1.749 |
leys | -1.748 |
Token | Murray |
Token ID | 12164 |
Feature activation | -4.07e-01 |
Pos logit contributions | |
Armstrong | +0.148 |
ppo | +0.137 |
pos | +0.133 |
Cot | +0.130 |
闘 | +0.126 |
Neg logit contributions | |
quartered | -0.163 |
uphem | -0.158 |
hern | -0.154 |
leys | -0.152 |
uum | -0.152 |
Token | the |
Token ID | 262 |
Feature activation | +2.75e-02 |
Pos logit contributions | |
quartered | +0.908 |
uphem | +0.879 |
hern | +0.865 |
leys | +0.856 |
uum | +0.852 |
Neg logit contributions | |
Armstrong | -0.723 |
ppo | -0.672 |
pos | -0.650 |
Cot | -0.632 |
闘 | -0.619 |
Token | B |
Token ID | 347 |
Feature activation | -2.17e-01 |
Pos logit contributions | |
Armstrong | +0.051 |
ppo | +0.048 |
pos | +0.046 |
Cot | +0.045 |
闘 | +0.044 |
Neg logit contributions | |
quartered | -0.059 |
uphem | -0.057 |
hern | -0.056 |
leys | -0.055 |
uum | -0.055 |
Token | rawn |
Token ID | 5791 |
Feature activation | +1.96e+00 |
Pos logit contributions | |
quartered | +0.426 |
uphem | +0.414 |
hern | +0.404 |
leys | +0.399 |
uum | +0.398 |
Neg logit contributions | |
Armstrong | -0.449 |
ppo | -0.416 |
pos | -0.405 |
Cot | -0.397 |
闘 | -0.390 |
Token | – |
Token ID | 784 |
Feature activation | +1.38e-02 |
Pos logit contributions | |
Armstrong | +3.386 |
ppo | +3.077 |
pos | +2.988 |
Cot | +2.891 |
Chain | +2.789 |
Neg logit contributions | |
quartered | -4.505 |
uphem | -4.383 |
hern | -4.285 |
leys | -4.192 |
uum | -4.189 |
Token | but |
Token ID | 475 |
Feature activation | -1.61e-01 |
Pos logit contributions | |
Armstrong | +0.025 |
ppo | +0.023 |
pos | +0.023 |
Cot | +0.022 |
Recon | +0.021 |
Neg logit contributions | |
quartered | -0.031 |
uphem | -0.030 |
hern | -0.029 |
leys | -0.029 |
uum | -0.029 |
Token | the |
Token ID | 262 |
Feature activation | +2.00e-01 |
Pos logit contributions | |
quartered | +0.355 |
uphem | +0.346 |
hern | +0.337 |
uum | +0.333 |
leys | +0.333 |
Neg logit contributions | |
Armstrong | -0.295 |
ppo | -0.269 |
pos | -0.263 |
Cot | -0.256 |
Recon | -0.249 |
Token | unique |
Token ID | 3748 |
Feature activation | -9.96e-01 |
Pos logit contributions | |
Armstrong | +0.392 |
ppo | +0.359 |
pos | +0.353 |
Cot | +0.343 |
Recon | +0.335 |
Neg logit contributions | |
quartered | -0.417 |
uphem | -0.406 |
hern | -0.395 |
leys | -0.388 |
uum | -0.388 |
Token | locations |
Token ID | 7064 |
Feature activation | +1.30e-01 |
Pos logit contributions | |
quartered | +2.036 |
uphem | +1.951 |
hern | +1.923 |
leys | +1.907 |
uum | +1.903 |
Neg logit contributions | |
Armstrong | -1.960 |
ppo | -1.836 |
pos | -1.754 |
Cot | -1.729 |
闘 | -1.700 |
Token | ass |
Token ID | 562 |
Feature activation | +5.01e-01 |
Pos logit contributions | |
Armstrong | +2.550 |
ppo | +2.344 |
pos | +2.279 |
Cot | +2.205 |
Recon | +2.143 |
Neg logit contributions | |
quartered | -3.149 |
uphem | -3.044 |
hern | -2.971 |
leys | -2.944 |
uum | -2.924 |
Token | as |
Token ID | 355 |
Feature activation | -6.48e-02 |
Pos logit contributions | |
Armstrong | +0.943 |
ppo | +0.865 |
pos | +0.847 |
Cot | +0.823 |
Recon | +0.799 |
Neg logit contributions | |
quartered | -1.083 |
uphem | -1.049 |
hern | -1.025 |
uum | -1.011 |
leys | -1.011 |
Token | bait |
Token ID | 26536 |
Feature activation | +7.45e-01 |
Pos logit contributions | |
quartered | +0.138 |
uphem | +0.134 |
hern | +0.132 |
leys | +0.130 |
uum | +0.129 |
Neg logit contributions | |
Armstrong | -0.122 |
ppo | -0.114 |
pos | -0.108 |
Cot | -0.107 |
闘 | -0.105 |
Token | , |
Token ID | 11 |
Feature activation | +1.65e-01 |
Pos logit contributions | |
Armstrong | +1.341 |
ppo | +1.234 |
pos | +1.190 |
Cot | +1.164 |
闘 | +1.143 |
Neg logit contributions | |
quartered | -1.667 |
uphem | -1.616 |
hern | -1.585 |
leys | -1.564 |
uum | -1.561 |
Token | they |
Token ID | 484 |
Feature activation | +2.43e-01 |
Pos logit contributions | |
Armstrong | +0.308 |
ppo | +0.284 |
pos | +0.276 |
Cot | +0.269 |
闘 | +0.262 |
Neg logit contributions | |
quartered | -0.359 |
uphem | -0.348 |
hern | -0.341 |
leys | -0.336 |
uum | -0.336 |
Token | caught |
Token ID | 4978 |
Feature activation | +1.50e+00 |
Pos logit contributions | |
Armstrong | +0.493 |
ppo | +0.458 |
pos | +0.448 |
Cot | +0.435 |
Recon | +0.424 |
Neg logit contributions | |
quartered | -0.487 |
uphem | -0.471 |
hern | -0.460 |
uum | -0.453 |
leys | -0.453 |
Token | the |
Token ID | 262 |
Feature activation | +3.84e-02 |
Pos logit contributions | |
Armstrong | +2.696 |
ppo | +2.435 |
pos | +2.412 |
Cot | +2.356 |
eff | +2.237 |
Neg logit contributions | |
quartered | -3.410 |
uphem | -3.306 |
hern | -3.214 |
uum | -3.206 |
leys | -3.190 |
Token | beast |
Token ID | 13824 |
Feature activation | -7.32e-02 |
Pos logit contributions | |
Armstrong | +0.074 |
ppo | +0.069 |
pos | +0.067 |
Cot | +0.065 |
闘 | +0.064 |
Neg logit contributions | |
quartered | -0.080 |
uphem | -0.078 |
hern | -0.076 |
leys | -0.075 |
uum | -0.075 |
Token | and |
Token ID | 290 |
Feature activation | -1.43e-01 |
Pos logit contributions | |
quartered | +0.161 |
uphem | +0.156 |
hern | +0.153 |
leys | +0.151 |
uum | +0.151 |
Neg logit contributions | |
Armstrong | -0.134 |
ppo | -0.124 |
pos | -0.120 |
Cot | -0.117 |
闘 | -0.114 |
Token | it |
Token ID | 340 |
Feature activation | +2.14e-01 |
Pos logit contributions | |
quartered | +0.281 |
uphem | +0.272 |
hern | +0.265 |
leys | +0.262 |
uum | +0.261 |
Neg logit contributions | |
Armstrong | -0.298 |
ppo | -0.277 |
pos | -0.271 |
Cot | -0.264 |
闘 | -0.257 |
Token | 's |
Token ID | 338 |
Feature activation | -3.32e-01 |
Pos logit contributions | |
Armstrong | +0.403 |
ppo | +0.372 |
pos | +0.361 |
Cot | +0.352 |
闘 | +0.342 |
Neg logit contributions | |
quartered | -0.462 |
uphem | -0.448 |
hern | -0.438 |
leys | -0.433 |
uum | -0.432 |
Token | in |
Token ID | 287 |
Feature activation | +2.01e-01 |
Pos logit contributions | |
Armstrong | +1.258 |
ppo | +1.133 |
pos | +1.129 |
Cot | +1.090 |
Recon | +1.046 |
Neg logit contributions | |
quartered | -1.645 |
uphem | -1.593 |
hern | -1.556 |
uum | -1.544 |
leys | -1.542 |
Token | a |
Token ID | 257 |
Feature activation | +1.46e-01 |
Pos logit contributions | |
Armstrong | +0.399 |
ppo | +0.367 |
pos | +0.360 |
Cot | +0.352 |
Recon | +0.340 |
Neg logit contributions | |
quartered | -0.416 |
uphem | -0.401 |
hern | -0.392 |
leys | -0.387 |
uum | -0.387 |
Token | fab |
Token ID | 7843 |
Feature activation | -2.40e-01 |
Pos logit contributions | |
Armstrong | +0.310 |
ppo | +0.285 |
pos | +0.283 |
Cot | +0.275 |
Recon | +0.267 |
Neg logit contributions | |
quartered | -0.283 |
uphem | -0.273 |
hern | -0.266 |
leys | -0.263 |
uum | -0.262 |
Token | lab |
Token ID | 2248 |
Feature activation | -3.87e-01 |
Pos logit contributions | |
quartered | +0.394 |
uphem | +0.378 |
hern | +0.367 |
uum | +0.361 |
leys | +0.360 |
Neg logit contributions | |
Armstrong | -0.576 |
ppo | -0.541 |
pos | -0.528 |
Cot | -0.519 |
闘 | -0.509 |
Token | but |
Token ID | 475 |
Feature activation | +2.56e-01 |
Pos logit contributions | |
quartered | +0.856 |
uphem | +0.832 |
hern | +0.816 |
uum | +0.806 |
leys | +0.802 |
Neg logit contributions | |
Armstrong | -0.704 |
ppo | -0.650 |
pos | -0.625 |
Cot | -0.613 |
闘 | -0.600 |
Token | instead |
Token ID | 2427 |
Feature activation | +1.65e+00 |
Pos logit contributions | |
Armstrong | +0.429 |
ppo | +0.391 |
pos | +0.379 |
Cot | +0.368 |
闘 | +0.355 |
Neg logit contributions | |
quartered | -0.607 |
uphem | -0.590 |
hern | -0.578 |
uum | -0.571 |
leys | -0.571 |
Token | , |
Token ID | 11 |
Feature activation | +1.99e-02 |
Pos logit contributions | |
Armstrong | +2.776 |
pos | +2.544 |
ppo | +2.505 |
Cot | +2.395 |
eff | +2.307 |
Neg logit contributions | |
quartered | -3.867 |
uphem | -3.772 |
hern | -3.645 |
uum | -3.626 |
leys | -3.614 |
Token | turns |
Token ID | 4962 |
Feature activation | -6.29e-01 |
Pos logit contributions | |
Armstrong | +0.035 |
ppo | +0.032 |
pos | +0.031 |
Cot | +0.030 |
闘 | +0.029 |
Neg logit contributions | |
quartered | -0.045 |
uphem | -0.044 |
hern | -0.043 |
leys | -0.043 |
uum | -0.043 |
Token | our |
Token ID | 674 |
Feature activation | +1.18e-01 |
Pos logit contributions | |
quartered | +1.329 |
uphem | +1.289 |
hern | +1.261 |
leys | +1.244 |
uum | +1.243 |
Neg logit contributions | |
Armstrong | -1.196 |
ppo | -1.114 |
pos | -1.066 |
Cot | -1.049 |
闘 | -1.034 |
Token | attention |
Token ID | 3241 |
Feature activation | -2.37e-01 |
Pos logit contributions | |
Armstrong | +0.243 |
ppo | +0.225 |
pos | +0.222 |
Cot | +0.215 |
Recon | +0.209 |
Neg logit contributions | |
quartered | -0.234 |
uphem | -0.227 |
hern | -0.221 |
uum | -0.217 |
leys | -0.217 |
Token | to |
Token ID | 284 |
Feature activation | -1.84e-01 |
Pos logit contributions | |
quartered | +0.478 |
uphem | +0.460 |
hern | +0.452 |
uum | +0.445 |
leys | +0.445 |
Neg logit contributions | |
Armstrong | -0.476 |
ppo | -0.441 |
pos | -0.427 |
Cot | -0.420 |
闘 | -0.414 |
Token | cramped |
Token ID | 43295 |
Feature activation | -6.32e-01 |
Pos logit contributions | |
quartered | +0.078 |
uphem | +0.075 |
hern | +0.073 |
leys | +0.072 |
uum | +0.072 |
Neg logit contributions | |
Armstrong | -0.086 |
ppo | -0.080 |
pos | -0.078 |
Cot | -0.077 |
闘 | -0.075 |
Token | rooms |
Token ID | 9649 |
Feature activation | -3.22e-01 |
Pos logit contributions | |
quartered | +1.137 |
uphem | +1.094 |
hern | +1.065 |
leys | +1.052 |
uum | +1.049 |
Neg logit contributions | |
Armstrong | -1.412 |
ppo | -1.324 |
pos | -1.283 |
Cot | -1.264 |
闘 | -1.239 |
Token | . |
Token ID | 13 |
Feature activation | +1.39e-01 |
Pos logit contributions | |
quartered | +0.699 |
uphem | +0.678 |
hern | +0.663 |
leys | +0.654 |
uum | +0.654 |
Neg logit contributions | |
Armstrong | -0.601 |
ppo | -0.554 |
pos | -0.537 |
Cot | -0.524 |
闘 | -0.511 |
Token | And |
Token ID | 843 |
Feature activation | -1.37e-01 |
Pos logit contributions | |
Armstrong | +0.256 |
ppo | +0.232 |
pos | +0.226 |
Cot | +0.222 |
Recon | +0.216 |
Neg logit contributions | |
quartered | -0.313 |
uphem | -0.303 |
hern | -0.294 |
uum | -0.291 |
leys | -0.290 |
Token | , |
Token ID | 11 |
Feature activation | -8.88e-02 |
Pos logit contributions | |
quartered | +0.312 |
uphem | +0.304 |
hern | +0.296 |
uum | +0.293 |
leys | +0.292 |
Neg logit contributions | |
Armstrong | -0.241 |
ppo | -0.220 |
pos | -0.214 |
Cot | -0.208 |
Recon | -0.201 |
Token | instead |
Token ID | 2427 |
Feature activation | +1.51e+00 |
Pos logit contributions | |
quartered | +0.196 |
uphem | +0.191 |
hern | +0.185 |
uum | +0.183 |
leys | +0.183 |
Neg logit contributions | |
Armstrong | -0.164 |
ppo | -0.149 |
pos | -0.147 |
Cot | -0.143 |
Recon | -0.137 |
Token | of |
Token ID | 286 |
Feature activation | +4.69e-01 |
Pos logit contributions | |
Armstrong | +2.576 |
pos | +2.334 |
ppo | +2.293 |
Cot | +2.200 |
eff | +2.139 |
Neg logit contributions | |
quartered | -3.441 |
uphem | -3.384 |
hern | -3.309 |
uum | -3.248 |
leys | -3.238 |
Token | allowing |
Token ID | 5086 |
Feature activation | +2.11e-01 |
Pos logit contributions | |
Armstrong | +0.938 |
ppo | +0.860 |
pos | +0.848 |
Cot | +0.826 |
Recon | +0.801 |
Neg logit contributions | |
quartered | -0.962 |
uphem | -0.933 |
hern | -0.907 |
uum | -0.897 |
leys | -0.895 |
Token | the |
Token ID | 262 |
Feature activation | +9.25e-03 |
Pos logit contributions | |
Armstrong | +0.395 |
ppo | +0.361 |
pos | +0.353 |
Cot | +0.344 |
Recon | +0.333 |
Neg logit contributions | |
quartered | -0.459 |
uphem | -0.444 |
hern | -0.434 |
uum | -0.429 |
leys | -0.428 |
Token | audience |
Token ID | 5386 |
Feature activation | +2.94e-01 |
Pos logit contributions | |
Armstrong | +0.019 |
ppo | +0.017 |
pos | +0.017 |
Cot | +0.016 |
闘 | +0.016 |
Neg logit contributions | |
quartered | -0.019 |
uphem | -0.018 |
hern | -0.018 |
leys | -0.018 |
uum | -0.018 |
Token | to |
Token ID | 284 |
Feature activation | +7.03e-02 |
Pos logit contributions | |
Armstrong | +0.576 |
ppo | +0.533 |
pos | +0.516 |
Cot | +0.506 |
闘 | +0.495 |
Neg logit contributions | |
quartered | -0.609 |
uphem | -0.590 |
hern | -0.576 |
leys | -0.568 |
uum | -0.567 |
Token | those |
Token ID | 883 |
Feature activation | -4.84e-01 |
Pos logit contributions | |
Armstrong | +0.263 |
ppo | +0.245 |
pos | +0.237 |
Cot | +0.232 |
闘 | +0.228 |
Neg logit contributions | |
quartered | -0.260 |
uphem | -0.252 |
hern | -0.246 |
leys | -0.243 |
uum | -0.243 |
Token | things |
Token ID | 1243 |
Feature activation | +3.68e-01 |
Pos logit contributions | |
quartered | +1.008 |
uphem | +0.980 |
hern | +0.957 |
uum | +0.950 |
leys | +0.947 |
Neg logit contributions | |
Armstrong | -0.937 |
ppo | -0.869 |
pos | -0.831 |
Cot | -0.820 |
闘 | -0.808 |
Token | directly |
Token ID | 3264 |
Feature activation | +4.04e-02 |
Pos logit contributions | |
Armstrong | +0.648 |
ppo | +0.590 |
pos | +0.581 |
Cot | +0.561 |
Recon | +0.541 |
Neg logit contributions | |
quartered | -0.844 |
uphem | -0.816 |
hern | -0.800 |
uum | -0.789 |
leys | -0.789 |
Token | . |
Token ID | 13 |
Feature activation | -6.60e-02 |
Pos logit contributions | |
Armstrong | +0.072 |
ppo | +0.066 |
pos | +0.064 |
Cot | +0.063 |
闘 | +0.061 |
Neg logit contributions | |
quartered | -0.091 |
uphem | -0.088 |
hern | -0.087 |
leys | -0.085 |
uum | -0.085 |
Token | We |
Token ID | 775 |
Feature activation | +3.90e-02 |
Pos logit contributions | |
quartered | +0.144 |
uphem | +0.138 |
hern | +0.135 |
leys | +0.133 |
uum | +0.133 |
Neg logit contributions | |
Armstrong | -0.125 |
ppo | -0.114 |
pos | -0.111 |
Cot | -0.109 |
Recon | -0.106 |
Token | just |
Token ID | 655 |
Feature activation | +4.89e-01 |
Pos logit contributions | |
Armstrong | +0.072 |
ppo | +0.066 |
pos | +0.065 |
Cot | +0.063 |
闘 | +0.061 |
Neg logit contributions | |
quartered | -0.086 |
uphem | -0.083 |
hern | -0.081 |
leys | -0.080 |
uum | -0.080 |
Token | more |
Token ID | 517 |
Feature activation | +2.49e-01 |
Pos logit contributions | |
Armstrong | +0.921 |
ppo | +0.849 |
pos | +0.837 |
Cot | +0.803 |
Recon | +0.783 |
Neg logit contributions | |
quartered | -1.059 |
uphem | -1.019 |
hern | -1.000 |
leys | -0.983 |
uum | -0.980 |
Token | like |
Token ID | 588 |
Feature activation | +4.07e-01 |
Pos logit contributions | |
Armstrong | +0.474 |
ppo | +0.437 |
pos | +0.423 |
Cot | +0.415 |
闘 | +0.404 |
Neg logit contributions | |
quartered | -0.531 |
uphem | -0.515 |
hern | -0.503 |
uum | -0.496 |
leys | -0.496 |
Token | to |
Token ID | 284 |
Feature activation | +2.35e-01 |
Pos logit contributions | |
Armstrong | +0.774 |
ppo | +0.713 |
pos | +0.698 |
Cot | +0.677 |
Recon | +0.657 |
Neg logit contributions | |
quartered | -0.870 |
uphem | -0.841 |
hern | -0.825 |
uum | -0.812 |
leys | -0.812 |
Token | , |
Token ID | 11 |
Feature activation | +3.87e-02 |
Pos logit contributions | |
Armstrong | +0.449 |
ppo | +0.415 |
pos | +0.407 |
Cot | +0.394 |
Recon | +0.383 |
Neg logit contributions | |
quartered | -0.501 |
uphem | -0.483 |
hern | -0.473 |
leys | -0.467 |
uum | -0.466 |
Token | at |
Token ID | 379 |
Feature activation | +2.65e-01 |
Pos logit contributions | |
Armstrong | +0.071 |
ppo | +0.065 |
pos | +0.064 |
Cot | +0.062 |
Recon | +0.060 |
Neg logit contributions | |
quartered | -0.086 |
uphem | -0.083 |
hern | -0.081 |
leys | -0.080 |
uum | -0.080 |
Token | ," |
Token ID | 553 |
Feature activation | +1.78e-01 |
Pos logit contributions | |
Armstrong | +0.077 |
ppo | +0.071 |
pos | +0.069 |
Cot | +0.067 |
Recon | +0.066 |
Neg logit contributions | |
quartered | -0.086 |
uphem | -0.084 |
hern | -0.082 |
leys | -0.081 |
uum | -0.080 |
Token | I |
Token ID | 314 |
Feature activation | +9.57e-02 |
Pos logit contributions | |
Armstrong | +0.323 |
ppo | +0.295 |
pos | +0.288 |
Cot | +0.281 |
Recon | +0.272 |
Neg logit contributions | |
quartered | -0.401 |
uphem | -0.388 |
hern | -0.379 |
leys | -0.374 |
uum | -0.374 |
Token | basically |
Token ID | 6209 |
Feature activation | -5.84e-02 |
Pos logit contributions | |
Armstrong | +0.192 |
ppo | +0.177 |
pos | +0.173 |
Cot | +0.169 |
闘 | +0.164 |
Neg logit contributions | |
quartered | -0.196 |
uphem | -0.189 |
hern | -0.184 |
leys | -0.182 |
uum | -0.182 |
Token | sat |
Token ID | 3332 |
Feature activation | -1.11e+00 |
Pos logit contributions | |
quartered | +0.113 |
uphem | +0.109 |
hern | +0.106 |
leys | +0.104 |
uum | +0.104 |
Neg logit contributions | |
Armstrong | -0.123 |
ppo | -0.115 |
pos | -0.112 |
Cot | -0.109 |
Recon | -0.107 |
Token | down |
Token ID | 866 |
Feature activation | +6.97e-01 |
Pos logit contributions | |
quartered | +2.404 |
uphem | +2.338 |
hern | +2.294 |
uum | +2.268 |
leys | +2.264 |
Neg logit contributions | |
Armstrong | -2.053 |
ppo | -1.895 |
pos | -1.833 |
Cot | -1.793 |
闘 | -1.768 |
Token | in |
Token ID | 287 |
Feature activation | +3.21e-01 |
Pos logit contributions | |
Armstrong | +1.226 |
ppo | +1.112 |
pos | +1.091 |
Cot | +1.057 |
Recon | +1.020 |
Neg logit contributions | |
quartered | -1.607 |
uphem | -1.558 |
hern | -1.508 |
uum | -1.495 |
leys | -1.494 |
Token | my |
Token ID | 616 |
Feature activation | +1.81e-02 |
Pos logit contributions | |
Armstrong | +0.588 |
ppo | +0.538 |
pos | +0.525 |
Cot | +0.510 |
Recon | +0.495 |
Neg logit contributions | |
quartered | -0.711 |
uphem | -0.688 |
hern | -0.674 |
leys | -0.666 |
uum | -0.665 |
Token | design |
Token ID | 1486 |
Feature activation | +6.69e-01 |
Pos logit contributions | |
Armstrong | +0.042 |
ppo | +0.039 |
pos | +0.038 |
Cot | +0.038 |
Recon | +0.037 |
Neg logit contributions | |
quartered | -0.031 |
uphem | -0.030 |
hern | -0.029 |
leys | -0.029 |
uum | -0.029 |
Token | doc |
Token ID | 2205 |
Feature activation | -5.16e-01 |
Pos logit contributions | |
Armstrong | +1.598 |
ppo | +1.491 |
pos | +1.469 |
Cot | +1.436 |
Recon | +1.404 |
Neg logit contributions | |
quartered | -1.110 |
uphem | -1.063 |
hern | -1.036 |
leys | -1.016 |
uum | -1.012 |
Token | and |
Token ID | 290 |
Feature activation | -1.19e-01 |
Pos logit contributions | |
quartered | +1.106 |
uphem | +1.077 |
hern | +1.055 |
uum | +1.046 |
leys | +1.041 |
Neg logit contributions | |
Armstrong | -0.968 |
ppo | -0.895 |
pos | -0.862 |
Cot | -0.850 |
闘 | -0.831 |
Token | I |
Token ID | 314 |
Feature activation | +2.20e-01 |
Pos logit contributions | |
quartered | +0.236 |
uphem | +0.228 |
hern | +0.222 |
leys | +0.218 |
uum | +0.218 |
Neg logit contributions | |
Armstrong | -0.248 |
ppo | -0.230 |
pos | -0.226 |
Cot | -0.220 |
Recon | -0.214 |
Token | . |
Token ID | 13 |
Feature activation | +6.02e-02 |
Pos logit contributions | |
Armstrong | +0.607 |
ppo | +0.557 |
pos | +0.544 |
Cot | +0.527 |
Recon | +0.509 |
Neg logit contributions | |
quartered | -0.755 |
uphem | -0.731 |
hern | -0.714 |
leys | -0.705 |
uum | -0.705 |
Token | My |
Token ID | 2011 |
Feature activation | -3.53e-01 |
Pos logit contributions | |
Armstrong | +0.113 |
ppo | +0.103 |
pos | +0.100 |
Cot | +0.099 |
Recon | +0.096 |
Neg logit contributions | |
quartered | -0.133 |
uphem | -0.129 |
hern | -0.125 |
uum | -0.124 |
leys | -0.123 |
Token | deeds |
Token ID | 23777 |
Feature activation | +7.02e-03 |
Pos logit contributions | |
quartered | +0.734 |
uphem | +0.710 |
hern | +0.694 |
leys | +0.685 |
uum | +0.684 |
Neg logit contributions | |
Armstrong | -0.689 |
ppo | -0.638 |
pos | -0.617 |
Cot | -0.604 |
闘 | -0.592 |
Token | , |
Token ID | 11 |
Feature activation | +3.74e-01 |
Pos logit contributions | |
Armstrong | +0.013 |
ppo | +0.012 |
pos | +0.012 |
Cot | +0.012 |
Recon | +0.011 |
Neg logit contributions | |
quartered | -0.015 |
uphem | -0.015 |
hern | -0.014 |
uum | -0.014 |
leys | -0.014 |
Token | whatever |
Token ID | 4232 |
Feature activation | +1.01e+00 |
Pos logit contributions | |
Armstrong | +0.664 |
ppo | +0.597 |
pos | +0.595 |
Cot | +0.574 |
Recon | +0.553 |
Neg logit contributions | |
quartered | -0.855 |
uphem | -0.829 |
hern | -0.808 |
uum | -0.801 |
leys | -0.798 |
Token | they |
Token ID | 484 |
Feature activation | +3.62e-01 |
Pos logit contributions | |
Armstrong | +1.728 |
ppo | +1.561 |
pos | +1.550 |
Cot | +1.491 |
Recon | +1.437 |
Neg logit contributions | |
quartered | -2.342 |
uphem | -2.276 |
hern | -2.220 |
uum | -2.200 |
leys | -2.191 |
Token | may |
Token ID | 743 |
Feature activation | -3.92e-01 |
Pos logit contributions | |
Armstrong | +0.641 |
ppo | +0.589 |
pos | +0.567 |
Cot | +0.555 |
闘 | +0.541 |
Neg logit contributions | |
quartered | -0.819 |
uphem | -0.797 |
hern | -0.780 |
leys | -0.770 |
uum | -0.770 |
Token | be |
Token ID | 307 |
Feature activation | -1.30e-01 |
Pos logit contributions | |
quartered | +0.835 |
uphem | +0.816 |
hern | +0.798 |
leys | +0.786 |
uum | +0.784 |
Neg logit contributions | |
Armstrong | -0.738 |
ppo | -0.685 |
Cot | -0.645 |
pos | -0.641 |
闘 | -0.634 |
Token | , |
Token ID | 11 |
Feature activation | +2.32e-01 |
Pos logit contributions | |
quartered | +0.291 |
uphem | +0.282 |
hern | +0.277 |
leys | +0.273 |
uum | +0.273 |
Neg logit contributions | |
Armstrong | -0.234 |
ppo | -0.215 |
pos | -0.206 |
Cot | -0.203 |
闘 | -0.199 |
Token | will |
Token ID | 481 |
Feature activation | +3.57e-01 |
Pos logit contributions | |
Armstrong | +0.409 |
ppo | +0.372 |
pos | +0.365 |
Cot | +0.353 |
Recon | +0.342 |
Neg logit contributions | |
quartered | -0.532 |
uphem | -0.516 |
hern | -0.504 |
uum | -0.499 |
leys | -0.497 |
Token | be |
Token ID | 307 |
Feature activation | -1.01e-01 |
Pos logit contributions | |
Armstrong | +0.643 |
ppo | +0.589 |
pos | +0.576 |
Cot | +0.557 |
Recon | +0.540 |
Neg logit contributions | |
quartered | -0.800 |
uphem | -0.776 |
hern | -0.757 |
uum | -0.751 |
leys | -0.749 |
Token | of |
Token ID | 286 |
Feature activation | -3.17e-01 |
Pos logit contributions | |
Armstrong | +0.213 |
ppo | +0.198 |
pos | +0.191 |
Cot | +0.188 |
Recon | +0.184 |
Neg logit contributions | |
quartered | -0.213 |
uphem | -0.204 |
hern | -0.202 |
leys | -0.199 |
uum | -0.197 |
Token | new |
Token ID | 649 |
Feature activation | +3.20e-01 |
Pos logit contributions | |
quartered | +0.627 |
uphem | +0.606 |
hern | +0.592 |
leys | +0.583 |
uum | +0.583 |
Neg logit contributions | |
Armstrong | -0.653 |
ppo | -0.607 |
pos | -0.590 |
Cot | -0.577 |
闘 | -0.565 |
Token | army |
Token ID | 5428 |
Feature activation | +3.47e-01 |
Pos logit contributions | |
Armstrong | +0.644 |
ppo | +0.595 |
pos | +0.586 |
Cot | +0.571 |
Recon | +0.554 |
Neg logit contributions | |
quartered | -0.655 |
uphem | -0.635 |
hern | -0.613 |
uum | -0.605 |
leys | -0.602 |
Token | - |
Token ID | 12 |
Feature activation | -2.55e-01 |
Pos logit contributions | |
Armstrong | +0.678 |
ppo | +0.628 |
pos | +0.616 |
Cot | +0.596 |
Recon | +0.579 |
Neg logit contributions | |
quartered | -0.723 |
uphem | -0.704 |
hern | -0.683 |
uum | -0.675 |
leys | -0.672 |
Token | building |
Token ID | 16894 |
Feature activation | -3.05e-01 |
Pos logit contributions | |
quartered | +0.489 |
uphem | +0.471 |
hern | +0.461 |
uum | +0.453 |
leys | +0.452 |
Neg logit contributions | |
Armstrong | -0.542 |
ppo | -0.504 |
pos | -0.490 |
Cot | -0.482 |
闘 | -0.471 |
Token | options |
Token ID | 3689 |
Feature activation | +1.87e-01 |
Pos logit contributions | |
quartered | +0.632 |
uphem | +0.612 |
hern | +0.600 |
leys | +0.591 |
uum | +0.590 |
Neg logit contributions | |
Armstrong | -0.596 |
ppo | -0.551 |
pos | -0.534 |
Cot | -0.523 |
闘 | -0.512 |
Token | and |
Token ID | 290 |
Feature activation | +1.00e-01 |
Pos logit contributions | |
Armstrong | +0.346 |
ppo | +0.318 |
pos | +0.309 |
Cot | +0.301 |
闘 | +0.293 |
Neg logit contributions | |
quartered | -0.409 |
uphem | -0.397 |
hern | -0.387 |
leys | -0.383 |
uum | -0.383 |
Token | give |
Token ID | 1577 |
Feature activation | -1.91e-01 |
Pos logit contributions | |
Armstrong | +0.199 |
ppo | +0.183 |
pos | +0.179 |
Cot | +0.175 |
Recon | +0.170 |
Neg logit contributions | |
quartered | -0.207 |
uphem | -0.200 |
hern | -0.195 |
uum | -0.192 |
leys | -0.192 |
Token | many |
Token ID | 867 |
Feature activation | +1.17e-01 |
Pos logit contributions | |
quartered | +0.401 |
uphem | +0.389 |
hern | +0.380 |
leys | +0.375 |
uum | +0.375 |
Neg logit contributions | |
Armstrong | -0.371 |
ppo | -0.343 |
pos | -0.333 |
Cot | -0.325 |
闘 | -0.317 |
Token | models |
Token ID | 4981 |
Feature activation | +4.67e-01 |
Pos logit contributions | |
Armstrong | +0.221 |
ppo | +0.203 |
pos | +0.199 |
Cot | +0.194 |
Recon | +0.188 |
Neg logit contributions | |
quartered | -0.250 |
uphem | -0.243 |
hern | -0.238 |
leys | -0.234 |
uum | -0.234 |
Token | that |
Token ID | 326 |
Feature activation | -6.08e-02 |
Pos logit contributions | |
Armstrong | +0.878 |
ppo | +0.805 |
pos | +0.790 |
Cot | +0.766 |
闘 | +0.741 |
Neg logit contributions | |
quartered | -1.015 |
uphem | -0.986 |
hern | -0.958 |
uum | -0.947 |
leys | -0.945 |
Token | construction |
Token ID | 5103 |
Feature activation | +6.85e-02 |
Pos logit contributions | |
quartered | +0.254 |
uphem | +0.247 |
hern | +0.243 |
leys | +0.238 |
uum | +0.237 |
Neg logit contributions | |
Armstrong | -0.281 |
ppo | -0.262 |
pos | -0.254 |
Cot | -0.250 |
闘 | -0.245 |
Token | , |
Token ID | 11 |
Feature activation | +1.20e-01 |
Pos logit contributions | |
Armstrong | +0.124 |
ppo | +0.114 |
pos | +0.110 |
Cot | +0.108 |
闘 | +0.105 |
Neg logit contributions | |
quartered | -0.152 |
uphem | -0.148 |
hern | -0.145 |
leys | -0.143 |
uum | -0.143 |
Token | managing |
Token ID | 11149 |
Feature activation | -2.68e-02 |
Pos logit contributions | |
Armstrong | +0.244 |
ppo | +0.226 |
pos | +0.220 |
Cot | +0.215 |
闘 | +0.210 |
Neg logit contributions | |
quartered | -0.240 |
uphem | -0.232 |
hern | -0.226 |
leys | -0.223 |
uum | -0.223 |
Token | the |
Token ID | 262 |
Feature activation | +8.22e-03 |
Pos logit contributions | |
quartered | +0.055 |
uphem | +0.053 |
hern | +0.052 |
leys | +0.051 |
uum | +0.051 |
Neg logit contributions | |
Armstrong | -0.053 |
ppo | -0.049 |
pos | -0.048 |
Cot | -0.047 |
闘 | -0.046 |
Token | existing |
Token ID | 4683 |
Feature activation | +1.72e-01 |
Pos logit contributions | |
Armstrong | +0.016 |
ppo | +0.015 |
pos | +0.014 |
Cot | +0.014 |
闘 | +0.014 |
Neg logit contributions | |
quartered | -0.017 |
uphem | -0.017 |
hern | -0.016 |
leys | -0.016 |
uum | -0.016 |
Token | reservoirs |
Token ID | 46637 |
Feature activation | +2.14e-01 |
Pos logit contributions | |
Armstrong | +0.329 |
ppo | +0.303 |
pos | +0.296 |
Cot | +0.289 |
闘 | +0.280 |
Neg logit contributions | |
quartered | -0.364 |
uphem | -0.353 |
hern | -0.345 |
uum | -0.340 |
leys | -0.340 |
Token | and |
Token ID | 290 |
Feature activation | +6.58e-02 |
Pos logit contributions | |
Armstrong | +0.382 |
ppo | +0.351 |
pos | +0.341 |
Cot | +0.332 |
闘 | +0.322 |
Neg logit contributions | |
quartered | -0.481 |
uphem | -0.467 |
hern | -0.456 |
leys | -0.451 |
uum | -0.451 |
Token | regulating |
Token ID | 26379 |
Feature activation | +5.43e-02 |
Pos logit contributions | |
Armstrong | +0.136 |
ppo | +0.127 |
pos | +0.123 |
Cot | +0.120 |
闘 | +0.118 |
Neg logit contributions | |
quartered | -0.129 |
uphem | -0.125 |
hern | -0.122 |
leys | -0.120 |
uum | -0.120 |
Token | the |
Token ID | 262 |
Feature activation | -4.20e-02 |
Pos logit contributions | |
Armstrong | +0.105 |
ppo | +0.097 |
pos | +0.094 |
Cot | +0.092 |
闘 | +0.090 |
Neg logit contributions | |
quartered | -0.114 |
uphem | -0.111 |
hern | -0.108 |
leys | -0.107 |
uum | -0.107 |
Token | use |
Token ID | 779 |
Feature activation | +3.49e-01 |
Pos logit contributions | |
quartered | +0.089 |
uphem | +0.087 |
hern | +0.085 |
leys | +0.084 |
uum | +0.084 |
Neg logit contributions | |
Armstrong | -0.080 |
ppo | -0.074 |
pos | -0.072 |
Cot | -0.070 |
闘 | -0.068 |
Token | of |
Token ID | 286 |
Feature activation | -8.10e-02 |
Pos logit contributions | |
Armstrong | +0.705 |
ppo | +0.655 |
pos | +0.632 |
Cot | +0.622 |
Recon | +0.607 |
Neg logit contributions | |
quartered | -0.701 |
uphem | -0.682 |
hern | -0.669 |
leys | -0.662 |
uum | -0.656 |
Token | things |
Token ID | 1243 |
Feature activation | +2.53e-01 |
Pos logit contributions | |
quartered | +2.388 |
uphem | +2.310 |
hern | +2.255 |
leys | +2.226 |
uum | +2.222 |
Neg logit contributions | |
Armstrong | -2.373 |
ppo | -2.200 |
pos | -2.134 |
Cot | -2.091 |
闘 | -2.045 |
Token | are |
Token ID | 389 |
Feature activation | +2.37e-02 |
Pos logit contributions | |
Armstrong | +0.423 |
ppo | +0.386 |
pos | +0.377 |
Cot | +0.362 |
Recon | +0.351 |
Neg logit contributions | |
quartered | -0.602 |
uphem | -0.584 |
hern | -0.573 |
uum | -0.565 |
leys | -0.564 |
Token | highlighted |
Token ID | 14537 |
Feature activation | -3.37e-01 |
Pos logit contributions | |
Armstrong | +0.044 |
pos | +0.040 |
ppo | +0.040 |
Cot | +0.038 |
Recon | +0.037 |
Neg logit contributions | |
quartered | -0.053 |
uphem | -0.051 |
hern | -0.049 |
uum | -0.049 |
leys | -0.049 |
Token | and |
Token ID | 290 |
Feature activation | -2.69e-02 |
Pos logit contributions | |
quartered | +0.751 |
uphem | +0.729 |
hern | +0.715 |
leys | +0.705 |
uum | +0.704 |
Neg logit contributions | |
Armstrong | -0.608 |
ppo | -0.560 |
pos | -0.540 |
Cot | -0.529 |
闘 | -0.515 |
Token | certain |
Token ID | 1728 |
Feature activation | -1.27e+00 |
Pos logit contributions | |
quartered | +0.061 |
uphem | +0.060 |
hern | +0.058 |
leys | +0.058 |
uum | +0.058 |
Neg logit contributions | |
Armstrong | -0.048 |
ppo | -0.043 |
pos | -0.043 |
Cot | -0.041 |
Recon | -0.040 |
Token | things |
Token ID | 1243 |
Feature activation | +7.89e-02 |
Pos logit contributions | |
quartered | +2.500 |
uphem | +2.417 |
hern | +2.357 |
leys | +2.335 |
uum | +2.323 |
Neg logit contributions | |
Armstrong | -2.626 |
ppo | -2.441 |
pos | -2.351 |
Cot | -2.316 |
闘 | -2.279 |
Token | removed |
Token ID | 4615 |
Feature activation | -4.91e-01 |
Pos logit contributions | |
Armstrong | +0.149 |
ppo | +0.138 |
pos | +0.134 |
Cot | +0.130 |
闘 | +0.127 |
Neg logit contributions | |
quartered | -0.170 |
uphem | -0.164 |
hern | -0.161 |
leys | -0.159 |
uum | -0.158 |
Token | . |
Token ID | 13 |
Feature activation | +4.26e-02 |
Pos logit contributions | |
quartered | +1.083 |
uphem | +1.054 |
hern | +1.033 |
uum | +1.019 |
leys | +1.018 |
Neg logit contributions | |
Armstrong | -0.892 |
ppo | -0.821 |
pos | -0.789 |
Cot | -0.777 |
闘 | -0.758 |
Token | This |
Token ID | 770 |
Feature activation | -3.43e-01 |
Pos logit contributions | |
Armstrong | +0.080 |
ppo | +0.073 |
pos | +0.071 |
Cot | +0.070 |
Recon | +0.069 |
Neg logit contributions | |
quartered | -0.095 |
uphem | -0.091 |
hern | -0.089 |
uum | -0.087 |
leys | -0.087 |
Token | often |
Token ID | 1690 |
Feature activation | +3.66e-01 |
Pos logit contributions | |
quartered | +0.805 |
uphem | +0.782 |
hern | +0.768 |
leys | +0.759 |
uum | +0.758 |
Neg logit contributions | |
Armstrong | -0.577 |
ppo | -0.529 |
pos | -0.507 |
Cot | -0.495 |
闘 | -0.483 |
Token | is |
Token ID | 318 |
Feature activation | +4.37e-01 |
Pos logit contributions | |
Armstrong | +0.571 |
ppo | +0.516 |
pos | +0.501 |
Cot | +0.483 |
闘 | +0.466 |
Neg logit contributions | |
quartered | -0.911 |
uphem | -0.886 |
hern | -0.868 |
uum | -0.858 |
leys | -0.858 |
Token | joined |
Token ID | 5399 |
Feature activation | -1.63e-01 |
Pos logit contributions | |
Armstrong | +0.178 |
ppo | +0.164 |
pos | +0.160 |
Cot | +0.155 |
Recon | +0.151 |
Neg logit contributions | |
quartered | -0.197 |
uphem | -0.191 |
hern | -0.185 |
leys | -0.184 |
uum | -0.183 |
Token | Panasonic |
Token ID | 46049 |
Feature activation | -3.57e-01 |
Pos logit contributions | |
quartered | +0.353 |
uphem | +0.341 |
hern | +0.335 |
uum | +0.330 |
leys | +0.330 |
Neg logit contributions | |
Armstrong | -0.306 |
ppo | -0.283 |
pos | -0.274 |
Cot | -0.267 |
闘 | -0.262 |
Token | to |
Token ID | 284 |
Feature activation | -1.39e-01 |
Pos logit contributions | |
quartered | +0.751 |
uphem | +0.727 |
hern | +0.712 |
uum | +0.702 |
leys | +0.701 |
Neg logit contributions | |
Armstrong | -0.688 |
ppo | -0.637 |
pos | -0.618 |
Cot | -0.602 |
闘 | -0.589 |
Token | help |
Token ID | 1037 |
Feature activation | -5.58e-01 |
Pos logit contributions | |
quartered | +0.307 |
uphem | +0.298 |
hern | +0.291 |
leys | +0.288 |
uum | +0.287 |
Neg logit contributions | |
Armstrong | -0.256 |
ppo | -0.235 |
pos | -0.228 |
Cot | -0.222 |
Recon | -0.215 |
Token | them |
Token ID | 606 |
Feature activation | +2.00e-01 |
Pos logit contributions | |
quartered | +1.140 |
uphem | +1.102 |
hern | +1.077 |
uum | +1.062 |
leys | +1.062 |
Neg logit contributions | |
Armstrong | -1.113 |
ppo | -1.031 |
pos | -1.002 |
Cot | -0.980 |
闘 | -0.957 |
Token | get |
Token ID | 651 |
Feature activation | +9.38e-02 |
Pos logit contributions | |
Armstrong | +0.398 |
ppo | +0.368 |
pos | +0.358 |
Cot | +0.350 |
Recon | +0.340 |
Neg logit contributions | |
quartered | -0.414 |
uphem | -0.401 |
hern | -0.391 |
leys | -0.386 |
uum | -0.384 |
Token | into |
Token ID | 656 |
Feature activation | -9.17e-01 |
Pos logit contributions | |
Armstrong | +0.176 |
ppo | +0.161 |
pos | +0.157 |
Cot | +0.154 |
Recon | +0.149 |
Neg logit contributions | |
quartered | -0.205 |
uphem | -0.199 |
hern | -0.193 |
leys | -0.192 |
uum | -0.191 |
Token | the |
Token ID | 262 |
Feature activation | -1.28e-01 |
Pos logit contributions | |
quartered | +1.931 |
uphem | +1.851 |
hern | +1.838 |
uum | +1.815 |
leys | +1.787 |
Neg logit contributions | |
Armstrong | -1.728 |
ppo | -1.620 |
pos | -1.550 |
闘 | -1.513 |
Cot | -1.509 |
Token | medical |
Token ID | 3315 |
Feature activation | -1.94e-01 |
Pos logit contributions | |
quartered | +0.227 |
uphem | +0.218 |
hern | +0.213 |
uum | +0.209 |
leys | +0.209 |
Neg logit contributions | |
Armstrong | -0.291 |
ppo | -0.272 |
pos | -0.265 |
Cot | -0.260 |
闘 | -0.256 |
Token | industry |
Token ID | 2831 |
Feature activation | -1.01e-01 |
Pos logit contributions | |
quartered | +0.419 |
uphem | +0.405 |
hern | +0.397 |
uum | +0.392 |
leys | +0.391 |
Neg logit contributions | |
Armstrong | -0.363 |
ppo | -0.335 |
pos | -0.324 |
Cot | -0.317 |
闘 | -0.311 |
Token | and |
Token ID | 290 |
Feature activation | -1.18e-01 |
Pos logit contributions | |
quartered | +0.229 |
uphem | +0.222 |
hern | +0.218 |
leys | +0.215 |
uum | +0.215 |
Neg logit contributions | |
Armstrong | -0.179 |
ppo | -0.164 |
pos | -0.159 |
Cot | -0.155 |
闘 | -0.151 |
Token | by |
Token ID | 416 |
Feature activation | +1.33e-01 |
Pos logit contributions | |
quartered | +0.831 |
uphem | +0.806 |
hern | +0.786 |
uum | +0.773 |
leys | +0.770 |
Neg logit contributions | |
Armstrong | -0.884 |
ppo | -0.838 |
pos | -0.797 |
Cot | -0.793 |
闘 | -0.771 |
Token | the |
Token ID | 262 |
Feature activation | -5.66e-02 |
Pos logit contributions | |
Armstrong | +0.247 |
ppo | +0.227 |
pos | +0.220 |
Cot | +0.215 |
闘 | +0.209 |
Neg logit contributions | |
quartered | -0.291 |
uphem | -0.282 |
hern | -0.276 |
leys | -0.273 |
uum | -0.273 |
Token | James |
Token ID | 3700 |
Feature activation | -1.82e-01 |
Pos logit contributions | |
quartered | +0.103 |
uphem | +0.099 |
hern | +0.096 |
leys | +0.095 |
uum | +0.095 |
Neg logit contributions | |
Armstrong | -0.126 |
ppo | -0.117 |
pos | -0.115 |
Cot | -0.113 |
Recon | -0.110 |
Token | Sim |
Token ID | 3184 |
Feature activation | -5.98e-01 |
Pos logit contributions | |
quartered | +0.372 |
uphem | +0.360 |
hern | +0.357 |
leys | +0.351 |
uum | +0.349 |
Neg logit contributions | |
Armstrong | -0.350 |
ppo | -0.329 |
pos | -0.321 |
Cot | -0.309 |
闘 | -0.307 |
Token | ons |
Token ID | 684 |
Feature activation | +6.33e-02 |
Pos logit contributions | |
quartered | +1.229 |
uphem | +1.187 |
hern | +1.169 |
leys | +1.158 |
uum | +1.143 |
Neg logit contributions | |
Armstrong | -1.174 |
ppo | -1.089 |
pos | -1.050 |
Cot | -1.035 |
闘 | -1.014 |
Token | and |
Token ID | 290 |
Feature activation | -7.68e-02 |
Pos logit contributions | |
Armstrong | +0.128 |
ppo | +0.117 |
pos | +0.114 |
Cot | +0.112 |
Recon | +0.109 |
Neg logit contributions | |
quartered | -0.129 |
uphem | -0.126 |
hern | -0.121 |
leys | -0.120 |
uum | -0.120 |
Token | Laurie |
Token ID | 40811 |
Feature activation | -5.19e-01 |
Pos logit contributions | |
quartered | +0.158 |
uphem | +0.153 |
hern | +0.149 |
uum | +0.147 |
leys | +0.147 |
Neg logit contributions | |
Armstrong | -0.153 |
ppo | -0.141 |
pos | -0.138 |
Cot | -0.135 |
Recon | -0.131 |
Token | Land |
Token ID | 6379 |
Feature activation | -1.36e-01 |
Pos logit contributions | |
quartered | +1.084 |
uphem | +1.058 |
hern | +1.044 |
leys | +1.030 |
uum | +1.029 |
Neg logit contributions | |
Armstrong | -0.972 |
ppo | -0.912 |
pos | -0.887 |
闘 | -0.852 |
Cot | -0.849 |
Token | e |
Token ID | 68 |
Feature activation | -2.15e-01 |
Pos logit contributions | |
quartered | +0.248 |
uphem | +0.239 |
hern | +0.233 |
leys | +0.228 |
uum | +0.228 |
Neg logit contributions | |
Armstrong | -0.302 |
ppo | -0.282 |
pos | -0.274 |
Cot | -0.269 |
Recon | -0.262 |
Token | au |
Token ID | 559 |
Feature activation | -1.61e-01 |
Pos logit contributions | |
quartered | +0.429 |
uphem | +0.416 |
hern | +0.405 |
leys | +0.399 |
uum | +0.399 |
Neg logit contributions | |
Armstrong | -0.439 |
ppo | -0.407 |
pos | -0.397 |
Cot | -0.387 |
Recon | -0.378 |
Token | foundations |
Token ID | 19369 |
Feature activation | -1.47e-01 |
Pos logit contributions | |
quartered | +0.326 |
uphem | +0.318 |
hern | +0.308 |
leys | +0.304 |
uum | +0.304 |
Neg logit contributions | |
Armstrong | -0.324 |
ppo | -0.297 |
pos | -0.291 |
Cot | -0.283 |
Recon | -0.277 |
Token | and |
Token ID | 290 |
Feature activation | -2.37e-02 |
Pos logit contributions | |
quartered | +0.234 |
uphem | +0.227 |
hern | +0.222 |
leys | +0.218 |
uum | +0.218 |
Neg logit contributions | |
Armstrong | -0.224 |
ppo | -0.208 |
pos | -0.201 |
Cot | -0.197 |
闘 | -0.193 |
Token | radio |
Token ID | 5243 |
Feature activation | -6.36e-01 |
Pos logit contributions | |
quartered | +0.048 |
uphem | +0.047 |
hern | +0.046 |
leys | +0.045 |
uum | +0.045 |
Neg logit contributions | |
Armstrong | -0.048 |
ppo | -0.044 |
pos | -0.043 |
Cot | -0.042 |
闘 | -0.041 |
Token | discipline |
Token ID | 12883 |
Feature activation | -4.22e-01 |
Pos logit contributions | |
quartered | +1.247 |
uphem | +1.207 |
hern | +1.184 |
leys | +1.162 |
uum | +1.153 |
Neg logit contributions | |
Armstrong | -1.304 |
ppo | -1.223 |
pos | -1.173 |
Cot | -1.160 |
闘 | -1.138 |
Token | , |
Token ID | 11 |
Feature activation | -1.76e-01 |
Pos logit contributions | |
quartered | +0.933 |
uphem | +0.906 |
hern | +0.887 |
leys | +0.874 |
uum | +0.874 |
Neg logit contributions | |
Armstrong | -0.769 |
ppo | -0.708 |
pos | -0.685 |
Cot | -0.669 |
闘 | -0.652 |
Token | out |
Token ID | 503 |
Feature activation | -4.11e-01 |
Pos logit contributions | |
quartered | +0.379 |
uphem | +0.367 |
hern | +0.359 |
leys | +0.354 |
uum | +0.354 |
Neg logit contributions | |
Armstrong | -0.334 |
ppo | -0.307 |
pos | -0.299 |
Cot | -0.292 |
闘 | -0.283 |
Token | the |
Token ID | 262 |
Feature activation | -2.03e-01 |
Pos logit contributions | |
quartered | +0.873 |
uphem | +0.845 |
hern | +0.827 |
leys | +0.816 |
uum | +0.815 |
Neg logit contributions | |
Armstrong | -0.788 |
ppo | -0.728 |
pos | -0.707 |
Cot | -0.690 |
闘 | -0.673 |
Token | window |
Token ID | 4324 |
Feature activation | -2.09e-01 |
Pos logit contributions | |
quartered | +0.428 |
uphem | +0.415 |
hern | +0.405 |
leys | +0.400 |
uum | +0.399 |
Neg logit contributions | |
Armstrong | -0.394 |
ppo | -0.364 |
pos | -0.354 |
Cot | -0.345 |
闘 | -0.336 |
Token | ... |
Token ID | 986 |
Feature activation | -1.22e-01 |
Pos logit contributions | |
quartered | +0.463 |
uphem | +0.449 |
hern | +0.440 |
leys | +0.434 |
uum | +0.434 |
Neg logit contributions | |
Armstrong | -0.380 |
ppo | -0.350 |
pos | -0.339 |
Cot | -0.330 |
闘 | -0.322 |
Token | no |
Token ID | 3919 |
Feature activation | -6.35e-01 |
Pos logit contributions | |
quartered | +0.256 |
uphem | +0.248 |
hern | +0.242 |
leys | +0.239 |
uum | +0.239 |
Neg logit contributions | |
Armstrong | -0.238 |
ppo | -0.220 |
pos | -0.214 |
Cot | -0.209 |
闘 | -0.203 |
Token | call |
Token ID | 869 |
Feature activation | -2.64e-01 |
Pos logit contributions | |
quartered | +1.312 |
uphem | +1.274 |
hern | +1.244 |
leys | +1.228 |
uum | +1.225 |
Neg logit contributions | |
Armstrong | -1.238 |
ppo | -1.153 |
pos | -1.113 |
Cot | -1.093 |
闘 | -1.074 |
Token | sign |
Token ID | 1051 |
Feature activation | -3.71e-01 |
Pos logit contributions | |
quartered | +0.570 |
uphem | +0.554 |
hern | +0.541 |
leys | +0.536 |
uum | +0.535 |
Neg logit contributions | |
Armstrong | -0.490 |
ppo | -0.453 |
pos | -0.439 |
Cot | -0.429 |
闘 | -0.419 |
Token | sey |
Token ID | 4397 |
Feature activation | -2.11e-01 |
Pos logit contributions | |
quartered | +1.388 |
uphem | +1.328 |
hern | +1.313 |
leys | +1.299 |
uum | +1.281 |
Neg logit contributions | |
Armstrong | -2.422 |
ppo | -2.299 |
pos | -2.263 |
闘 | -2.234 |
Cot | -2.203 |
Token | Gram |
Token ID | 20159 |
Feature activation | -6.97e-01 |
Pos logit contributions | |
quartered | +0.446 |
uphem | +0.434 |
hern | +0.426 |
leys | +0.420 |
uum | +0.419 |
Neg logit contributions | |
Armstrong | -0.398 |
ppo | -0.371 |
pos | -0.359 |
Cot | -0.348 |
闘 | -0.345 |
Token | mer |
Token ID | 647 |
Feature activation | -4.17e-01 |
Pos logit contributions | |
quartered | +1.314 |
uphem | +1.262 |
hern | +1.242 |
leys | +1.230 |
uum | +1.220 |
Neg logit contributions | |
Armstrong | -1.481 |
ppo | -1.385 |
pos | -1.356 |
Cot | -1.329 |
闘 | -1.322 |
Token |
|
Token ID | 198 |
Feature activation | -1.01e-02 |
Pos logit contributions | |
quartered | +0.880 |
uphem | +0.853 |
hern | +0.839 |
leys | +0.836 |
uum | +0.833 |
Neg logit contributions | |
Armstrong | -0.773 |
ppo | -0.729 |
pos | -0.702 |
闘 | -0.688 |
Cot | -0.686 |
Token |
|
Token ID | 198 |
Feature activation | +1.13e-01 |
Pos logit contributions | |
quartered | +0.025 |
uphem | +0.024 |
hern | +0.023 |
leys | +0.023 |
uum | +0.023 |
Neg logit contributions | |
Armstrong | -0.017 |
ppo | -0.015 |
pos | -0.015 |
Cot | -0.014 |
Recon | -0.014 |
Token | 1992 |
Token ID | 23847 |
Feature activation | -4.49e-01 |
Pos logit contributions | |
Armstrong | +0.217 |
ppo | +0.200 |
pos | +0.195 |
Cot | +0.192 |
闘 | +0.188 |
Neg logit contributions | |
quartered | -0.236 |
uphem | -0.229 |
hern | -0.225 |
leys | -0.224 |
uum | -0.223 |
Token | B |
Token ID | 347 |
Feature activation | -3.76e-01 |
Pos logit contributions | |
quartered | +0.902 |
uphem | +0.874 |
hern | +0.854 |
leys | +0.842 |
uum | +0.841 |
Neg logit contributions | |
Armstrong | -0.905 |
ppo | -0.843 |
pos | -0.818 |
Cot | -0.799 |
闘 | -0.784 |
Token | urt |
Token ID | 3325 |
Feature activation | +4.55e-01 |
Pos logit contributions | |
quartered | +0.672 |
uphem | +0.650 |
hern | +0.635 |
leys | +0.625 |
uum | +0.624 |
Neg logit contributions | |
Armstrong | -0.839 |
ppo | -0.784 |
pos | -0.766 |
Cot | -0.751 |
闘 | -0.739 |
Token | Reynolds |
Token ID | 21995 |
Feature activation | +3.20e-03 |
Pos logit contributions | |
Armstrong | +0.848 |
ppo | +0.756 |
pos | +0.736 |
Cot | +0.726 |
Recon | +0.714 |
Neg logit contributions | |
quartered | -1.045 |
uphem | -1.001 |
hern | -0.953 |
uum | -0.952 |
leys | -0.946 |
Token |
|
Token ID | 198 |
Feature activation | +6.67e-02 |
Pos logit contributions | |
Armstrong | +0.006 |
ppo | +0.006 |
pos | +0.005 |
闘 | +0.005 |
Cot | +0.005 |
Neg logit contributions | |
quartered | -0.007 |
uphem | -0.007 |
hern | -0.006 |
leys | -0.006 |
uum | -0.006 |
Token |
|
Token ID | 198 |
Feature activation | +1.16e-01 |
Pos logit contributions | |
Armstrong | +0.110 |
ppo | +0.099 |
pos | +0.097 |
Cot | +0.093 |
Recon | +0.091 |
Neg logit contributions | |
quartered | -0.167 |
uphem | -0.159 |
hern | -0.154 |
leys | -0.153 |
uum | -0.152 |
Token | badass |
Token ID | 37289 |
Feature activation | -3.10e-01 |
Pos logit contributions | |
Armstrong | +0.725 |
ppo | +0.666 |
pos | +0.647 |
Cot | +0.631 |
闘 | +0.613 |
Neg logit contributions | |
quartered | -0.862 |
uphem | -0.835 |
hern | -0.817 |
uum | -0.807 |
leys | -0.806 |
Token | ery |
Token ID | 1924 |
Feature activation | -4.53e-01 |
Pos logit contributions | |
quartered | +0.489 |
uphem | +0.469 |
hern | +0.457 |
leys | +0.449 |
uum | +0.447 |
Neg logit contributions | |
Armstrong | -0.760 |
ppo | -0.718 |
pos | -0.698 |
Cot | -0.686 |
闘 | -0.675 |
Token | , |
Token ID | 11 |
Feature activation | +1.57e-01 |
Pos logit contributions | |
quartered | +0.987 |
uphem | +0.958 |
hern | +0.945 |
leys | +0.930 |
uum | +0.927 |
Neg logit contributions | |
Armstrong | -0.828 |
ppo | -0.767 |
pos | -0.733 |
Cot | -0.721 |
闘 | -0.707 |
Token | and |
Token ID | 290 |
Feature activation | +1.66e-01 |
Pos logit contributions | |
Armstrong | +0.293 |
ppo | +0.271 |
pos | +0.261 |
Cot | +0.255 |
闘 | +0.250 |
Neg logit contributions | |
quartered | -0.339 |
uphem | -0.329 |
hern | -0.323 |
leys | -0.318 |
uum | -0.317 |
Token | one |
Token ID | 530 |
Feature activation | +2.06e-01 |
Pos logit contributions | |
Armstrong | +0.311 |
ppo | +0.287 |
pos | +0.278 |
Cot | +0.272 |
闘 | +0.265 |
Neg logit contributions | |
quartered | -0.357 |
uphem | -0.346 |
hern | -0.339 |
leys | -0.334 |
uum | -0.334 |
Token | hell |
Token ID | 5968 |
Feature activation | -1.29e+00 |
Pos logit contributions | |
Armstrong | +0.419 |
ppo | +0.389 |
pos | +0.379 |
Cot | +0.370 |
闘 | +0.361 |
Neg logit contributions | |
quartered | -0.412 |
uphem | -0.398 |
hern | -0.389 |
leys | -0.383 |
uum | -0.383 |
Token | of |
Token ID | 286 |
Feature activation | -3.32e-01 |
Pos logit contributions | |
quartered | +2.405 |
uphem | +2.311 |
uum | +2.273 |
leys | +2.252 |
hern | +2.249 |
Neg logit contributions | |
Armstrong | -2.733 |
ppo | -2.541 |
Cot | -2.463 |
pos | -2.435 |
Recon | -2.406 |
Token | an |
Token ID | 281 |
Feature activation | +3.03e-01 |
Pos logit contributions | |
quartered | +0.696 |
uphem | +0.674 |
hern | +0.659 |
leys | +0.649 |
uum | +0.649 |
Neg logit contributions | |
Armstrong | -0.645 |
ppo | -0.597 |
pos | -0.578 |
Cot | -0.566 |
闘 | -0.553 |
Token | epic |
Token ID | 12191 |
Feature activation | -5.98e-02 |
Pos logit contributions | |
Armstrong | +0.584 |
ppo | +0.533 |
pos | +0.524 |
Cot | +0.509 |
Recon | +0.495 |
Neg logit contributions | |
quartered | -0.644 |
uphem | -0.627 |
hern | -0.610 |
uum | -0.602 |
leys | -0.602 |
Token | superhero |
Token ID | 17343 |
Feature activation | +2.19e-01 |
Pos logit contributions | |
quartered | +0.126 |
uphem | +0.122 |
hern | +0.120 |
leys | +0.118 |
uum | +0.118 |
Neg logit contributions | |
Armstrong | -0.115 |
ppo | -0.106 |
pos | -0.103 |
Cot | -0.101 |
闘 | -0.099 |
Token | crossover |
Token ID | 27668 |
Feature activation | -2.39e-01 |
Pos logit contributions | |
Armstrong | +0.461 |
ppo | +0.428 |
pos | +0.418 |
Cot | +0.409 |
闘 | +0.398 |
Neg logit contributions | |
quartered | -0.426 |
uphem | -0.411 |
hern | -0.400 |
leys | -0.395 |
uum | -0.394 |
Token | some |
Token ID | 617 |
Feature activation | -6.39e-01 |
Pos logit contributions | |
Armstrong | +0.414 |
ppo | +0.377 |
pos | +0.370 |
Cot | +0.361 |
Recon | +0.350 |
Neg logit contributions | |
quartered | -0.502 |
uphem | -0.489 |
hern | -0.479 |
leys | -0.470 |
uum | -0.470 |
Token | notable |
Token ID | 12411 |
Feature activation | +4.29e-01 |
Pos logit contributions | |
quartered | +1.278 |
uphem | +1.235 |
hern | +1.204 |
leys | +1.192 |
uum | +1.190 |
Neg logit contributions | |
Armstrong | -1.294 |
ppo | -1.208 |
pos | -1.159 |
Cot | -1.141 |
闘 | -1.123 |
Token | exceptions |
Token ID | 13269 |
Feature activation | -7.20e-01 |
Pos logit contributions | |
Armstrong | +0.849 |
ppo | +0.778 |
pos | +0.771 |
Cot | +0.748 |
Recon | +0.726 |
Neg logit contributions | |
quartered | -0.892 |
uphem | -0.862 |
hern | -0.842 |
uum | -0.830 |
leys | -0.825 |
Token | , |
Token ID | 11 |
Feature activation | -2.54e-01 |
Pos logit contributions | |
quartered | +1.559 |
uphem | +1.514 |
hern | +1.492 |
uum | +1.464 |
leys | +1.461 |
Neg logit contributions | |
Armstrong | -1.308 |
ppo | -1.210 |
pos | -1.164 |
Cot | -1.143 |
闘 | -1.126 |
Token | public |
Token ID | 1171 |
Feature activation | +9.93e-01 |
Pos logit contributions | |
quartered | +0.552 |
uphem | +0.535 |
hern | +0.524 |
leys | +0.517 |
uum | +0.517 |
Neg logit contributions | |
Armstrong | -0.473 |
ppo | -0.436 |
pos | -0.423 |
Cot | -0.413 |
闘 | -0.402 |
Token | reaction |
Token ID | 6317 |
Feature activation | -1.55e+00 |
Pos logit contributions | |
Armstrong | +2.119 |
ppo | +1.953 |
pos | +1.946 |
Cot | +1.891 |
Recon | +1.831 |
Neg logit contributions | |
quartered | -1.892 |
uphem | -1.836 |
hern | -1.778 |
uum | -1.743 |
leys | -1.734 |
Token | was |
Token ID | 373 |
Feature activation | -5.59e-02 |
Pos logit contributions | |
quartered | +3.104 |
uphem | +2.971 |
hern | +2.926 |
leys | +2.874 |
uum | +2.873 |
Neg logit contributions | |
Armstrong | -3.074 |
ppo | -2.870 |
Cot | -2.746 |
闘 | -2.712 |
pos | -2.709 |
Token | swift |
Token ID | 14622 |
Feature activation | -6.79e-01 |
Pos logit contributions | |
quartered | +0.116 |
uphem | +0.112 |
hern | +0.109 |
uum | +0.108 |
leys | +0.108 |
Neg logit contributions | |
Armstrong | -0.111 |
ppo | -0.102 |
pos | -0.100 |
Cot | -0.097 |
闘 | -0.094 |
Token | and |
Token ID | 290 |
Feature activation | -1.03e-01 |
Pos logit contributions | |
quartered | +1.444 |
uphem | +1.400 |
hern | +1.373 |
leys | +1.361 |
uum | +1.356 |
Neg logit contributions | |
Armstrong | -1.265 |
ppo | -1.178 |
pos | -1.129 |
Cot | -1.121 |
闘 | -1.099 |
Token | harsh |
Token ID | 11859 |
Feature activation | -1.78e-01 |
Pos logit contributions | |
quartered | +0.205 |
uphem | +0.199 |
hern | +0.194 |
leys | +0.191 |
uum | +0.191 |
Neg logit contributions | |
Armstrong | -0.209 |
ppo | -0.194 |
pos | -0.188 |
Cot | -0.185 |
闘 | -0.181 |
Token | . |
Token ID | 13 |
Feature activation | -1.18e-01 |
Pos logit contributions | |
quartered | +0.393 |
uphem | +0.381 |
hern | +0.372 |
leys | +0.368 |
uum | +0.368 |
Neg logit contributions | |
Armstrong | -0.325 |
ppo | -0.301 |
pos | -0.290 |
Cot | -0.284 |
闘 | -0.279 |
Token | in |
Token ID | 287 |
Feature activation | +6.64e-02 |
Pos logit contributions | |
quartered | +1.370 |
uphem | +1.329 |
hern | +1.304 |
leys | +1.285 |
uum | +1.280 |
Neg logit contributions | |
Armstrong | -1.159 |
ppo | -1.074 |
pos | -1.031 |
Cot | -1.009 |
闘 | -0.993 |
Token | his |
Token ID | 465 |
Feature activation | +4.70e-02 |
Pos logit contributions | |
Armstrong | +0.131 |
ppo | +0.121 |
pos | +0.118 |
Cot | +0.115 |
Recon | +0.112 |
Neg logit contributions | |
quartered | -0.138 |
uphem | -0.133 |
hern | -0.130 |
uum | -0.128 |
leys | -0.128 |
Token | individual |
Token ID | 1981 |
Feature activation | +2.45e-02 |
Pos logit contributions | |
Armstrong | +0.094 |
ppo | +0.086 |
pos | +0.085 |
Cot | +0.082 |
Recon | +0.080 |
Neg logit contributions | |
quartered | -0.097 |
uphem | -0.094 |
hern | -0.092 |
uum | -0.090 |
leys | -0.090 |
Token | capacity |
Token ID | 5339 |
Feature activation | +1.17e-01 |
Pos logit contributions | |
Armstrong | +0.043 |
ppo | +0.039 |
pos | +0.038 |
Cot | +0.037 |
Recon | +0.036 |
Neg logit contributions | |
quartered | -0.056 |
uphem | -0.055 |
hern | -0.054 |
uum | -0.053 |
leys | -0.053 |
Token | whole |
Token ID | 2187 |
Feature activation | -1.43e+00 |
Pos logit contributions | |
Armstrong | +0.208 |
ppo | +0.189 |
pos | +0.186 |
Cot | +0.180 |
Recon | +0.174 |
Neg logit contributions | |
quartered | -0.266 |
uphem | -0.259 |
hern | -0.252 |
uum | -0.250 |
leys | -0.249 |
Token | living |
Token ID | 2877 |
Feature activation | -1.60e+00 |
Pos logit contributions | |
quartered | +3.150 |
hern | +2.993 |
uphem | +2.991 |
leys | +2.968 |
uum | +2.922 |
Neg logit contributions | |
Armstrong | -2.515 |
ppo | -2.359 |
Cot | -2.218 |
pos | -2.209 |
闘 | -2.205 |
Token | in |
Token ID | 287 |
Feature activation | -2.65e-01 |
Pos logit contributions | |
quartered | +3.345 |
uphem | +3.251 |
hern | +3.199 |
leys | +3.162 |
uum | +3.113 |
Neg logit contributions | |
Armstrong | -3.022 |
ppo | -2.857 |
pos | -2.681 |
闘 | -2.651 |
Cot | -2.648 |
Token | London |
Token ID | 3576 |
Feature activation | -1.91e-01 |
Pos logit contributions | |
quartered | +0.516 |
uphem | +0.499 |
hern | +0.488 |
leys | +0.481 |
uum | +0.478 |
Neg logit contributions | |
Armstrong | -0.551 |
ppo | -0.513 |
pos | -0.497 |
Cot | -0.487 |
闘 | -0.480 |
Token | for |
Token ID | 329 |
Feature activation | -3.28e-01 |
Pos logit contributions | |
quartered | +0.426 |
uphem | +0.414 |
hern | +0.405 |
uum | +0.400 |
leys | +0.398 |
Neg logit contributions | |
Armstrong | -0.345 |
ppo | -0.316 |
pos | -0.309 |
Cot | -0.300 |
Recon | -0.290 |
Token | years |
Token ID | 812 |
Feature activation | -5.82e-01 |
Pos logit contributions | |
quartered | +0.713 |
uphem | +0.691 |
hern | +0.675 |
uum | +0.668 |
leys | +0.666 |
Neg logit contributions | |
Armstrong | -0.614 |
ppo | -0.562 |
pos | -0.550 |
Cot | -0.535 |
Recon | -0.520 |
Token | . |
Token ID | 13 |
Feature activation | -1.54e-02 |
Pos logit contributions | |
quartered | +1.307 |
uphem | +1.268 |
hern | +1.244 |
leys | +1.228 |
uum | +1.226 |
Neg logit contributions | |
Armstrong | -1.040 |
ppo | -0.956 |
pos | -0.923 |
Cot | -0.901 |
闘 | -0.879 |
Token |
|
Token ID | 198 |
Feature activation | -1.65e-01 |
Pos logit contributions | |
quartered | +0.693 |
uphem | +0.662 |
hern | +0.645 |
leys | +0.637 |
uum | +0.637 |
Neg logit contributions | |
Armstrong | -0.479 |
ppo | -0.435 |
pos | -0.425 |
Cot | -0.408 |
Recon | -0.397 |
Token | -- |
Token ID | 438 |
Feature activation | -8.57e-01 |
Pos logit contributions | |
quartered | +0.334 |
uphem | +0.324 |
hern | +0.317 |
leys | +0.312 |
uum | +0.312 |
Neg logit contributions | |
Armstrong | -0.332 |
ppo | -0.308 |
pos | -0.300 |
Cot | -0.293 |
闘 | -0.287 |
Token |
|
Token ID | 198 |
Feature activation | +1.42e-01 |
Pos logit contributions | |
quartered | +1.821 |
uphem | +1.785 |
hern | +1.756 |
uum | +1.722 |
leys | +1.716 |
Neg logit contributions | |
Armstrong | -1.563 |
ppo | -1.467 |
pos | -1.404 |
闘 | -1.374 |
Cot | -1.366 |
Token |
|
Token ID | 198 |
Feature activation | +1.42e-01 |
Pos logit contributions | |
Armstrong | +0.231 |
ppo | +0.209 |
pos | +0.205 |
Cot | +0.195 |
Recon | +0.191 |
Neg logit contributions | |
quartered | -0.360 |
uphem | -0.341 |
hern | -0.331 |
uum | -0.328 |
leys | -0.328 |
Token | You |
Token ID | 1639 |
Feature activation | -9.16e-01 |
Pos logit contributions | |
Armstrong | +0.275 |
ppo | +0.255 |
pos | +0.247 |
Cot | +0.242 |
闘 | +0.236 |
Neg logit contributions | |
quartered | -0.297 |
uphem | -0.288 |
hern | -0.281 |
leys | -0.278 |
uum | -0.277 |
Token | might |
Token ID | 1244 |
Feature activation | -1.18e+00 |
Pos logit contributions | |
quartered | +2.437 |
uphem | +2.378 |
hern | +2.336 |
leys | +2.314 |
uum | +2.313 |
Neg logit contributions | |
Armstrong | -1.247 |
ppo | -1.109 |
pos | -1.062 |
Cot | -1.036 |
闘 | -0.999 |
Token | also |
Token ID | 635 |
Feature activation | -4.62e-01 |
Pos logit contributions | |
quartered | +2.225 |
uphem | +2.157 |
hern | +2.120 |
uum | +2.085 |
leys | +2.077 |
Neg logit contributions | |
Armstrong | -2.482 |
ppo | -2.322 |
Cot | -2.218 |
pos | -2.203 |
闘 | -2.197 |
Token | like |
Token ID | 588 |
Feature activation | -2.47e-01 |
Pos logit contributions | |
quartered | +1.013 |
uphem | +0.980 |
hern | +0.971 |
uum | +0.953 |
leys | +0.950 |
Neg logit contributions | |
Armstrong | -0.840 |
ppo | -0.775 |
pos | -0.738 |
Cot | -0.735 |
闘 | -0.722 |
Token | : |
Token ID | 25 |
Feature activation | -5.13e-01 |
Pos logit contributions | |
quartered | +0.527 |
uphem | +0.510 |
hern | +0.502 |
uum | +0.496 |
leys | +0.492 |
Neg logit contributions | |
Armstrong | -0.462 |
ppo | -0.432 |
pos | -0.412 |
Cot | -0.403 |
闘 | -0.401 |
Token |
|
Token ID | 198 |
Feature activation | -4.34e-01 |
Pos logit contributions | |
quartered | +1.143 |
uphem | +1.116 |
hern | +1.098 |
uum | +1.084 |
leys | +1.081 |
Neg logit contributions | |
Armstrong | -0.899 |
ppo | -0.835 |
pos | -0.797 |
Cot | -0.773 |
闘 | -0.772 |
Token |
|
Token ID | 198 |
Feature activation | -3.50e-01 |
Pos logit contributions | |
quartered | +1.030 |
uphem | +0.989 |
hern | +0.965 |
leys | +0.954 |
uum | +0.953 |
Neg logit contributions | |
Armstrong | -0.745 |
ppo | -0.679 |
pos | -0.660 |
Cot | -0.638 |
Recon | -0.620 |
Token | in |
Token ID | 287 |
Feature activation | -3.04e-02 |
Pos logit contributions | |
quartered | +0.680 |
uphem | +0.660 |
hern | +0.646 |
leys | +0.638 |
uum | +0.637 |
Neg logit contributions | |
Armstrong | -0.548 |
ppo | -0.503 |
pos | -0.487 |
Cot | -0.475 |
闘 | -0.462 |
Token | the |
Token ID | 262 |
Feature activation | +1.87e-01 |
Pos logit contributions | |
quartered | +0.062 |
uphem | +0.060 |
hern | +0.059 |
uum | +0.058 |
leys | +0.058 |
Neg logit contributions | |
Armstrong | -0.061 |
ppo | -0.056 |
pos | -0.055 |
Cot | -0.053 |
闘 | -0.052 |
Token | deciding |
Token ID | 14615 |
Feature activation | +1.96e-01 |
Pos logit contributions | |
Armstrong | +0.380 |
ppo | +0.352 |
pos | +0.343 |
Cot | +0.335 |
闘 | +0.326 |
Neg logit contributions | |
quartered | -0.377 |
uphem | -0.365 |
hern | -0.356 |
leys | -0.351 |
uum | -0.351 |
Token | game |
Token ID | 983 |
Feature activation | -4.76e-01 |
Pos logit contributions | |
Armstrong | +0.339 |
ppo | +0.311 |
pos | +0.300 |
Cot | +0.292 |
闘 | +0.285 |
Neg logit contributions | |
quartered | -0.450 |
uphem | -0.437 |
hern | -0.429 |
uum | -0.424 |
leys | -0.424 |
Token | . |
Token ID | 13 |
Feature activation | -2.15e-02 |
Pos logit contributions | |
quartered | +1.038 |
uphem | +1.008 |
hern | +0.987 |
uum | +0.974 |
leys | +0.974 |
Neg logit contributions | |
Armstrong | -0.877 |
ppo | -0.810 |
pos | -0.781 |
Cot | -0.766 |
闘 | -0.746 |
Token | Still |
Token ID | 7831 |
Feature activation | -1.17e+00 |
Pos logit contributions | |
quartered | +0.046 |
uphem | +0.045 |
hern | +0.043 |
leys | +0.043 |
uum | +0.043 |
Neg logit contributions | |
Armstrong | -0.042 |
ppo | -0.038 |
pos | -0.037 |
Cot | -0.036 |
Recon | -0.035 |
Token | , |
Token ID | 11 |
Feature activation | -4.41e-02 |
Pos logit contributions | |
quartered | +2.516 |
hern | +2.402 |
uphem | +2.386 |
uum | +2.351 |
leys | +2.327 |
Neg logit contributions | |
Armstrong | -2.171 |
ppo | -2.010 |
pos | -1.933 |
Cot | -1.897 |
闘 | -1.867 |
Token | he |
Token ID | 339 |
Feature activation | +1.93e-02 |
Pos logit contributions | |
quartered | +0.099 |
uphem | +0.096 |
hern | +0.094 |
leys | +0.093 |
uum | +0.092 |
Neg logit contributions | |
Armstrong | -0.080 |
ppo | -0.073 |
pos | -0.071 |
Cot | -0.069 |
闘 | -0.067 |
Token | clearly |
Token ID | 4084 |
Feature activation | +1.80e-01 |
Pos logit contributions | |
Armstrong | +0.036 |
ppo | +0.033 |
pos | +0.032 |
Cot | +0.031 |
闘 | +0.030 |
Neg logit contributions | |
quartered | -0.042 |
uphem | -0.041 |
hern | -0.040 |
leys | -0.039 |
uum | -0.039 |
Token | cannot |
Token ID | 2314 |
Feature activation | +1.41e-01 |
Pos logit contributions | |
Armstrong | +0.331 |
ppo | +0.304 |
pos | +0.296 |
Cot | +0.288 |
闘 | +0.279 |
Neg logit contributions | |
quartered | -0.398 |
uphem | -0.386 |
hern | -0.377 |
leys | -0.372 |
uum | -0.371 |
Token | fill |
Token ID | 6070 |
Feature activation | +2.19e-01 |
Pos logit contributions | |
Armstrong | +0.260 |
ppo | +0.240 |
pos | +0.233 |
Cot | +0.227 |
闘 | +0.220 |
Neg logit contributions | |
quartered | -0.310 |
uphem | -0.300 |
hern | -0.294 |
leys | -0.290 |
uum | -0.290 |