Token | � |
Token ID | 241 |
Feature activation | +1.46e-01 |
Pos logit contributions | |
igsaw | +0.404 |
oak | +0.355 |
magazine | +0.338 |
newspaper | +0.335 |
pine | +0.321 |
Neg logit contributions | |
ered | -0.378 |
bled | -0.374 |
izes | -0.332 |
vowed | -0.328 |
hadn | -0.323 |
Token | Good |
Token ID | 10248 |
Feature activation | +6.13e-02 |
Pos logit contributions | |
igsaw | +0.256 |
oak | +0.224 |
newspaper | +0.210 |
pine | +0.206 |
ant | +0.205 |
Neg logit contributions | |
ered | -0.292 |
bled | -0.287 |
izes | -0.260 |
icked | -0.252 |
vowed | -0.251 |
Token | job |
Token ID | 1693 |
Feature activation | +2.73e-01 |
Pos logit contributions | |
igsaw | +0.085 |
oak | +0.070 |
magazine | +0.064 |
newspaper | +0.063 |
pine | +0.059 |
Neg logit contributions | |
ered | -0.149 |
bled | -0.149 |
izes | -0.134 |
vowed | -0.133 |
eking | -0.132 |
Token | , |
Token ID | 11 |
Feature activation | +2.55e-01 |
Pos logit contributions | |
igsaw | +0.343 |
oak | +0.264 |
newspaper | +0.247 |
magazine | +0.236 |
ant | +0.224 |
Neg logit contributions | |
bled | -0.697 |
ered | -0.661 |
izes | -0.620 |
vowed | -0.619 |
itates | -0.619 |
Token | mighty |
Token ID | 18680 |
Feature activation | +2.53e-01 |
Pos logit contributions | |
igsaw | +0.459 |
oak | +0.418 |
newspaper | +0.384 |
pine | +0.380 |
magazine | +0.378 |
Neg logit contributions | |
bled | -0.515 |
ered | -0.498 |
izes | -0.445 |
eking | -0.439 |
icked | -0.438 |
Token | ant |
Token ID | 1885 |
Feature activation | +3.89e-01 |
Pos logit contributions | |
igsaw | +0.353 |
oak | +0.292 |
magazine | +0.269 |
newspaper | +0.265 |
pine | +0.250 |
Neg logit contributions | |
ered | -0.606 |
bled | -0.603 |
izes | -0.547 |
vowed | -0.543 |
hadn | -0.537 |
Token | ! |
Token ID | 0 |
Feature activation | +1.92e-01 |
Pos logit contributions | |
igsaw | +0.564 |
oak | +0.530 |
ant | +0.480 |
pine | +0.467 |
newspaper | +0.465 |
Neg logit contributions | |
bled | -0.909 |
ered | -0.860 |
izes | -0.769 |
eking | -0.766 |
itates | -0.754 |
Token | â |
Token ID | 22940 |
Feature activation | +6.21e-02 |
Pos logit contributions | |
igsaw | +0.370 |
oak | +0.324 |
magazine | +0.299 |
newspaper | +0.293 |
pine | +0.292 |
Neg logit contributions | |
bled | -0.368 |
ered | -0.366 |
izes | -0.324 |
eking | -0.323 |
vowed | -0.311 |
Token | € |
Token ID | 26391 |
Feature activation | +1.49e-01 |
Pos logit contributions | |
igsaw | +0.118 |
oak | +0.101 |
magazine | +0.094 |
newspaper | +0.092 |
pine | +0.088 |
Neg logit contributions | |
bled | -0.119 |
ered | -0.118 |
vowed | -0.104 |
izes | -0.103 |
hadn | -0.103 |
Token | |
Token ID | 220 |
Feature activation | -8.06e-03 |
Pos logit contributions | |
igsaw | +0.279 |
oak | +0.242 |
magazine | +0.227 |
newspaper | +0.224 |
pine | +0.216 |
Neg logit contributions | |
ered | -0.290 |
bled | -0.290 |
izes | -0.255 |
vowed | -0.251 |
eking | -0.247 |
Token |
|
Token ID | 198 |
Feature activation | +1.02e-01 |
Pos logit contributions | |
ered | +0.018 |
bled | +0.018 |
vowed | +0.017 |
izes | +0.017 |
eking | +0.016 |
Neg logit contributions | |
igsaw | -0.013 |
oak | -0.010 |
magazine | -0.009 |
newspaper | -0.009 |
pine | -0.009 |
Token | to |
Token ID | 284 |
Feature activation | +2.70e-01 |
Pos logit contributions | |
ered | +0.019 |
bled | +0.018 |
hadn | +0.017 |
vowed | +0.016 |
izes | +0.016 |
Neg logit contributions | |
igsaw | -0.022 |
oak | -0.019 |
magazine | -0.018 |
newspaper | -0.018 |
ant | -0.018 |
Token | see |
Token ID | 766 |
Feature activation | +1.08e-02 |
Pos logit contributions | |
igsaw | +0.468 |
oak | +0.424 |
magazine | +0.389 |
newspaper | +0.384 |
pine | +0.372 |
Neg logit contributions | |
bled | -0.558 |
ered | -0.527 |
izes | -0.491 |
vowed | -0.480 |
hadn | -0.479 |
Token | a |
Token ID | 257 |
Feature activation | +2.45e-01 |
Pos logit contributions | |
igsaw | +0.019 |
oak | +0.017 |
magazine | +0.016 |
newspaper | +0.016 |
pine | +0.015 |
Neg logit contributions | |
bled | -0.022 |
ered | -0.021 |
izes | -0.019 |
vowed | -0.019 |
hadn | -0.019 |
Token | little |
Token ID | 1310 |
Feature activation | +3.69e-01 |
Pos logit contributions | |
igsaw | +0.340 |
oak | +0.271 |
magazine | +0.246 |
newspaper | +0.242 |
pine | +0.227 |
Neg logit contributions | |
ered | -0.575 |
bled | -0.567 |
izes | -0.526 |
hadn | -0.521 |
vowed | -0.521 |
Token | red |
Token ID | 2266 |
Feature activation | +3.37e-01 |
Pos logit contributions | |
igsaw | +0.445 |
oak | +0.319 |
magazine | +0.285 |
newspaper | +0.284 |
pine | +0.257 |
Neg logit contributions | |
ered | -0.920 |
bled | -0.890 |
izes | -0.846 |
vowed | -0.842 |
apologised | -0.837 |
Token | ant |
Token ID | 1885 |
Feature activation | +3.88e-01 |
Pos logit contributions | |
igsaw | +0.464 |
oak | +0.305 |
Mic | +0.278 |
newspaper | +0.268 |
magazine | +0.268 |
Neg logit contributions | |
ered | -0.806 |
bled | -0.788 |
vowed | -0.714 |
izes | -0.713 |
wouldn | -0.707 |
Token | . |
Token ID | 13 |
Feature activation | +2.04e-01 |
Pos logit contributions | |
igsaw | +0.545 |
oak | +0.526 |
magazine | +0.475 |
ant | +0.465 |
pine | +0.458 |
Neg logit contributions | |
bled | -0.912 |
ered | -0.868 |
izes | -0.785 |
vowed | -0.772 |
uffed | -0.768 |
Token | He |
Token ID | 679 |
Feature activation | -2.52e-02 |
Pos logit contributions | |
igsaw | +0.342 |
oak | +0.288 |
magazine | +0.273 |
newspaper | +0.269 |
pine | +0.253 |
Neg logit contributions | |
ered | -0.425 |
bled | -0.419 |
vowed | -0.381 |
hadn | -0.379 |
izes | -0.376 |
Token | sque |
Token ID | 35776 |
Feature activation | -1.40e-02 |
Pos logit contributions | |
bled | +0.049 |
eking | +0.041 |
ered | +0.039 |
bling | +0.038 |
vere | +0.038 |
Neg logit contributions | |
oak | -0.047 |
pine | -0.047 |
bull | -0.044 |
ant | -0.044 |
igsaw | -0.043 |
Token | aked |
Token ID | 4335 |
Feature activation | +1.96e-02 |
Pos logit contributions | |
bled | +0.023 |
ered | +0.022 |
hadn | +0.021 |
wouldn | +0.021 |
vowed | +0.020 |
Neg logit contributions | |
igsaw | -0.030 |
oak | -0.024 |
olive | -0.024 |
magazine | -0.024 |
ant | -0.023 |
Token | , |
Token ID | 11 |
Feature activation | +2.42e-01 |
Pos logit contributions | |
igsaw | +0.035 |
oak | +0.030 |
magazine | +0.027 |
newspaper | +0.027 |
pine | +0.026 |
Neg logit contributions | |
ered | -0.041 |
bled | -0.040 |
vowed | -0.035 |
apologised | -0.035 |
izes | -0.035 |
Token | He |
Token ID | 679 |
Feature activation | +4.60e-02 |
Pos logit contributions | |
igsaw | +0.181 |
oak | +0.158 |
magazine | +0.148 |
newspaper | +0.146 |
pine | +0.140 |
Neg logit contributions | |
ered | -0.205 |
bled | -0.203 |
vowed | -0.181 |
izes | -0.181 |
hadn | -0.177 |
Token | felt |
Token ID | 2936 |
Feature activation | -1.03e-01 |
Pos logit contributions | |
oak | +0.081 |
pine | +0.079 |
lion | +0.076 |
tree | +0.074 |
cane | +0.073 |
Neg logit contributions | |
bled | -0.092 |
bling | -0.076 |
eking | -0.075 |
vere | -0.071 |
uffed | -0.069 |
Token | like |
Token ID | 588 |
Feature activation | +5.71e-02 |
Pos logit contributions | |
bled | +0.224 |
ered | +0.222 |
izes | +0.208 |
uffed | +0.207 |
eking | +0.204 |
Neg logit contributions | |
igsaw | -0.124 |
oak | -0.122 |
ant | -0.122 |
olive | -0.115 |
pine | -0.113 |
Token | a |
Token ID | 257 |
Feature activation | +2.28e-01 |
Pos logit contributions | |
igsaw | +0.088 |
oak | +0.079 |
magazine | +0.073 |
newspaper | +0.073 |
pine | +0.071 |
Neg logit contributions | |
ered | -0.127 |
bled | -0.124 |
izes | -0.116 |
vowed | -0.111 |
eking | -0.109 |
Token | small |
Token ID | 1402 |
Feature activation | +2.91e-01 |
Pos logit contributions | |
igsaw | +0.348 |
oak | +0.279 |
magazine | +0.248 |
newspaper | +0.243 |
olive | +0.223 |
Neg logit contributions | |
bled | -0.481 |
ered | -0.478 |
hadn | -0.461 |
vowed | -0.456 |
izes | -0.449 |
Token | ant |
Token ID | 1885 |
Feature activation | +3.87e-01 |
Pos logit contributions | |
igsaw | +0.403 |
oak | +0.298 |
magazine | +0.274 |
newspaper | +0.270 |
Mic | +0.250 |
Neg logit contributions | |
ered | -0.663 |
bled | -0.631 |
vowed | -0.619 |
hadn | -0.612 |
apologised | -0.604 |
Token | compared |
Token ID | 3688 |
Feature activation | +7.42e-03 |
Pos logit contributions | |
igsaw | +0.552 |
oak | +0.474 |
magazine | +0.437 |
newspaper | +0.428 |
pine | +0.413 |
Neg logit contributions | |
bled | -0.910 |
ered | -0.909 |
izes | -0.817 |
vowed | -0.804 |
eking | -0.794 |
Token | to |
Token ID | 284 |
Feature activation | +2.22e-01 |
Pos logit contributions | |
igsaw | +0.012 |
oak | +0.009 |
magazine | +0.009 |
newspaper | +0.009 |
pine | +0.008 |
Neg logit contributions | |
ered | -0.017 |
bled | -0.016 |
vowed | -0.015 |
izes | -0.015 |
hadn | -0.015 |
Token | its |
Token ID | 663 |
Feature activation | +1.51e-01 |
Pos logit contributions | |
igsaw | +0.325 |
oak | +0.294 |
newspaper | +0.268 |
magazine | +0.266 |
pine | +0.264 |
Neg logit contributions | |
bled | -0.515 |
ered | -0.495 |
izes | -0.461 |
uffed | -0.449 |
vowed | -0.446 |
Token | greatness |
Token ID | 27951 |
Feature activation | +9.93e-02 |
Pos logit contributions | |
igsaw | +0.247 |
oak | +0.215 |
magazine | +0.199 |
newspaper | +0.197 |
pine | +0.190 |
Neg logit contributions | |
ered | -0.325 |
bled | -0.325 |
izes | -0.291 |
vowed | -0.287 |
hadn | -0.283 |
Token | . |
Token ID | 13 |
Feature activation | +1.84e-01 |
Pos logit contributions | |
igsaw | +0.128 |
oak | +0.106 |
magazine | +0.098 |
newspaper | +0.096 |
pine | +0.090 |
Neg logit contributions | |
bled | -0.251 |
ered | -0.245 |
izes | -0.224 |
vowed | -0.220 |
eking | -0.220 |
Token | , |
Token ID | 11 |
Feature activation | +2.86e-01 |
Pos logit contributions | |
igsaw | +0.037 |
oak | +0.033 |
newspaper | +0.031 |
magazine | +0.031 |
pine | +0.030 |
Neg logit contributions | |
ered | -0.031 |
bled | -0.031 |
hadn | -0.028 |
vowed | -0.028 |
apologised | -0.028 |
Token | there |
Token ID | 612 |
Feature activation | +1.70e-01 |
Pos logit contributions | |
igsaw | +0.405 |
oak | +0.387 |
newspaper | +0.354 |
magazine | +0.352 |
pine | +0.341 |
Neg logit contributions | |
bled | -0.656 |
ered | -0.654 |
izes | -0.586 |
vowed | -0.572 |
eking | -0.572 |
Token | was |
Token ID | 373 |
Feature activation | -7.80e-02 |
Pos logit contributions | |
igsaw | +0.271 |
oak | +0.267 |
newspaper | +0.263 |
magazine | +0.259 |
dog | +0.252 |
Neg logit contributions | |
bled | -0.332 |
eking | -0.308 |
ered | -0.296 |
izes | -0.286 |
uffed | -0.280 |
Token | a |
Token ID | 257 |
Feature activation | +2.32e-01 |
Pos logit contributions | |
ered | +0.183 |
bled | +0.183 |
izes | +0.171 |
eking | +0.166 |
vowed | +0.164 |
Neg logit contributions | |
igsaw | -0.112 |
oak | -0.095 |
magazine | -0.092 |
newspaper | -0.091 |
pine | -0.084 |
Token | little |
Token ID | 1310 |
Feature activation | +3.59e-01 |
Pos logit contributions | |
igsaw | +0.341 |
oak | +0.264 |
magazine | +0.241 |
newspaper | +0.237 |
pine | +0.220 |
Neg logit contributions | |
bled | -0.510 |
ered | -0.510 |
hadn | -0.482 |
izes | -0.475 |
vowed | -0.474 |
Token | ant |
Token ID | 1885 |
Feature activation | +3.84e-01 |
Pos logit contributions | |
igsaw | +0.489 |
oak | +0.347 |
newspaper | +0.332 |
magazine | +0.329 |
pine | +0.284 |
Neg logit contributions | |
ered | -0.812 |
bled | -0.794 |
izes | -0.757 |
hadn | -0.756 |
vowed | -0.756 |
Token | . |
Token ID | 13 |
Feature activation | +1.98e-01 |
Pos logit contributions | |
igsaw | +0.573 |
oak | +0.521 |
magazine | +0.473 |
newspaper | +0.464 |
pine | +0.458 |
Neg logit contributions | |
bled | -0.873 |
ered | -0.857 |
izes | -0.772 |
vowed | -0.757 |
eking | -0.747 |
Token | It |
Token ID | 632 |
Feature activation | -3.53e-02 |
Pos logit contributions | |
igsaw | +0.327 |
oak | +0.273 |
magazine | +0.261 |
newspaper | +0.258 |
pine | +0.239 |
Neg logit contributions | |
ered | -0.417 |
bled | -0.408 |
hadn | -0.372 |
vowed | -0.372 |
izes | -0.365 |
Token | was |
Token ID | 373 |
Feature activation | -1.09e-01 |
Pos logit contributions | |
bled | +0.072 |
ered | +0.063 |
eking | +0.062 |
izes | +0.061 |
vere | +0.057 |
Neg logit contributions | |
oak | -0.063 |
pine | -0.060 |
ant | -0.059 |
elephant | -0.058 |
cat | -0.057 |
Token | very |
Token ID | 845 |
Feature activation | +9.62e-02 |
Pos logit contributions | |
ered | +0.223 |
bled | +0.221 |
izes | +0.202 |
vowed | +0.197 |
eking | +0.193 |
Neg logit contributions | |
igsaw | -0.188 |
oak | -0.165 |
magazine | -0.155 |
newspaper | -0.153 |
pine | -0.149 |
Token | scared |
Token ID | 12008 |
Feature activation | +3.51e-02 |
Pos logit contributions | |
igsaw | +0.141 |
oak | +0.122 |
pine | +0.109 |
ant | +0.108 |
newspaper | +0.107 |
Neg logit contributions | |
ered | -0.226 |
bled | -0.226 |
vowed | -0.205 |
izes | -0.205 |
eking | -0.200 |
Token | , |
Token ID | 11 |
Feature activation | +2.71e-01 |
Pos logit contributions | |
igsaw | +0.159 |
magazine | +0.134 |
oak | +0.134 |
newspaper | +0.128 |
ant | +0.119 |
Neg logit contributions | |
ered | -0.236 |
bled | -0.232 |
vowed | -0.211 |
hadn | -0.209 |
apologised | -0.209 |
Token | he |
Token ID | 339 |
Feature activation | +2.66e-02 |
Pos logit contributions | |
igsaw | +0.357 |
oak | +0.348 |
magazine | +0.310 |
newspaper | +0.308 |
pine | +0.303 |
Neg logit contributions | |
ered | -0.646 |
bled | -0.636 |
izes | -0.584 |
icked | -0.567 |
vowed | -0.560 |
Token | saw |
Token ID | 2497 |
Feature activation | +7.01e-03 |
Pos logit contributions | |
oak | +0.046 |
pine | +0.044 |
lion | +0.043 |
dog | +0.043 |
ant | +0.043 |
Neg logit contributions | |
bled | -0.050 |
ered | -0.045 |
vere | -0.042 |
eking | -0.042 |
fed | -0.039 |
Token | a |
Token ID | 257 |
Feature activation | +2.23e-01 |
Pos logit contributions | |
igsaw | +0.011 |
oak | +0.010 |
magazine | +0.009 |
newspaper | +0.009 |
pine | +0.009 |
Neg logit contributions | |
ered | -0.015 |
bled | -0.015 |
izes | -0.014 |
vowed | -0.013 |
hadn | -0.013 |
Token | tiny |
Token ID | 7009 |
Feature activation | +2.74e-01 |
Pos logit contributions | |
igsaw | +0.340 |
oak | +0.257 |
magazine | +0.231 |
newspaper | +0.227 |
olive | +0.217 |
Neg logit contributions | |
bled | -0.467 |
ered | -0.464 |
hadn | -0.451 |
izes | -0.433 |
vowed | -0.428 |
Token | ant |
Token ID | 1885 |
Feature activation | +3.80e-01 |
Pos logit contributions | |
igsaw | +0.384 |
Mic | +0.262 |
newspaper | +0.261 |
magazine | +0.245 |
oak | +0.245 |
Neg logit contributions | |
ered | -0.559 |
bled | -0.523 |
hadn | -0.521 |
vowed | -0.507 |
apologised | -0.500 |
Token | working |
Token ID | 1762 |
Feature activation | +1.43e-03 |
Pos logit contributions | |
igsaw | +0.667 |
oak | +0.615 |
magazine | +0.568 |
newspaper | +0.557 |
pine | +0.552 |
Neg logit contributions | |
bled | -0.759 |
ered | -0.750 |
izes | -0.665 |
vowed | -0.657 |
apologised | -0.641 |
Token | hard |
Token ID | 1327 |
Feature activation | +1.66e-01 |
Pos logit contributions | |
igsaw | +0.003 |
oak | +0.002 |
magazine | +0.002 |
newspaper | +0.002 |
pine | +0.002 |
Neg logit contributions | |
ered | -0.003 |
bled | -0.003 |
izes | -0.002 |
vowed | -0.002 |
apologised | -0.002 |
Token | to |
Token ID | 284 |
Feature activation | +2.17e-01 |
Pos logit contributions | |
igsaw | +0.254 |
oak | +0.212 |
magazine | +0.199 |
newspaper | +0.195 |
pine | +0.184 |
Neg logit contributions | |
ered | -0.377 |
bled | -0.375 |
izes | -0.339 |
vowed | -0.337 |
hadn | -0.332 |
Token | carry |
Token ID | 3283 |
Feature activation | +2.79e-02 |
Pos logit contributions | |
igsaw | +0.444 |
oak | +0.401 |
magazine | +0.372 |
newspaper | +0.369 |
pine | +0.363 |
Neg logit contributions | |
bled | -0.382 |
ered | -0.374 |
izes | -0.334 |
vowed | -0.329 |
hadn | -0.323 |
Token | a |
Token ID | 257 |
Feature activation | +2.01e-01 |
Pos logit contributions | |
igsaw | +0.047 |
oak | +0.039 |
magazine | +0.037 |
newspaper | +0.037 |
bull | +0.034 |
Neg logit contributions | |
ered | -0.060 |
bled | -0.058 |
vowed | -0.052 |
izes | -0.052 |
hadn | -0.051 |
Token | said |
Token ID | 531 |
Feature activation | -2.56e-02 |
Pos logit contributions | |
oak | +0.293 |
igsaw | +0.284 |
pine | +0.273 |
newspaper | +0.265 |
elephant | +0.260 |
Neg logit contributions | |
bled | -0.524 |
ered | -0.478 |
izes | -0.455 |
vere | -0.445 |
eking | -0.443 |
Token | , |
Token ID | 11 |
Feature activation | +2.68e-01 |
Pos logit contributions | |
ered | +0.074 |
bled | +0.074 |
izes | +0.069 |
eking | +0.068 |
vowed | +0.068 |
Neg logit contributions | |
igsaw | -0.019 |
oak | -0.014 |
magazine | -0.012 |
newspaper | -0.011 |
pine | -0.011 |
Token | " |
Token ID | 366 |
Feature activation | +7.84e-02 |
Pos logit contributions | |
igsaw | +0.400 |
oak | +0.351 |
magazine | +0.328 |
newspaper | +0.326 |
bull | +0.313 |
Neg logit contributions | |
ered | -0.612 |
bled | -0.603 |
izes | -0.544 |
eking | -0.535 |
vowed | -0.534 |
Token | Hi |
Token ID | 17250 |
Feature activation | +2.81e-03 |
Pos logit contributions | |
igsaw | +0.169 |
oak | +0.151 |
magazine | +0.143 |
newspaper | +0.142 |
pine | +0.138 |
Neg logit contributions | |
ered | -0.130 |
bled | -0.127 |
izes | -0.110 |
vowed | -0.109 |
eking | -0.107 |
Token | , |
Token ID | 11 |
Feature activation | +2.48e-01 |
Pos logit contributions | |
igsaw | +0.005 |
oak | +0.004 |
magazine | +0.004 |
newspaper | +0.004 |
cane | +0.004 |
Neg logit contributions | |
ered | -0.005 |
ok | -0.005 |
bled | -0.005 |
okay | -0.005 |
alright | -0.005 |
Token | ant |
Token ID | 1885 |
Feature activation | +3.75e-01 |
Pos logit contributions | |
igsaw | +0.426 |
oak | +0.345 |
magazine | +0.332 |
newspaper | +0.324 |
cane | +0.302 |
Neg logit contributions | |
ered | -0.510 |
bled | -0.500 |
hadn | -0.463 |
izes | -0.450 |
vowed | -0.448 |
Token | ! |
Token ID | 0 |
Feature activation | +1.86e-01 |
Pos logit contributions | |
igsaw | +0.605 |
oak | +0.482 |
magazine | +0.462 |
newspaper | +0.453 |
pine | +0.414 |
Neg logit contributions | |
ered | -0.831 |
bled | -0.808 |
izes | -0.749 |
vowed | -0.747 |
hadn | -0.738 |
Token | Do |
Token ID | 2141 |
Feature activation | -9.51e-02 |
Pos logit contributions | |
igsaw | +0.376 |
oak | +0.328 |
magazine | +0.313 |
newspaper | +0.310 |
pine | +0.297 |
Neg logit contributions | |
ered | -0.329 |
bled | -0.325 |
izes | -0.286 |
vowed | -0.284 |
hadn | -0.281 |
Token | you |
Token ID | 345 |
Feature activation | -2.97e-02 |
Pos logit contributions | |
ered | +0.139 |
bled | +0.134 |
izes | +0.115 |
vowed | +0.113 |
hadn | +0.113 |
Neg logit contributions | |
igsaw | -0.227 |
oak | -0.198 |
magazine | -0.193 |
newspaper | -0.191 |
pine | -0.182 |
Token | need |
Token ID | 761 |
Feature activation | -2.36e-02 |
Pos logit contributions | |
bled | +0.068 |
ered | +0.059 |
izes | +0.056 |
uffed | +0.056 |
vowed | +0.054 |
Neg logit contributions | |
igsaw | -0.048 |
oak | -0.045 |
olive | -0.040 |
elephant | -0.039 |
magazine | -0.039 |
Token | help |
Token ID | 1037 |
Feature activation | +9.09e-02 |
Pos logit contributions | |
bled | +0.049 |
ered | +0.049 |
izes | +0.045 |
vowed | +0.044 |
hadn | +0.044 |
Neg logit contributions | |
igsaw | -0.039 |
oak | -0.034 |
magazine | -0.033 |
newspaper | -0.033 |
pine | -0.032 |
Token | blanket |
Token ID | 18447 |
Feature activation | +2.69e-01 |
Pos logit contributions | |
igsaw | +0.463 |
oak | +0.381 |
magazine | +0.350 |
newspaper | +0.343 |
pine | +0.321 |
Neg logit contributions | |
ered | -0.892 |
bled | -0.883 |
izes | -0.805 |
vowed | -0.801 |
hadn | -0.790 |
Token | . |
Token ID | 13 |
Feature activation | +2.20e-01 |
Pos logit contributions | |
igsaw | +0.244 |
oak | +0.244 |
magazine | +0.223 |
newspaper | +0.206 |
pine | +0.188 |
Neg logit contributions | |
ered | -0.701 |
bled | -0.697 |
izes | -0.643 |
eking | -0.640 |
eling | -0.637 |
Token | He |
Token ID | 679 |
Feature activation | -1.40e-02 |
Pos logit contributions | |
igsaw | +0.346 |
oak | +0.293 |
magazine | +0.273 |
newspaper | +0.267 |
pine | +0.258 |
Neg logit contributions | |
ered | -0.494 |
bled | -0.491 |
izes | -0.444 |
eking | -0.439 |
vowed | -0.432 |
Token | lifts |
Token ID | 27103 |
Feature activation | -3.01e-02 |
Pos logit contributions | |
bled | +0.028 |
vere | +0.023 |
eking | +0.023 |
bling | +0.022 |
itates | +0.022 |
Neg logit contributions | |
pine | -0.025 |
oak | -0.024 |
cane | -0.022 |
newspaper | -0.022 |
lion | -0.022 |
Token | the |
Token ID | 262 |
Feature activation | +3.15e-01 |
Pos logit contributions | |
ered | +0.059 |
bled | +0.056 |
vowed | +0.049 |
izes | +0.049 |
apologised | +0.047 |
Neg logit contributions | |
igsaw | -0.059 |
oak | -0.050 |
magazine | -0.047 |
newspaper | -0.046 |
dictionary | -0.044 |
Token | ostr |
Token ID | 42387 |
Feature activation | +3.75e-01 |
Pos logit contributions | |
igsaw | +0.387 |
oak | +0.320 |
magazine | +0.298 |
newspaper | +0.292 |
pine | +0.269 |
Neg logit contributions | |
ered | -0.808 |
bled | -0.800 |
izes | -0.731 |
vowed | -0.729 |
hadn | -0.718 |
Token | ich |
Token ID | 488 |
Feature activation | +2.02e-01 |
Pos logit contributions | |
igsaw | +0.589 |
oak | +0.495 |
magazine | +0.463 |
newspaper | +0.455 |
pine | +0.431 |
Neg logit contributions | |
ered | -0.839 |
bled | -0.831 |
izes | -0.751 |
vowed | -0.746 |
hadn | -0.736 |
Token | off |
Token ID | 572 |
Feature activation | +1.21e-01 |
Pos logit contributions | |
igsaw | +0.359 |
oak | +0.306 |
magazine | +0.289 |
newspaper | +0.286 |
pine | +0.272 |
Neg logit contributions | |
ered | -0.409 |
bled | -0.402 |
izes | -0.361 |
vowed | -0.358 |
hadn | -0.352 |
Token | the |
Token ID | 262 |
Feature activation | +2.96e-01 |
Pos logit contributions | |
igsaw | +0.198 |
oak | +0.158 |
newspaper | +0.149 |
magazine | +0.149 |
dictionary | +0.135 |
Neg logit contributions | |
ered | -0.267 |
bled | -0.255 |
vowed | -0.237 |
hadn | -0.230 |
izes | -0.230 |
Token | fence |
Token ID | 13990 |
Feature activation | +2.25e-01 |
Pos logit contributions | |
igsaw | +0.392 |
oak | +0.289 |
magazine | +0.267 |
newspaper | +0.256 |
olive | +0.237 |
Neg logit contributions | |
ered | -0.724 |
bled | -0.714 |
izes | -0.662 |
vowed | -0.657 |
hadn | -0.651 |
Token | . |
Token ID | 13 |
Feature activation | +1.88e-01 |
Pos logit contributions | |
igsaw | +0.291 |
oak | +0.242 |
magazine | +0.221 |
newspaper | +0.217 |
pine | +0.204 |
Neg logit contributions | |
ered | -0.559 |
bled | -0.558 |
izes | -0.509 |
vowed | -0.503 |
hadn | -0.497 |
Token | a |
Token ID | 257 |
Feature activation | +2.40e-01 |
Pos logit contributions | |
igsaw | +0.057 |
oak | +0.049 |
newspaper | +0.047 |
magazine | +0.047 |
bull | +0.044 |
Neg logit contributions | |
bled | -0.085 |
ered | -0.085 |
izes | -0.078 |
vowed | -0.077 |
hadn | -0.076 |
Token | big |
Token ID | 1263 |
Feature activation | +3.52e-01 |
Pos logit contributions | |
igsaw | +0.378 |
oak | +0.294 |
magazine | +0.259 |
newspaper | +0.253 |
olive | +0.248 |
Neg logit contributions | |
bled | -0.498 |
ered | -0.489 |
hadn | -0.469 |
izes | -0.462 |
vowed | -0.450 |
Token | net |
Token ID | 2010 |
Feature activation | +2.22e-01 |
Pos logit contributions | |
igsaw | +0.537 |
Mic | +0.380 |
oak | +0.349 |
magazine | +0.343 |
newspaper | +0.339 |
Neg logit contributions | |
ered | -0.715 |
bled | -0.684 |
izes | -0.652 |
vowed | -0.643 |
sorry | -0.638 |
Token | around |
Token ID | 1088 |
Feature activation | +1.17e-01 |
Pos logit contributions | |
igsaw | +0.285 |
oak | +0.275 |
magazine | +0.259 |
newspaper | +0.248 |
olive | +0.238 |
Neg logit contributions | |
ered | -0.509 |
bled | -0.502 |
vowed | -0.466 |
izes | -0.458 |
apologised | -0.449 |
Token | the |
Token ID | 262 |
Feature activation | +3.17e-01 |
Pos logit contributions | |
igsaw | +0.147 |
oak | +0.126 |
magazine | +0.116 |
newspaper | +0.114 |
ant | +0.109 |
Neg logit contributions | |
bled | -0.287 |
ered | -0.286 |
izes | -0.269 |
vowed | -0.259 |
eking | -0.258 |
Token | oct |
Token ID | 19318 |
Feature activation | +3.75e-01 |
Pos logit contributions | |
igsaw | +0.442 |
oak | +0.330 |
magazine | +0.296 |
newspaper | +0.289 |
pine | +0.284 |
Neg logit contributions | |
bled | -0.745 |
ered | -0.729 |
izes | -0.686 |
vowed | -0.676 |
hadn | -0.672 |
Token | opus |
Token ID | 25790 |
Feature activation | +2.20e-01 |
Pos logit contributions | |
igsaw | +0.531 |
oak | +0.461 |
newspaper | +0.415 |
magazine | +0.414 |
pine | +0.404 |
Neg logit contributions | |
bled | -0.869 |
ered | -0.859 |
izes | -0.794 |
eking | -0.772 |
hadn | -0.767 |
Token | . |
Token ID | 13 |
Feature activation | +1.95e-01 |
Pos logit contributions | |
igsaw | +0.249 |
oak | +0.223 |
magazine | +0.207 |
newspaper | +0.197 |
olive | +0.188 |
Neg logit contributions | |
bled | -0.569 |
ered | -0.561 |
izes | -0.510 |
vowed | -0.502 |
apologised | -0.493 |
Token | It |
Token ID | 632 |
Feature activation | -4.47e-02 |
Pos logit contributions | |
igsaw | +0.328 |
oak | +0.280 |
magazine | +0.263 |
newspaper | +0.259 |
pine | +0.246 |
Neg logit contributions | |
ered | -0.409 |
bled | -0.405 |
vowed | -0.363 |
izes | -0.362 |
hadn | -0.361 |
Token | was |
Token ID | 373 |
Feature activation | -1.15e-01 |
Pos logit contributions | |
bled | +0.095 |
ered | +0.088 |
eking | +0.085 |
izes | +0.082 |
vere | +0.081 |
Neg logit contributions | |
lion | -0.068 |
pine | -0.068 |
tiger | -0.068 |
rh | -0.067 |
oak | -0.067 |
Token | stuck |
Token ID | 7819 |
Feature activation | +6.92e-02 |
Pos logit contributions | |
ered | +0.255 |
bled | +0.248 |
izes | +0.232 |
vowed | +0.226 |
apologised | +0.220 |
Neg logit contributions | |
igsaw | -0.178 |
oak | -0.157 |
magazine | -0.148 |
newspaper | -0.146 |
pine | -0.142 |
Token | a |
Token ID | 257 |
Feature activation | +2.26e-01 |
Pos logit contributions | |
igsaw | +0.150 |
oak | +0.128 |
magazine | +0.121 |
newspaper | +0.119 |
pine | +0.113 |
Neg logit contributions | |
ered | -0.174 |
bled | -0.172 |
izes | -0.152 |
vowed | -0.152 |
hadn | -0.151 |
Token | time |
Token ID | 640 |
Feature activation | +2.52e-01 |
Pos logit contributions | |
igsaw | +0.443 |
oak | +0.399 |
magazine | +0.373 |
newspaper | +0.349 |
automobile | +0.341 |
Neg logit contributions | |
ered | -0.368 |
bled | -0.359 |
zed | -0.341 |
hadn | -0.332 |
eking | -0.332 |
Token | there |
Token ID | 612 |
Feature activation | +1.36e-01 |
Pos logit contributions | |
igsaw | +0.385 |
oak | +0.324 |
newspaper | +0.320 |
magazine | +0.320 |
cane | +0.281 |
Neg logit contributions | |
bled | -0.585 |
ered | -0.580 |
hadn | -0.516 |
eking | -0.508 |
icked | -0.506 |
Token | was |
Token ID | 373 |
Feature activation | -9.91e-02 |
Pos logit contributions | |
igsaw | +0.304 |
oak | +0.285 |
pine | +0.268 |
magazine | +0.267 |
newspaper | +0.266 |
Neg logit contributions | |
bled | -0.192 |
ered | -0.190 |
izes | -0.171 |
eking | -0.164 |
apologised | -0.161 |
Token | an |
Token ID | 281 |
Feature activation | +2.00e-01 |
Pos logit contributions | |
bled | +0.137 |
ered | +0.126 |
hes | +0.121 |
hadn | +0.109 |
icked | +0.108 |
Neg logit contributions | |
igsaw | -0.233 |
oak | -0.210 |
dictionary | -0.189 |
magazine | -0.187 |
newspaper | -0.184 |
Token | ant |
Token ID | 1885 |
Feature activation | +3.74e-01 |
Pos logit contributions | |
igsaw | +0.383 |
magazine | +0.297 |
newspaper | +0.279 |
ro | +0.268 |
cane | +0.268 |
Neg logit contributions | |
ered | -0.360 |
bled | -0.329 |
eking | -0.300 |
vowed | -0.295 |
izes | -0.289 |
Token | . |
Token ID | 13 |
Feature activation | +1.92e-01 |
Pos logit contributions | |
igsaw | +0.841 |
magazine | +0.636 |
newspaper | +0.629 |
oak | +0.628 |
gol | +0.580 |
Neg logit contributions | |
ered | -0.605 |
bled | -0.551 |
hadn | -0.550 |
vowed | -0.541 |
wouldn | -0.518 |
Token | The |
Token ID | 383 |
Feature activation | +1.87e-01 |
Pos logit contributions | |
igsaw | +0.355 |
magazine | +0.298 |
newspaper | +0.293 |
oak | +0.288 |
cane | +0.270 |
Neg logit contributions | |
ered | -0.343 |
hadn | -0.313 |
bled | -0.312 |
vowed | -0.293 |
icked | -0.291 |
Token | ant |
Token ID | 1885 |
Feature activation | +3.60e-01 |
Pos logit contributions | |
igsaw | +0.212 |
magazine | +0.141 |
newspaper | +0.113 |
automobile | +0.107 |
gol | +0.097 |
Neg logit contributions | |
bled | -0.465 |
ered | -0.465 |
wouldn | -0.431 |
hadn | -0.428 |
vowed | -0.423 |
Token | was |
Token ID | 373 |
Feature activation | -1.23e-01 |
Pos logit contributions | |
oak | +0.731 |
igsaw | +0.693 |
pine | +0.682 |
ant | +0.678 |
magazine | +0.652 |
Neg logit contributions | |
bled | -0.630 |
ered | -0.561 |
izes | -0.510 |
eking | -0.507 |
vere | -0.485 |
Token | purple |
Token ID | 14032 |
Feature activation | +2.79e-01 |
Pos logit contributions | |
ered | +0.200 |
bled | +0.197 |
izes | +0.173 |
vowed | +0.170 |
hadn | +0.166 |
Neg logit contributions | |
igsaw | -0.267 |
oak | -0.240 |
magazine | -0.228 |
newspaper | -0.225 |
pine | -0.220 |
Token | that |
Token ID | 326 |
Feature activation | +2.12e-01 |
Pos logit contributions | |
igsaw | +0.081 |
oak | +0.070 |
magazine | +0.065 |
newspaper | +0.064 |
pine | +0.061 |
Neg logit contributions | |
ered | -0.104 |
bled | -0.104 |
izes | -0.093 |
vowed | -0.092 |
hadn | -0.091 |
Token | day |
Token ID | 1110 |
Feature activation | +2.33e-01 |
Pos logit contributions | |
igsaw | +0.195 |
oak | +0.158 |
magazine | +0.136 |
newspaper | +0.132 |
ant | +0.124 |
Neg logit contributions | |
ered | -0.613 |
bled | -0.612 |
izes | -0.559 |
eking | -0.555 |
uffed | -0.549 |
Token | on |
Token ID | 319 |
Feature activation | +1.49e-01 |
Pos logit contributions | |
igsaw | +0.331 |
oak | +0.281 |
magazine | +0.270 |
newspaper | +0.254 |
pine | +0.250 |
Neg logit contributions | |
ered | -0.508 |
bled | -0.502 |
eking | -0.466 |
izes | -0.464 |
apologised | -0.457 |
Token | , |
Token ID | 11 |
Feature activation | +2.48e-01 |
Pos logit contributions | |
igsaw | +0.134 |
oak | +0.098 |
magazine | +0.083 |
newspaper | +0.079 |
pine | +0.075 |
Neg logit contributions | |
bled | -0.432 |
ered | -0.427 |
izes | -0.399 |
vowed | -0.394 |
eking | -0.393 |
Token | the |
Token ID | 262 |
Feature activation | +3.03e-01 |
Pos logit contributions | |
igsaw | +0.429 |
oak | +0.350 |
newspaper | +0.348 |
magazine | +0.346 |
cane | +0.311 |
Neg logit contributions | |
ered | -0.491 |
bled | -0.473 |
hadn | -0.461 |
vowed | -0.451 |
wouldn | -0.437 |
Token | ant |
Token ID | 1885 |
Feature activation | +3.74e-01 |
Pos logit contributions | |
igsaw | +0.419 |
oak | +0.356 |
magazine | +0.327 |
newspaper | +0.322 |
pine | +0.307 |
Neg logit contributions | |
ered | -0.735 |
bled | -0.722 |
izes | -0.660 |
vowed | -0.654 |
eking | -0.641 |
Token | crawled |
Token ID | 45668 |
Feature activation | -3.22e-02 |
Pos logit contributions | |
oak | +0.749 |
igsaw | +0.696 |
pine | +0.680 |
ant | +0.676 |
magazine | +0.650 |
Neg logit contributions | |
bled | -0.660 |
ered | -0.588 |
eking | -0.532 |
vere | -0.517 |
uffed | -0.513 |
Token | around |
Token ID | 1088 |
Feature activation | +9.90e-02 |
Pos logit contributions | |
ered | +0.063 |
bled | +0.063 |
izes | +0.056 |
vowed | +0.055 |
hadn | +0.055 |
Neg logit contributions | |
igsaw | -0.060 |
oak | -0.051 |
magazine | -0.049 |
newspaper | -0.048 |
pine | -0.046 |
Token | singing |
Token ID | 13777 |
Feature activation | +9.78e-03 |
Pos logit contributions | |
igsaw | +0.161 |
oak | +0.129 |
magazine | +0.123 |
newspaper | +0.123 |
automobile | +0.110 |
Neg logit contributions | |
ered | -0.216 |
bled | -0.209 |
vowed | -0.194 |
izes | -0.189 |
hadn | -0.188 |
Token | the |
Token ID | 262 |
Feature activation | +2.86e-01 |
Pos logit contributions | |
igsaw | +0.016 |
oak | +0.013 |
magazine | +0.012 |
newspaper | +0.012 |
pine | +0.011 |
Neg logit contributions | |
ered | -0.021 |
bled | -0.021 |
izes | -0.019 |
vowed | -0.019 |
hadn | -0.019 |
Token | beautiful |
Token ID | 4950 |
Feature activation | +2.01e-01 |
Pos logit contributions | |
igsaw | +0.409 |
oak | +0.335 |
magazine | +0.311 |
newspaper | +0.305 |
pine | +0.286 |
Neg logit contributions | |
ered | -0.674 |
bled | -0.669 |
izes | -0.609 |
vowed | -0.605 |
hadn | -0.599 |
Token | He |
Token ID | 679 |
Feature activation | +1.15e-02 |
Pos logit contributions | |
igsaw | +0.377 |
oak | +0.323 |
magazine | +0.299 |
newspaper | +0.293 |
pine | +0.282 |
Neg logit contributions | |
ered | -0.543 |
bled | -0.542 |
izes | -0.486 |
eking | -0.484 |
vowed | -0.473 |
Token | carefully |
Token ID | 7773 |
Feature activation | +7.47e-02 |
Pos logit contributions | |
oak | +0.020 |
pine | +0.020 |
lion | +0.019 |
dog | +0.019 |
rh | +0.019 |
Neg logit contributions | |
bled | -0.023 |
eking | -0.018 |
bling | -0.018 |
vere | -0.018 |
ered | -0.017 |
Token | carried |
Token ID | 5281 |
Feature activation | -3.40e-02 |
Pos logit contributions | |
igsaw | +0.149 |
oak | +0.139 |
pine | +0.134 |
newspaper | +0.132 |
magazine | +0.129 |
Neg logit contributions | |
bled | -0.126 |
itates | -0.111 |
ered | -0.108 |
vere | -0.106 |
izes | -0.104 |
Token | the |
Token ID | 262 |
Feature activation | +3.08e-01 |
Pos logit contributions | |
ered | +0.065 |
bled | +0.064 |
vowed | +0.056 |
izes | +0.055 |
hadn | +0.054 |
Neg logit contributions | |
igsaw | -0.065 |
oak | -0.057 |
magazine | -0.054 |
newspaper | -0.053 |
pine | -0.050 |
Token | little |
Token ID | 1310 |
Feature activation | +3.48e-01 |
Pos logit contributions | |
igsaw | +0.393 |
oak | +0.308 |
magazine | +0.276 |
newspaper | +0.270 |
pine | +0.254 |
Neg logit contributions | |
ered | -0.761 |
bled | -0.756 |
izes | -0.692 |
vowed | -0.688 |
hadn | -0.687 |
Token | ant |
Token ID | 1885 |
Feature activation | +3.74e-01 |
Pos logit contributions | |
igsaw | +0.436 |
oak | +0.332 |
magazine | +0.302 |
newspaper | +0.295 |
pine | +0.267 |
Neg logit contributions | |
ered | -0.858 |
bled | -0.846 |
vowed | -0.783 |
hadn | -0.782 |
izes | -0.779 |
Token | down |
Token ID | 866 |
Feature activation | +1.11e-01 |
Pos logit contributions | |
igsaw | +0.618 |
oak | +0.510 |
magazine | +0.480 |
newspaper | +0.478 |
pine | +0.447 |
Neg logit contributions | |
ered | -0.802 |
bled | -0.773 |
vowed | -0.708 |
izes | -0.708 |
apologised | -0.686 |
Token | the |
Token ID | 262 |
Feature activation | +2.96e-01 |
Pos logit contributions | |
igsaw | +0.147 |
oak | +0.123 |
magazine | +0.111 |
newspaper | +0.109 |
pine | +0.106 |
Neg logit contributions | |
ered | -0.273 |
bled | -0.272 |
izes | -0.250 |
vowed | -0.244 |
eking | -0.243 |
Token | tree |
Token ID | 5509 |
Feature activation | +2.96e-01 |
Pos logit contributions | |
igsaw | +0.362 |
oak | +0.259 |
magazine | +0.255 |
newspaper | +0.240 |
olive | +0.211 |
Neg logit contributions | |
ered | -0.717 |
bled | -0.716 |
hadn | -0.658 |
izes | -0.652 |
vowed | -0.652 |
Token | . |
Token ID | 13 |
Feature activation | +1.86e-01 |
Pos logit contributions | |
igsaw | +0.430 |
oak | +0.353 |
magazine | +0.329 |
newspaper | +0.323 |
pine | +0.303 |
Neg logit contributions | |
ered | -0.694 |
bled | -0.685 |
izes | -0.623 |
vowed | -0.621 |
hadn | -0.613 |
Token |
|
Token ID | 198 |
Feature activation | +1.02e-01 |
Pos logit contributions | |
igsaw | +0.314 |
oak | +0.268 |
magazine | +0.252 |
newspaper | +0.249 |
pine | +0.236 |
Neg logit contributions | |
ered | -0.391 |
bled | -0.387 |
izes | -0.347 |
vowed | -0.346 |
hadn | -0.342 |
Token | time |
Token ID | 640 |
Feature activation | +2.68e-01 |
Pos logit contributions | |
igsaw | +0.463 |
oak | +0.408 |
magazine | +0.379 |
newspaper | +0.365 |
olive | +0.349 |
Neg logit contributions | |
ered | -0.468 |
bled | -0.449 |
hadn | -0.426 |
vowed | -0.424 |
izes | -0.422 |
Token | there |
Token ID | 612 |
Feature activation | +1.54e-01 |
Pos logit contributions | |
igsaw | +0.295 |
oak | +0.230 |
magazine | +0.193 |
pine | +0.190 |
newspaper | +0.189 |
Neg logit contributions | |
ered | -0.708 |
bled | -0.707 |
izes | -0.661 |
vowed | -0.658 |
apologised | -0.650 |
Token | was |
Token ID | 373 |
Feature activation | -8.63e-02 |
Pos logit contributions | |
igsaw | +0.272 |
oak | +0.272 |
pine | +0.254 |
ant | +0.246 |
newspaper | +0.246 |
Neg logit contributions | |
bled | -0.268 |
ered | -0.256 |
izes | -0.245 |
eking | -0.245 |
apologised | -0.232 |
Token | a |
Token ID | 257 |
Feature activation | +2.23e-01 |
Pos logit contributions | |
ered | +0.183 |
bled | +0.182 |
izes | +0.164 |
vowed | +0.162 |
hadn | +0.159 |
Neg logit contributions | |
igsaw | -0.145 |
oak | -0.123 |
magazine | -0.116 |
newspaper | -0.114 |
pine | -0.109 |
Token | ready |
Token ID | 3492 |
Feature activation | +1.49e-02 |
Pos logit contributions | |
igsaw | +0.361 |
oak | +0.283 |
magazine | +0.267 |
newspaper | +0.264 |
olive | +0.241 |
Neg logit contributions | |
bled | -0.455 |
ered | -0.451 |
hadn | -0.430 |
vowed | -0.420 |
izes | -0.419 |
Token | oct |
Token ID | 19318 |
Feature activation | +3.74e-01 |
Pos logit contributions | |
igsaw | +0.020 |
oak | +0.016 |
magazine | +0.015 |
newspaper | +0.014 |
pine | +0.013 |
Neg logit contributions | |
ered | -0.035 |
bled | -0.035 |
hadn | -0.032 |
vowed | -0.032 |
izes | -0.031 |
Token | opus |
Token ID | 25790 |
Feature activation | +2.24e-01 |
Pos logit contributions | |
igsaw | +0.559 |
oak | +0.484 |
magazine | +0.449 |
newspaper | +0.447 |
pine | +0.425 |
Neg logit contributions | |
bled | -0.833 |
ered | -0.828 |
izes | -0.752 |
eking | -0.745 |
hadn | -0.733 |
Token | . |
Token ID | 13 |
Feature activation | +1.93e-01 |
Pos logit contributions | |
igsaw | +0.375 |
oak | +0.311 |
magazine | +0.293 |
newspaper | +0.289 |
pine | +0.272 |
Neg logit contributions | |
ered | -0.476 |
bled | -0.468 |
vowed | -0.424 |
izes | -0.423 |
hadn | -0.420 |
Token | She |
Token ID | 1375 |
Feature activation | -3.06e-02 |
Pos logit contributions | |
igsaw | +0.323 |
oak | +0.270 |
magazine | +0.259 |
newspaper | +0.256 |
pine | +0.237 |
Neg logit contributions | |
ered | -0.407 |
bled | -0.399 |
vowed | -0.363 |
hadn | -0.361 |
izes | -0.356 |
Token | wanted |
Token ID | 2227 |
Feature activation | -1.40e-01 |
Pos logit contributions | |
bled | +0.062 |
eking | +0.053 |
izes | +0.052 |
vere | +0.052 |
ered | +0.052 |
Neg logit contributions | |
pine | -0.056 |
oak | -0.055 |
rh | -0.048 |
bull | -0.048 |
ostr | -0.048 |
Token | to |
Token ID | 284 |
Feature activation | +2.04e-01 |
Pos logit contributions | |
bled | +0.391 |
ered | +0.386 |
izes | +0.363 |
vowed | +0.352 |
hadn | +0.349 |
Neg logit contributions | |
igsaw | -0.139 |
oak | -0.118 |
newspaper | -0.104 |
magazine | -0.101 |
pine | -0.098 |
Token | couldn |
Token ID | 3521 |
Feature activation | -6.38e-02 |
Pos logit contributions | |
igsaw | +0.250 |
oak | +0.239 |
pine | +0.228 |
ant | +0.219 |
newspaper | +0.214 |
Neg logit contributions | |
bled | -0.247 |
vere | -0.224 |
ered | -0.216 |
izes | -0.214 |
eking | -0.207 |
Token | 't |
Token ID | 470 |
Feature activation | -2.04e-02 |
Pos logit contributions | |
bled | +0.109 |
ered | +0.102 |
eking | +0.097 |
izes | +0.094 |
vowed | +0.090 |
Neg logit contributions | |
igsaw | -0.137 |
oak | -0.117 |
magazine | -0.108 |
pine | -0.104 |
newspaper | -0.103 |
Token | see |
Token ID | 766 |
Feature activation | -1.54e-02 |
Pos logit contributions | |
bled | +0.040 |
ered | +0.035 |
uffed | +0.035 |
izes | +0.034 |
icked | +0.033 |
Neg logit contributions | |
igsaw | -0.039 |
oak | -0.036 |
pine | -0.031 |
ant | -0.031 |
magazine | -0.031 |
Token | what |
Token ID | 644 |
Feature activation | +6.25e-02 |
Pos logit contributions | |
ered | +0.030 |
bled | +0.028 |
vowed | +0.025 |
hadn | +0.024 |
eking | +0.024 |
Neg logit contributions | |
igsaw | -0.030 |
oak | -0.025 |
magazine | -0.024 |
newspaper | -0.024 |
dictionary | -0.022 |
Token | the |
Token ID | 262 |
Feature activation | +3.03e-01 |
Pos logit contributions | |
igsaw | +0.117 |
oak | +0.103 |
magazine | +0.099 |
newspaper | +0.098 |
ant | +0.094 |
Neg logit contributions | |
bled | -0.121 |
ered | -0.118 |
izes | -0.107 |
eking | -0.106 |
vowed | -0.102 |
Token | ostr |
Token ID | 42387 |
Feature activation | +3.72e-01 |
Pos logit contributions | |
igsaw | +0.394 |
oak | +0.318 |
magazine | +0.291 |
newspaper | +0.285 |
pine | +0.266 |
Neg logit contributions | |
ered | -0.754 |
bled | -0.747 |
izes | -0.684 |
vowed | -0.680 |
hadn | -0.673 |
Token | ich |
Token ID | 488 |
Feature activation | +2.06e-01 |
Pos logit contributions | |
igsaw | +0.539 |
oak | +0.468 |
magazine | +0.421 |
newspaper | +0.413 |
pine | +0.402 |
Neg logit contributions | |
ered | -0.858 |
bled | -0.847 |
izes | -0.771 |
vowed | -0.770 |
hadn | -0.758 |
Token | was |
Token ID | 373 |
Feature activation | -1.09e-01 |
Pos logit contributions | |
oak | +0.429 |
igsaw | +0.408 |
magazine | +0.394 |
newspaper | +0.386 |
pine | +0.385 |
Neg logit contributions | |
bled | -0.346 |
ered | -0.308 |
izes | -0.279 |
itates | -0.262 |
uffed | -0.262 |
Token | hiding |
Token ID | 11816 |
Feature activation | +5.09e-02 |
Pos logit contributions | |
bled | +0.226 |
ered | +0.216 |
vowed | +0.198 |
izes | +0.194 |
uffed | +0.194 |
Neg logit contributions | |
igsaw | -0.178 |
oak | -0.167 |
magazine | -0.154 |
newspaper | -0.154 |
bull | -0.146 |
Token | , |
Token ID | 11 |
Feature activation | +2.26e-01 |
Pos logit contributions | |
igsaw | +0.077 |
oak | +0.067 |
magazine | +0.062 |
newspaper | +0.060 |
ant | +0.058 |
Neg logit contributions | |
ered | -0.112 |
bled | -0.110 |
izes | -0.102 |
vowed | -0.100 |
apologised | -0.100 |
Token | but |
Token ID | 475 |
Feature activation | -1.01e-02 |
Pos logit contributions | |
igsaw | +0.279 |
oak | +0.277 |
newspaper | +0.245 |
magazine | +0.244 |
pine | +0.241 |
Neg logit contributions | |
bled | -0.552 |
ered | -0.541 |
izes | -0.502 |
icked | -0.493 |
eking | -0.475 |
Token | boat |
Token ID | 8848 |
Feature activation | +2.41e-01 |
Pos logit contributions | |
igsaw | +0.450 |
oak | +0.379 |
magazine | +0.349 |
newspaper | +0.344 |
pine | +0.323 |
Neg logit contributions | |
ered | -0.825 |
bled | -0.821 |
izes | -0.744 |
vowed | -0.741 |
hadn | -0.726 |
Token | . |
Token ID | 13 |
Feature activation | +2.15e-01 |
Pos logit contributions | |
igsaw | +0.281 |
oak | +0.270 |
magazine | +0.234 |
newspaper | +0.229 |
pine | +0.223 |
Neg logit contributions | |
bled | -0.590 |
ered | -0.572 |
hadn | -0.527 |
itates | -0.521 |
icked | -0.520 |
Token |
|
Token ID | 198 |
Feature activation | +1.27e-01 |
Pos logit contributions | |
igsaw | +0.353 |
oak | +0.300 |
magazine | +0.279 |
newspaper | +0.275 |
pine | +0.264 |
Neg logit contributions | |
ered | -0.468 |
bled | -0.463 |
izes | -0.420 |
eking | -0.411 |
vowed | -0.410 |
Token |
|
Token ID | 198 |
Feature activation | +1.20e-01 |
Pos logit contributions | |
igsaw | +0.194 |
oak | +0.163 |
magazine | +0.151 |
newspaper | +0.150 |
pine | +0.147 |
Neg logit contributions | |
bled | -0.288 |
ered | -0.286 |
izes | -0.257 |
eking | -0.255 |
vowed | -0.254 |
Token | The |
Token ID | 464 |
Feature activation | +2.15e-01 |
Pos logit contributions | |
igsaw | +0.222 |
oak | +0.191 |
magazine | +0.181 |
newspaper | +0.180 |
pine | +0.174 |
Neg logit contributions | |
ered | -0.234 |
bled | -0.233 |
izes | -0.203 |
vowed | -0.202 |
eking | -0.201 |
Token | oct |
Token ID | 19318 |
Feature activation | +3.71e-01 |
Pos logit contributions | |
igsaw | +0.261 |
oak | +0.228 |
magazine | +0.204 |
newspaper | +0.200 |
bull | +0.191 |
Neg logit contributions | |
ered | -0.558 |
bled | -0.540 |
izes | -0.496 |
icked | -0.494 |
vowed | -0.493 |
Token | opus |
Token ID | 25790 |
Feature activation | +2.17e-01 |
Pos logit contributions | |
igsaw | +0.415 |
oak | +0.303 |
magazine | +0.278 |
newspaper | +0.267 |
ant | +0.253 |
Neg logit contributions | |
ered | -1.005 |
bled | -0.976 |
vowed | -0.917 |
izes | -0.895 |
hadn | -0.893 |
Token | was |
Token ID | 373 |
Feature activation | -1.13e-01 |
Pos logit contributions | |
oak | +0.381 |
lion | +0.372 |
par | +0.371 |
ant | +0.364 |
pine | +0.361 |
Neg logit contributions | |
bled | -0.379 |
ered | -0.343 |
vere | -0.324 |
eking | -0.313 |
izes | -0.311 |
Token | having |
Token ID | 1719 |
Feature activation | -1.40e-01 |
Pos logit contributions | |
ered | +0.215 |
bled | +0.214 |
izes | +0.205 |
vowed | +0.194 |
wouldn | +0.187 |
Neg logit contributions | |
igsaw | -0.175 |
oak | -0.167 |
pine | -0.161 |
ant | -0.161 |
bull | -0.158 |
Token | so |
Token ID | 523 |
Feature activation | +5.27e-02 |
Pos logit contributions | |
bled | +0.323 |
ered | +0.315 |
izes | +0.304 |
eking | +0.283 |
itates | +0.283 |
Neg logit contributions | |
igsaw | -0.214 |
oak | -0.185 |
magazine | -0.178 |
newspaper | -0.174 |
pine | -0.171 |
Token | much |
Token ID | 881 |
Feature activation | +8.36e-02 |
Pos logit contributions | |
igsaw | +0.110 |
oak | +0.095 |
pine | +0.091 |
newspaper | +0.088 |
magazine | +0.088 |
Neg logit contributions | |
bled | -0.098 |
ered | -0.093 |
eking | -0.084 |
izes | -0.083 |
hadn | -0.080 |
Token | on |
Token ID | 319 |
Feature activation | +1.66e-01 |
Pos logit contributions | |
igsaw | +0.081 |
oak | +0.074 |
magazine | +0.074 |
newspaper | +0.070 |
pine | +0.067 |
Neg logit contributions | |
bled | -0.132 |
ered | -0.131 |
vowed | -0.122 |
izes | -0.121 |
hadn | -0.120 |
Token | the |
Token ID | 262 |
Feature activation | +3.25e-01 |
Pos logit contributions | |
igsaw | +0.276 |
oak | +0.236 |
magazine | +0.213 |
newspaper | +0.208 |
bull | +0.204 |
Neg logit contributions | |
ered | -0.349 |
bled | -0.334 |
vowed | -0.310 |
hadn | -0.301 |
izes | -0.299 |
Token | ground |
Token ID | 2323 |
Feature activation | +1.46e-01 |
Pos logit contributions | |
igsaw | +0.440 |
oak | +0.349 |
magazine | +0.319 |
newspaper | +0.311 |
pine | +0.292 |
Neg logit contributions | |
ered | -0.789 |
bled | -0.783 |
izes | -0.717 |
vowed | -0.710 |
hadn | -0.703 |
Token | and |
Token ID | 290 |
Feature activation | +1.78e-01 |
Pos logit contributions | |
igsaw | +0.194 |
oak | +0.163 |
magazine | +0.150 |
newspaper | +0.148 |
pine | +0.139 |
Neg logit contributions | |
ered | -0.354 |
bled | -0.354 |
izes | -0.321 |
vowed | -0.320 |
hadn | -0.315 |
Token | the |
Token ID | 262 |
Feature activation | +2.99e-01 |
Pos logit contributions | |
oak | +0.308 |
igsaw | +0.301 |
pine | +0.295 |
newspaper | +0.293 |
magazine | +0.285 |
Neg logit contributions | |
bled | -0.315 |
ered | -0.304 |
vere | -0.281 |
eking | -0.280 |
itates | -0.278 |
Token | ant |
Token ID | 1885 |
Feature activation | +3.70e-01 |
Pos logit contributions | |
igsaw | +0.378 |
oak | +0.315 |
magazine | +0.288 |
newspaper | +0.283 |
pine | +0.265 |
Neg logit contributions | |
ered | -0.757 |
bled | -0.746 |
izes | -0.684 |
vowed | -0.680 |
hadn | -0.668 |
Token | started |
Token ID | 2067 |
Feature activation | -2.88e-02 |
Pos logit contributions | |
oak | +0.636 |
igsaw | +0.588 |
pine | +0.587 |
magazine | +0.565 |
newspaper | +0.565 |
Neg logit contributions | |
bled | -0.701 |
ered | -0.624 |
itates | -0.623 |
izes | -0.593 |
vere | -0.568 |
Token | to |
Token ID | 284 |
Feature activation | +2.11e-01 |
Pos logit contributions | |
bled | +0.077 |
ered | +0.076 |
izes | +0.072 |
vowed | +0.069 |
itates | +0.069 |
Neg logit contributions | |
igsaw | -0.030 |
oak | -0.026 |
pine | -0.022 |
magazine | -0.022 |
newspaper | -0.022 |
Token | eat |
Token ID | 4483 |
Feature activation | +2.80e-02 |
Pos logit contributions | |
igsaw | +0.332 |
oak | +0.324 |
pine | +0.292 |
newspaper | +0.273 |
magazine | +0.272 |
Neg logit contributions | |
bled | -0.472 |
ered | -0.428 |
icked | -0.408 |
izes | -0.405 |
uffed | -0.400 |
Token | . |
Token ID | 13 |
Feature activation | +1.82e-01 |
Pos logit contributions | |
igsaw | +0.046 |
oak | +0.039 |
magazine | +0.037 |
newspaper | +0.036 |
pine | +0.035 |
Neg logit contributions | |
ered | -0.060 |
bled | -0.060 |
izes | -0.054 |
vowed | -0.053 |
hadn | -0.053 |
Token | Lily |
Token ID | 20037 |
Feature activation | +8.61e-02 |
Pos logit contributions | |
igsaw | +0.296 |
oak | +0.252 |
magazine | +0.234 |
newspaper | +0.231 |
pine | +0.223 |
Neg logit contributions | |
ered | -0.399 |
bled | -0.395 |
izes | -0.358 |
eking | -0.351 |
vowed | -0.350 |
Token | walked |
Token ID | 6807 |
Feature activation | +1.44e-02 |
Pos logit contributions | |
oak | +0.048 |
dog | +0.044 |
cat | +0.043 |
elephant | +0.042 |
lion | +0.041 |
Neg logit contributions | |
bled | -0.054 |
eking | -0.048 |
uffed | -0.047 |
ered | -0.045 |
bling | -0.045 |
Token | , |
Token ID | 11 |
Feature activation | +2.67e-01 |
Pos logit contributions | |
igsaw | +0.020 |
oak | +0.018 |
magazine | +0.017 |
newspaper | +0.016 |
pine | +0.015 |
Neg logit contributions | |
ered | -0.032 |
bled | -0.032 |
hadn | -0.029 |
eking | -0.029 |
eling | -0.029 |
Token | they |
Token ID | 484 |
Feature activation | -2.17e-02 |
Pos logit contributions | |
igsaw | +0.354 |
oak | +0.352 |
newspaper | +0.319 |
bull | +0.315 |
pine | +0.312 |
Neg logit contributions | |
ered | -0.598 |
bled | -0.582 |
izes | -0.541 |
icked | -0.529 |
vowed | -0.515 |
Token | saw |
Token ID | 2497 |
Feature activation | +1.79e-03 |
Pos logit contributions | |
bled | +0.044 |
ered | +0.041 |
izes | +0.038 |
its | +0.037 |
eking | +0.037 |
Neg logit contributions | |
oak | -0.035 |
pine | -0.030 |
dog | -0.030 |
par | -0.029 |
cat | -0.029 |
Token | an |
Token ID | 281 |
Feature activation | +2.07e-01 |
Pos logit contributions | |
igsaw | +0.002 |
oak | +0.002 |
newspaper | +0.002 |
magazine | +0.002 |
bull | +0.002 |
Neg logit contributions | |
bled | -0.004 |
ered | -0.004 |
izes | -0.004 |
eking | -0.004 |
vowed | -0.004 |
Token | ostr |
Token ID | 42387 |
Feature activation | +3.70e-01 |
Pos logit contributions | |
igsaw | +0.320 |
magazine | +0.238 |
newspaper | +0.224 |
oak | +0.221 |
Mic | +0.205 |
Neg logit contributions | |
ered | -0.486 |
bled | -0.447 |
vowed | -0.414 |
izes | -0.413 |
eking | -0.411 |
Token | ich |
Token ID | 488 |
Feature activation | +2.07e-01 |
Pos logit contributions | |
igsaw | +0.579 |
oak | +0.486 |
magazine | +0.454 |
newspaper | +0.447 |
pine | +0.423 |
Neg logit contributions | |
ered | -0.828 |
bled | -0.820 |
izes | -0.741 |
vowed | -0.736 |
hadn | -0.727 |
Token | . |
Token ID | 13 |
Feature activation | +1.92e-01 |
Pos logit contributions | |
oak | +0.263 |
igsaw | +0.250 |
magazine | +0.227 |
pine | +0.222 |
newspaper | +0.218 |
Neg logit contributions | |
bled | -0.501 |
ered | -0.484 |
izes | -0.440 |
vowed | -0.434 |
icked | -0.429 |
Token | " |
Token ID | 366 |
Feature activation | +5.52e-02 |
Pos logit contributions | |
igsaw | +0.313 |
oak | +0.263 |
magazine | +0.248 |
newspaper | +0.244 |
pine | +0.230 |
Neg logit contributions | |
ered | -0.409 |
bled | -0.405 |
hadn | -0.366 |
vowed | -0.365 |
izes | -0.360 |
Token | Look |
Token ID | 8567 |
Feature activation | -9.32e-02 |
Pos logit contributions | |
igsaw | +0.118 |
oak | +0.103 |
magazine | +0.099 |
newspaper | +0.097 |
pine | +0.094 |
Neg logit contributions | |
ered | -0.092 |
bled | -0.091 |
izes | -0.079 |
vowed | -0.079 |
hadn | -0.078 |
Token | !" |
Token ID | 2474 |
Feature activation | +1.15e-01 |
Pos logit contributions | |
ered | +0.226 |
bled | +0.216 |
eking | +0.192 |
izes | +0.191 |
icked | +0.191 |
Neg logit contributions | |
igsaw | -0.141 |
oak | -0.112 |
magazine | -0.106 |
newspaper | -0.105 |
pine | -0.095 |
Token | water |
Token ID | 1660 |
Feature activation | +2.02e-01 |
Pos logit contributions | |
igsaw | +0.428 |
oak | +0.342 |
magazine | +0.313 |
newspaper | +0.306 |
pine | +0.285 |
Neg logit contributions | |
ered | -0.870 |
bled | -0.862 |
izes | -0.790 |
vowed | -0.786 |
hadn | -0.777 |
Token | and |
Token ID | 290 |
Feature activation | +1.79e-01 |
Pos logit contributions | |
igsaw | +0.219 |
oak | +0.200 |
magazine | +0.172 |
newspaper | +0.169 |
pine | +0.159 |
Neg logit contributions | |
bled | -0.524 |
ered | -0.521 |
vowed | -0.473 |
izes | -0.471 |
hadn | -0.470 |
Token | wave |
Token ID | 6769 |
Feature activation | +8.24e-02 |
Pos logit contributions | |
oak | +0.283 |
igsaw | +0.273 |
pine | +0.258 |
newspaper | +0.253 |
ant | +0.250 |
Neg logit contributions | |
bled | -0.383 |
ered | -0.346 |
eking | -0.333 |
vere | -0.322 |
izes | -0.320 |
Token | at |
Token ID | 379 |
Feature activation | +1.78e-01 |
Pos logit contributions | |
igsaw | +0.061 |
newspaper | +0.054 |
magazine | +0.054 |
dog | +0.049 |
cane | +0.047 |
Neg logit contributions | |
bled | -0.218 |
ered | -0.214 |
eling | -0.208 |
eking | -0.208 |
hadn | -0.201 |
Token | the |
Token ID | 262 |
Feature activation | +3.02e-01 |
Pos logit contributions | |
igsaw | +0.285 |
oak | +0.247 |
magazine | +0.227 |
newspaper | +0.224 |
pine | +0.218 |
Neg logit contributions | |
bled | -0.385 |
ered | -0.385 |
izes | -0.352 |
vowed | -0.343 |
hadn | -0.342 |
Token | oct |
Token ID | 19318 |
Feature activation | +3.70e-01 |
Pos logit contributions | |
igsaw | +0.376 |
oak | +0.331 |
magazine | +0.304 |
newspaper | +0.302 |
ant | +0.283 |
Neg logit contributions | |
ered | -0.772 |
bled | -0.756 |
vowed | -0.688 |
izes | -0.687 |
icked | -0.683 |
Token | opus |
Token ID | 25790 |
Feature activation | +2.16e-01 |
Pos logit contributions | |
igsaw | +0.483 |
oak | +0.389 |
magazine | +0.358 |
newspaper | +0.350 |
pine | +0.326 |
Neg logit contributions | |
ered | -0.925 |
bled | -0.915 |
izes | -0.837 |
vowed | -0.834 |
hadn | -0.823 |
Token | . |
Token ID | 13 |
Feature activation | +1.88e-01 |
Pos logit contributions | |
igsaw | +0.208 |
oak | +0.183 |
magazine | +0.161 |
ant | +0.155 |
pine | +0.152 |
Neg logit contributions | |
bled | -0.593 |
ered | -0.569 |
izes | -0.515 |
hadn | -0.514 |
vowed | -0.507 |
Token | The |
Token ID | 383 |
Feature activation | +1.84e-01 |
Pos logit contributions | |
igsaw | +0.306 |
oak | +0.259 |
magazine | +0.243 |
newspaper | +0.239 |
pine | +0.227 |
Neg logit contributions | |
ered | -0.407 |
bled | -0.403 |
izes | -0.364 |
vowed | -0.362 |
hadn | -0.357 |
Token | oct |
Token ID | 19318 |
Feature activation | +3.56e-01 |
Pos logit contributions | |
igsaw | +0.169 |
oak | +0.126 |
magazine | +0.108 |
newspaper | +0.105 |
pine | +0.095 |
Neg logit contributions | |
ered | -0.531 |
bled | -0.525 |
izes | -0.487 |
vowed | -0.484 |
hadn | -0.478 |
Token | opus |
Token ID | 25790 |
Feature activation | +2.04e-01 |
Pos logit contributions | |
igsaw | +0.435 |
oak | +0.334 |
magazine | +0.309 |
newspaper | +0.299 |
ant | +0.277 |
Neg logit contributions | |
ered | -0.928 |
bled | -0.911 |
vowed | -0.841 |
izes | -0.834 |
hadn | -0.826 |
Token | bid |
Token ID | 8406 |
Feature activation | -5.66e-03 |
Pos logit contributions | |
oak | +0.500 |
ant | +0.483 |
cat | +0.476 |
pine | +0.469 |
cater | +0.465 |
Neg logit contributions | |
bled | -0.535 |
ered | -0.457 |
itates | -0.447 |
izes | -0.435 |
vere | -0.429 |
Token | goodbye |
Token ID | 24829 |
Feature activation | +1.59e-01 |
Pos logit contributions | |
bled | +0.012 |
ered | +0.012 |
hadn | +0.011 |
izes | +0.010 |
eking | +0.010 |
Neg logit contributions | |
igsaw | -0.009 |
oak | -0.008 |
magazine | -0.008 |
newspaper | -0.007 |
pine | -0.007 |
Token | to |
Token ID | 284 |
Feature activation | +2.17e-01 |
Pos logit contributions | |
igsaw | +0.212 |
oak | +0.174 |
magazine | +0.160 |
newspaper | +0.157 |
pine | +0.148 |
Neg logit contributions | |
ered | -0.388 |
bled | -0.384 |
izes | -0.351 |
vowed | -0.348 |
hadn | -0.345 |
Token | the |
Token ID | 262 |
Feature activation | +2.92e-01 |
Pos logit contributions | |
igsaw | +0.295 |
oak | +0.242 |
magazine | +0.223 |
newspaper | +0.218 |
pine | +0.206 |
Neg logit contributions | |
ered | -0.526 |
bled | -0.524 |
izes | -0.478 |
vowed | -0.471 |
hadn | -0.469 |
Token | huge |
Token ID | 3236 |
Feature activation | +2.62e-01 |
Pos logit contributions | |
igsaw | +0.300 |
oak | +0.240 |
magazine | +0.219 |
newspaper | +0.213 |
ant | +0.195 |
Neg logit contributions | |
ered | -0.822 |
bled | -0.811 |
izes | -0.744 |
vowed | -0.738 |
eking | -0.731 |
Token | ant |
Token ID | 1885 |
Feature activation | +3.68e-01 |
Pos logit contributions | |
igsaw | +0.286 |
oak | +0.209 |
newspaper | +0.184 |
magazine | +0.182 |
pine | +0.168 |
Neg logit contributions | |
ered | -0.679 |
bled | -0.658 |
vowed | -0.626 |
izes | -0.622 |
apologised | -0.616 |
Token | and |
Token ID | 290 |
Feature activation | +1.51e-01 |
Pos logit contributions | |
igsaw | +0.517 |
oak | +0.466 |
magazine | +0.424 |
newspaper | +0.407 |
pine | +0.405 |
Neg logit contributions | |
bled | -0.874 |
ered | -0.846 |
izes | -0.761 |
eking | -0.746 |
hadn | -0.745 |
Token | thanked |
Token ID | 26280 |
Feature activation | -2.77e-02 |
Pos logit contributions | |
oak | +0.292 |
pine | +0.277 |
igsaw | +0.267 |
par | +0.258 |
ant | +0.253 |
Neg logit contributions | |
bled | -0.251 |
ered | -0.242 |
eking | -0.227 |
itates | -0.227 |
its | -0.220 |
Token | him |
Token ID | 683 |
Feature activation | +3.12e-02 |
Pos logit contributions | |
ered | +0.062 |
bled | +0.061 |
vowed | +0.056 |
izes | +0.054 |
apologised | +0.054 |
Neg logit contributions | |
igsaw | -0.044 |
oak | -0.036 |
magazine | -0.034 |
newspaper | -0.034 |
pine | -0.031 |
Token | for |
Token ID | 329 |
Feature activation | +1.19e-02 |
Pos logit contributions | |
igsaw | +0.055 |
oak | +0.047 |
magazine | +0.044 |
newspaper | +0.044 |
pine | +0.042 |
Neg logit contributions | |
ered | -0.063 |
bled | -0.063 |
izes | -0.056 |
vowed | -0.056 |
hadn | -0.055 |
Token | showing |
Token ID | 4478 |
Feature activation | -2.32e-02 |
Pos logit contributions | |
igsaw | +0.018 |
oak | +0.018 |
magazine | +0.017 |
pine | +0.017 |
newspaper | +0.016 |
Neg logit contributions | |
ered | -0.023 |
bled | -0.022 |
itates | -0.022 |
uffed | -0.022 |
izes | -0.021 |
Token |
|
Token ID | 198 |
Feature activation | +1.34e-01 |
Pos logit contributions | |
igsaw | +0.244 |
oak | +0.207 |
magazine | +0.193 |
newspaper | +0.191 |
pine | +0.185 |
Neg logit contributions | |
ered | -0.318 |
bled | -0.318 |
izes | -0.284 |
vowed | -0.281 |
eking | -0.278 |
Token | " |
Token ID | 1 |
Feature activation | +1.72e-01 |
Pos logit contributions | |
igsaw | +0.250 |
oak | +0.216 |
magazine | +0.204 |
newspaper | +0.202 |
pine | +0.194 |
Neg logit contributions | |
ered | -0.261 |
bled | -0.258 |
izes | -0.229 |
vowed | -0.227 |
hadn | -0.223 |
Token | By |
Token ID | 3886 |
Feature activation | +7.84e-02 |
Pos logit contributions | |
igsaw | +0.335 |
oak | +0.295 |
newspaper | +0.287 |
magazine | +0.287 |
pine | +0.274 |
Neg logit contributions | |
ered | -0.323 |
bled | -0.309 |
izes | -0.273 |
eking | -0.273 |
vowed | -0.268 |
Token | e |
Token ID | 68 |
Feature activation | +1.60e-01 |
Pos logit contributions | |
igsaw | +0.099 |
oak | +0.085 |
ant | +0.083 |
magazine | +0.078 |
pine | +0.075 |
Neg logit contributions | |
bled | -0.172 |
ered | -0.167 |
uffed | -0.167 |
hadn | -0.164 |
wouldn | -0.163 |
Token | , |
Token ID | 11 |
Feature activation | +2.48e-01 |
Pos logit contributions | |
igsaw | +0.222 |
oak | +0.179 |
magazine | +0.165 |
newspaper | +0.163 |
pine | +0.151 |
Neg logit contributions | |
ered | -0.390 |
bled | -0.386 |
izes | -0.353 |
vowed | -0.350 |
hadn | -0.344 |
Token | oct |
Token ID | 19318 |
Feature activation | +3.68e-01 |
Pos logit contributions | |
igsaw | +0.355 |
oak | +0.293 |
magazine | +0.271 |
newspaper | +0.266 |
pine | +0.251 |
Neg logit contributions | |
ered | -0.585 |
bled | -0.580 |
izes | -0.528 |
vowed | -0.525 |
hadn | -0.518 |
Token | opus |
Token ID | 25790 |
Feature activation | +2.15e-01 |
Pos logit contributions | |
igsaw | +0.493 |
oak | +0.402 |
magazine | +0.370 |
newspaper | +0.362 |
pine | +0.339 |
Neg logit contributions | |
ered | -0.904 |
bled | -0.896 |
izes | -0.818 |
vowed | -0.814 |
hadn | -0.805 |
Token | !" |
Token ID | 2474 |
Feature activation | +1.17e-01 |
Pos logit contributions | |
igsaw | +0.258 |
oak | +0.234 |
pine | +0.208 |
magazine | +0.204 |
ant | +0.201 |
Neg logit contributions | |
bled | -0.545 |
ered | -0.522 |
hadn | -0.474 |
uffed | -0.470 |
vowed | -0.467 |
Token | Ben |
Token ID | 3932 |
Feature activation | +1.20e-01 |
Pos logit contributions | |
igsaw | +0.167 |
oak | +0.141 |
magazine | +0.131 |
newspaper | +0.129 |
pine | +0.124 |
Neg logit contributions | |
bled | -0.275 |
ered | -0.275 |
izes | -0.247 |
eking | -0.244 |
vowed | -0.243 |
Token | says |
Token ID | 1139 |
Feature activation | -6.08e-02 |
Pos logit contributions | |
igsaw | +0.188 |
oak | +0.180 |
pine | +0.173 |
magazine | +0.172 |
newspaper | +0.170 |
Neg logit contributions | |
bled | -0.258 |
ered | -0.237 |
eking | -0.214 |
hadn | -0.209 |
izes | -0.207 |
Token | . |
Token ID | 13 |
Feature activation | +1.97e-01 |
Pos logit contributions | |
bled | +0.160 |
ered | +0.160 |
izes | +0.151 |
hadn | +0.150 |
eling | +0.150 |
Neg logit contributions | |
igsaw | -0.055 |
oak | -0.039 |
magazine | -0.038 |
ant | -0.034 |
pine | -0.033 |
Token | orn |
Token ID | 1211 |
Feature activation | +3.15e-01 |
Pos logit contributions | |
ered | +0.515 |
bled | +0.506 |
izes | +0.455 |
vowed | +0.449 |
hadn | +0.437 |
Neg logit contributions | |
igsaw | -0.542 |
oak | -0.476 |
magazine | -0.453 |
newspaper | -0.446 |
pine | -0.430 |
Token | that |
Token ID | 326 |
Feature activation | +2.30e-01 |
Pos logit contributions | |
oak | +0.536 |
igsaw | +0.512 |
magazine | +0.478 |
pine | +0.476 |
olive | +0.474 |
Neg logit contributions | |
bled | -0.565 |
ered | -0.559 |
itates | -0.508 |
izes | -0.504 |
apologised | -0.499 |
Token | fell |
Token ID | 3214 |
Feature activation | -1.32e-02 |
Pos logit contributions | |
oak | +0.347 |
igsaw | +0.337 |
olive | +0.309 |
pine | +0.308 |
ant | +0.305 |
Neg logit contributions | |
bled | -0.515 |
ered | -0.485 |
uffed | -0.458 |
eking | -0.443 |
itates | -0.442 |
Token | from |
Token ID | 422 |
Feature activation | +1.95e-01 |
Pos logit contributions | |
ered | +0.028 |
bled | +0.028 |
izes | +0.025 |
vowed | +0.025 |
hadn | +0.024 |
Neg logit contributions | |
igsaw | -0.021 |
oak | -0.019 |
magazine | -0.018 |
newspaper | -0.017 |
pine | -0.017 |
Token | an |
Token ID | 281 |
Feature activation | +2.16e-01 |
Pos logit contributions | |
igsaw | +0.291 |
oak | +0.253 |
magazine | +0.230 |
newspaper | +0.225 |
pine | +0.220 |
Neg logit contributions | |
ered | -0.445 |
bled | -0.441 |
izes | -0.403 |
vowed | -0.397 |
hadn | -0.394 |
Token | oak |
Token ID | 26210 |
Feature activation | +3.67e-01 |
Pos logit contributions | |
igsaw | +0.370 |
magazine | +0.288 |
newspaper | +0.271 |
oak | +0.264 |
cane | +0.244 |
Neg logit contributions | |
ered | -0.471 |
bled | -0.436 |
vowed | -0.409 |
izes | -0.403 |
eking | -0.398 |
Token | tree |
Token ID | 5509 |
Feature activation | +3.08e-01 |
Pos logit contributions | |
igsaw | +0.491 |
newspaper | +0.318 |
magazine | +0.316 |
oak | +0.302 |
bull | +0.272 |
Neg logit contributions | |
ered | -0.903 |
bled | -0.873 |
hadn | -0.845 |
izes | -0.843 |
vowed | -0.837 |
Token | . |
Token ID | 13 |
Feature activation | +1.98e-01 |
Pos logit contributions | |
igsaw | +0.449 |
oak | +0.388 |
magazine | +0.355 |
newspaper | +0.347 |
pine | +0.333 |
Neg logit contributions | |
bled | -0.714 |
ered | -0.712 |
izes | -0.641 |
vowed | -0.635 |
hadn | -0.623 |
Token | It |
Token ID | 632 |
Feature activation | -3.88e-02 |
Pos logit contributions | |
igsaw | +0.304 |
oak | +0.256 |
magazine | +0.237 |
newspaper | +0.233 |
pine | +0.223 |
Neg logit contributions | |
ered | -0.449 |
bled | -0.446 |
izes | -0.404 |
vowed | -0.398 |
eking | -0.393 |
Token | lay |
Token ID | 3830 |
Feature activation | +3.64e-02 |
Pos logit contributions | |
bled | +0.081 |
bling | +0.067 |
eking | +0.064 |
vere | +0.064 |
ered | +0.062 |
Neg logit contributions | |
oak | -0.069 |
pine | -0.068 |
tree | -0.063 |
cat | -0.060 |
ant | -0.060 |
Token | on |
Token ID | 319 |
Feature activation | +1.22e-01 |
Pos logit contributions | |
igsaw | +0.054 |
oak | +0.050 |
magazine | +0.045 |
pine | +0.044 |
newspaper | +0.044 |
Neg logit contributions | |
ered | -0.080 |
bled | -0.076 |
izes | -0.072 |
vowed | -0.071 |
apologised | -0.071 |
Token | the |
Token ID | 262 |
Feature activation | +1.38e-01 |
Pos logit contributions | |
igsaw | +0.015 |
oak | +0.015 |
pine | +0.014 |
magazine | +0.014 |
newspaper | +0.014 |
Neg logit contributions | |
ered | -0.013 |
bled | -0.013 |
eking | -0.012 |
izes | -0.011 |
its | -0.011 |
Token | owner |
Token ID | 4870 |
Feature activation | +3.88e-02 |
Pos logit contributions | |
igsaw | +0.090 |
oak | +0.060 |
magazine | +0.049 |
newspaper | +0.047 |
pine | +0.037 |
Neg logit contributions | |
ered | -0.439 |
bled | -0.434 |
izes | -0.404 |
vowed | -0.402 |
hadn | -0.398 |
Token | smiled |
Token ID | 13541 |
Feature activation | -1.99e-01 |
Pos logit contributions | |
igsaw | +0.081 |
oak | +0.079 |
pine | +0.076 |
magazine | +0.073 |
ant | +0.073 |
Neg logit contributions | |
bled | -0.057 |
ered | -0.055 |
izes | -0.051 |
eking | -0.051 |
vere | -0.047 |
Token | . |
Token ID | 13 |
Feature activation | +2.85e-02 |
Pos logit contributions | |
ered | +0.439 |
bled | +0.435 |
izes | +0.393 |
vowed | +0.390 |
hadn | +0.385 |
Neg logit contributions | |
igsaw | -0.319 |
oak | -0.269 |
magazine | -0.252 |
newspaper | -0.248 |
pine | -0.235 |
Token | " |
Token ID | 366 |
Feature activation | -1.02e-01 |
Pos logit contributions | |
igsaw | +0.042 |
oak | +0.038 |
magazine | +0.033 |
newspaper | +0.033 |
pine | +0.033 |
Neg logit contributions | |
ered | -0.067 |
bled | -0.064 |
izes | -0.060 |
eking | -0.060 |
vowed | -0.057 |
Token | Looks |
Token ID | 41102 |
Feature activation | -3.57e-01 |
Pos logit contributions | |
ered | +0.165 |
bled | +0.159 |
izes | +0.139 |
vowed | +0.134 |
eking | +0.133 |
Neg logit contributions | |
igsaw | -0.226 |
oak | -0.203 |
magazine | -0.194 |
newspaper | -0.192 |
pine | -0.187 |
Token | like |
Token ID | 588 |
Feature activation | -1.28e-01 |
Pos logit contributions | |
bled | +0.823 |
ered | +0.800 |
eking | +0.717 |
izes | +0.710 |
vowed | +0.697 |
Neg logit contributions | |
igsaw | -0.530 |
oak | -0.464 |
magazine | -0.446 |
newspaper | -0.439 |
olive | -0.421 |
Token | you |
Token ID | 345 |
Feature activation | -1.89e-01 |
Pos logit contributions | |
ered | +0.274 |
bled | +0.274 |
izes | +0.246 |
eking | +0.235 |
uffed | +0.234 |
Neg logit contributions | |
igsaw | -0.205 |
oak | -0.195 |
magazine | -0.183 |
newspaper | -0.182 |
pine | -0.178 |
Token | two |
Token ID | 734 |
Feature activation | -2.40e-02 |
Pos logit contributions | |
bled | +0.393 |
ered | +0.369 |
izes | +0.334 |
zed | +0.324 |
uffed | +0.324 |
Neg logit contributions | |
oak | -0.315 |
igsaw | -0.302 |
elephant | -0.292 |
magazine | -0.289 |
olive | -0.281 |
Token | are |
Token ID | 389 |
Feature activation | -2.60e-01 |
Pos logit contributions | |
bled | +0.060 |
ered | +0.056 |
izes | +0.049 |
vowed | +0.048 |
itates | +0.048 |
Neg logit contributions | |
oak | -0.032 |
igsaw | -0.030 |
magazine | -0.030 |
newspaper | -0.028 |
pine | -0.028 |
Token | friends |
Token ID | 2460 |
Feature activation | -4.49e-02 |
Pos logit contributions | |
bled | +0.612 |
ered | +0.605 |
izes | +0.550 |
vowed | +0.540 |
hadn | +0.536 |
Neg logit contributions | |
igsaw | -0.353 |
oak | -0.329 |
magazine | -0.304 |
pine | -0.284 |
ant | -0.282 |
Token | felt |
Token ID | 2936 |
Feature activation | -3.20e-01 |
Pos logit contributions | |
bled | +0.112 |
ered | +0.109 |
eking | +0.101 |
vere | +0.098 |
izes | +0.090 |
Neg logit contributions | |
oak | -0.171 |
igsaw | -0.168 |
magazine | -0.164 |
olive | -0.163 |
ant | -0.161 |
Token | bad |
Token ID | 2089 |
Feature activation | +3.66e-02 |
Pos logit contributions | |
ered | +0.580 |
bled | +0.569 |
eking | +0.512 |
izes | +0.490 |
uffed | +0.479 |
Neg logit contributions | |
igsaw | -0.589 |
oak | -0.558 |
magazine | -0.531 |
olive | -0.512 |
newspaper | -0.510 |
Token | and |
Token ID | 290 |
Feature activation | -2.32e-03 |
Pos logit contributions | |
igsaw | +0.049 |
oak | +0.040 |
magazine | +0.038 |
newspaper | +0.037 |
pine | +0.034 |
Neg logit contributions | |
ered | -0.090 |
bled | -0.089 |
izes | -0.081 |
vowed | -0.080 |
hadn | -0.080 |
Token | wished |
Token ID | 16555 |
Feature activation | -2.32e-01 |
Pos logit contributions | |
ered | +0.003 |
bled | +0.003 |
eking | +0.003 |
uffed | +0.003 |
izes | +0.002 |
Neg logit contributions | |
oak | -0.005 |
igsaw | -0.005 |
olive | -0.005 |
magazine | -0.005 |
newspaper | -0.005 |
Token | he |
Token ID | 339 |
Feature activation | -1.69e-01 |
Pos logit contributions | |
ered | +0.515 |
bled | +0.491 |
izes | +0.460 |
vowed | +0.447 |
eking | +0.441 |
Neg logit contributions | |
igsaw | -0.359 |
oak | -0.314 |
magazine | -0.295 |
newspaper | -0.291 |
olive | -0.279 |
Token | hadn |
Token ID | 8020 |
Feature activation | -3.51e-01 |
Pos logit contributions | |
ered | +0.261 |
bled | +0.249 |
eking | +0.247 |
izes | +0.239 |
vere | +0.232 |
Neg logit contributions | |
oak | -0.304 |
magazine | -0.302 |
cat | -0.299 |
dog | -0.296 |
puppy | -0.293 |
Token | 't |
Token ID | 470 |
Feature activation | -2.15e-01 |
Pos logit contributions | |
ered | +0.508 |
bled | +0.497 |
izes | +0.434 |
vowed | +0.421 |
eking | +0.415 |
Neg logit contributions | |
igsaw | -0.825 |
oak | -0.741 |
magazine | -0.711 |
newspaper | -0.697 |
pine | -0.680 |
Token | done |
Token ID | 1760 |
Feature activation | -2.13e-01 |
Pos logit contributions | |
bled | +0.384 |
ered | +0.359 |
izes | +0.348 |
uffed | +0.323 |
itates | +0.312 |
Neg logit contributions | |
igsaw | -0.399 |
oak | -0.398 |
olive | -0.375 |
lime | -0.370 |
pine | -0.366 |
Token | that |
Token ID | 326 |
Feature activation | +1.39e-02 |
Pos logit contributions | |
ered | +0.423 |
bled | +0.418 |
izes | +0.380 |
vowed | +0.373 |
hadn | +0.365 |
Neg logit contributions | |
igsaw | -0.377 |
oak | -0.324 |
magazine | -0.315 |
newspaper | -0.308 |
pine | -0.290 |
Token | . |
Token ID | 13 |
Feature activation | +1.49e-02 |
Pos logit contributions | |
igsaw | +0.027 |
oak | +0.021 |
magazine | +0.021 |
newspaper | +0.021 |
bull | +0.020 |
Neg logit contributions | |
ered | -0.024 |
bled | -0.023 |
vowed | -0.023 |
hadn | -0.022 |
wouldn | -0.022 |
Token | He |
Token ID | 679 |
Feature activation | -2.03e-01 |
Pos logit contributions | |
igsaw | +0.026 |
oak | +0.022 |
magazine | +0.021 |
newspaper | +0.020 |
pine | +0.020 |
Neg logit contributions | |
ered | -0.030 |
bled | -0.030 |
vowed | -0.027 |
hadn | -0.027 |
izes | -0.027 |
Token | sad |
Token ID | 6507 |
Feature activation | -7.88e-02 |
Pos logit contributions | |
ered | +0.628 |
bled | +0.582 |
eking | +0.535 |
izes | +0.535 |
uffed | +0.509 |
Neg logit contributions | |
igsaw | -0.545 |
oak | -0.516 |
magazine | -0.500 |
olive | -0.487 |
lime | -0.485 |
Token | . |
Token ID | 13 |
Feature activation | +2.10e-02 |
Pos logit contributions | |
ered | +0.195 |
bled | +0.192 |
izes | +0.174 |
vowed | +0.174 |
apologised | +0.171 |
Neg logit contributions | |
igsaw | -0.105 |
oak | -0.086 |
magazine | -0.080 |
newspaper | -0.079 |
pine | -0.074 |
Token | He |
Token ID | 679 |
Feature activation | -1.99e-01 |
Pos logit contributions | |
igsaw | +0.033 |
oak | +0.028 |
magazine | +0.026 |
newspaper | +0.026 |
pine | +0.024 |
Neg logit contributions | |
ered | -0.047 |
bled | -0.047 |
izes | -0.042 |
vowed | -0.042 |
eking | -0.041 |
Token | wished |
Token ID | 16555 |
Feature activation | -2.30e-01 |
Pos logit contributions | |
bled | +0.409 |
vere | +0.347 |
bling | +0.346 |
ered | +0.327 |
eking | +0.323 |
Neg logit contributions | |
oak | -0.349 |
newspaper | -0.326 |
puppy | -0.325 |
magazine | -0.324 |
cat | -0.320 |
Token | he |
Token ID | 339 |
Feature activation | -1.67e-01 |
Pos logit contributions | |
ered | +0.483 |
bled | +0.471 |
izes | +0.433 |
vowed | +0.425 |
apologised | +0.414 |
Neg logit contributions | |
igsaw | -0.391 |
oak | -0.339 |
magazine | -0.314 |
newspaper | -0.312 |
pine | -0.299 |
Token | hadn |
Token ID | 8020 |
Feature activation | -3.50e-01 |
Pos logit contributions | |
ered | +0.297 |
bled | +0.282 |
izes | +0.272 |
eking | +0.270 |
its | +0.261 |
Neg logit contributions | |
magazine | -0.292 |
oak | -0.288 |
skeleton | -0.277 |
igsaw | -0.276 |
cat | -0.273 |
Token | 't |
Token ID | 470 |
Feature activation | -2.14e-01 |
Pos logit contributions | |
ered | +0.508 |
bled | +0.496 |
izes | +0.437 |
vowed | +0.419 |
eking | +0.418 |
Neg logit contributions | |
igsaw | -0.822 |
oak | -0.740 |
magazine | -0.711 |
newspaper | -0.693 |
pine | -0.679 |
Token | given |
Token ID | 1813 |
Feature activation | -1.66e-01 |
Pos logit contributions | |
bled | +0.372 |
ered | +0.341 |
izes | +0.338 |
uffed | +0.304 |
itates | +0.301 |
Neg logit contributions | |
igsaw | -0.414 |
oak | -0.410 |
pine | -0.384 |
olive | -0.382 |
magazine | -0.380 |
Token | away |
Token ID | 1497 |
Feature activation | +2.43e-02 |
Pos logit contributions | |
ered | +0.275 |
bled | +0.256 |
vowed | +0.223 |
wouldn | +0.214 |
hadn | +0.209 |
Neg logit contributions | |
igsaw | -0.375 |
oak | -0.325 |
newspaper | -0.307 |
bull | -0.304 |
magazine | -0.304 |
Token | his |
Token ID | 465 |
Feature activation | +1.80e-02 |
Pos logit contributions | |
igsaw | +0.049 |
oak | +0.037 |
newspaper | +0.037 |
magazine | +0.037 |
bull | +0.036 |
Neg logit contributions | |
ered | -0.047 |
bled | -0.042 |
vowed | -0.040 |
apologised | -0.039 |
hadn | -0.037 |
Token | special |
Token ID | 2041 |
Feature activation | +3.42e-02 |
Pos logit contributions | |
igsaw | +0.027 |
oak | +0.023 |
magazine | +0.021 |
newspaper | +0.021 |
pine | +0.019 |
Neg logit contributions | |
ered | -0.041 |
bled | -0.041 |
izes | -0.037 |
vowed | -0.037 |
hadn | -0.036 |
Token | of |
Token ID | 286 |
Feature activation | +3.53e-02 |
Pos logit contributions | |
ered | +0.094 |
bled | +0.093 |
izes | +0.085 |
vowed | +0.084 |
hadn | +0.083 |
Neg logit contributions | |
igsaw | -0.064 |
oak | -0.053 |
magazine | -0.049 |
newspaper | -0.049 |
pine | -0.047 |
Token | fun |
Token ID | 1257 |
Feature activation | -7.28e-02 |
Pos logit contributions | |
igsaw | +0.046 |
newspaper | +0.040 |
oak | +0.040 |
pine | +0.039 |
magazine | +0.038 |
Neg logit contributions | |
bled | -0.094 |
ered | -0.083 |
itates | -0.079 |
izes | -0.079 |
icked | -0.079 |
Token | , |
Token ID | 11 |
Feature activation | +7.34e-02 |
Pos logit contributions | |
bled | +0.172 |
ered | +0.168 |
izes | +0.156 |
vowed | +0.153 |
apologised | +0.153 |
Neg logit contributions | |
igsaw | -0.106 |
oak | -0.087 |
magazine | -0.080 |
newspaper | -0.080 |
pine | -0.076 |
Token | but |
Token ID | 475 |
Feature activation | -1.60e-01 |
Pos logit contributions | |
igsaw | +0.105 |
oak | +0.090 |
magazine | +0.084 |
newspaper | +0.084 |
pine | +0.078 |
Neg logit contributions | |
ered | -0.173 |
bled | -0.172 |
izes | -0.157 |
vowed | -0.152 |
eking | -0.150 |
Token | they |
Token ID | 484 |
Feature activation | -2.06e-01 |
Pos logit contributions | |
ered | +0.300 |
bled | +0.293 |
izes | +0.263 |
eking | +0.262 |
icked | +0.248 |
Neg logit contributions | |
igsaw | -0.307 |
oak | -0.271 |
newspaper | -0.254 |
magazine | -0.254 |
ant | -0.251 |
Token | hadn |
Token ID | 8020 |
Feature activation | -3.50e-01 |
Pos logit contributions | |
bled | +0.322 |
eking | +0.311 |
vere | +0.296 |
izes | +0.293 |
ered | +0.292 |
Neg logit contributions | |
oak | -0.409 |
igsaw | -0.376 |
ant | -0.362 |
pine | -0.361 |
tree | -0.357 |
Token | 't |
Token ID | 470 |
Feature activation | -2.13e-01 |
Pos logit contributions | |
ered | +0.520 |
bled | +0.511 |
izes | +0.449 |
eking | +0.434 |
vowed | +0.430 |
Neg logit contributions | |
igsaw | -0.819 |
oak | -0.727 |
magazine | -0.695 |
newspaper | -0.681 |
pine | -0.664 |
Token | found |
Token ID | 1043 |
Feature activation | -2.15e-01 |
Pos logit contributions | |
bled | +0.354 |
izes | +0.309 |
ered | +0.305 |
uffed | +0.282 |
itates | +0.279 |
Neg logit contributions | |
igsaw | -0.472 |
oak | -0.457 |
pine | -0.421 |
ant | -0.418 |
newspaper | -0.409 |
Token | any |
Token ID | 597 |
Feature activation | +8.74e-03 |
Pos logit contributions | |
ered | +0.449 |
bled | +0.446 |
izes | +0.401 |
vowed | +0.396 |
hadn | +0.391 |
Neg logit contributions | |
igsaw | -0.368 |
oak | -0.315 |
magazine | -0.296 |
newspaper | -0.292 |
pine | -0.279 |
Token | treasure |
Token ID | 14068 |
Feature activation | +2.12e-02 |
Pos logit contributions | |
igsaw | +0.015 |
oak | +0.013 |
magazine | +0.012 |
newspaper | +0.012 |
ant | +0.012 |
Neg logit contributions | |
bled | -0.019 |
ered | -0.018 |
izes | -0.017 |
uffed | -0.016 |
vowed | -0.016 |
Token | . |
Token ID | 13 |
Feature activation | +1.93e-02 |
Pos logit contributions | |
igsaw | +0.032 |
oak | +0.027 |
magazine | +0.025 |
newspaper | +0.025 |
pine | +0.023 |
Neg logit contributions | |
ered | -0.048 |
bled | -0.048 |
izes | -0.043 |
vowed | -0.043 |
apologised | -0.042 |
Token | been |
Token ID | 587 |
Feature activation | -2.76e-01 |
Pos logit contributions | |
bled | +0.530 |
ered | +0.504 |
izes | +0.481 |
uffed | +0.472 |
itates | +0.465 |
Neg logit contributions | |
newspaper | -0.521 |
igsaw | -0.504 |
magazine | -0.490 |
olive | -0.465 |
lime | -0.465 |
Token | lucky |
Token ID | 9670 |
Feature activation | -1.34e-02 |
Pos logit contributions | |
ered | +0.545 |
bled | +0.544 |
its | +0.499 |
izes | +0.495 |
uffed | +0.485 |
Neg logit contributions | |
ant | -0.412 |
igsaw | -0.409 |
oak | -0.380 |
magazine | -0.374 |
goat | -0.363 |
Token | that |
Token ID | 326 |
Feature activation | +1.51e-02 |
Pos logit contributions | |
bled | +0.035 |
ered | +0.035 |
izes | +0.032 |
eking | +0.031 |
vowed | +0.031 |
Neg logit contributions | |
igsaw | -0.016 |
oak | -0.013 |
magazine | -0.012 |
newspaper | -0.012 |
ant | -0.011 |
Token | the |
Token ID | 262 |
Feature activation | +1.28e-01 |
Pos logit contributions | |
igsaw | +0.021 |
oak | +0.018 |
magazine | +0.018 |
newspaper | +0.018 |
ant | +0.017 |
Neg logit contributions | |
ered | -0.036 |
bled | -0.036 |
eking | -0.032 |
izes | -0.032 |
uffed | -0.032 |
Token | steam |
Token ID | 13324 |
Feature activation | -2.53e-03 |
Pos logit contributions | |
igsaw | +0.146 |
oak | +0.132 |
bull | +0.129 |
newspaper | +0.128 |
ant | +0.126 |
Neg logit contributions | |
ered | -0.332 |
bled | -0.331 |
icked | -0.311 |
izes | -0.302 |
apologised | -0.296 |
Token | hadn |
Token ID | 8020 |
Feature activation | -3.49e-01 |
Pos logit contributions | |
ered | +0.004 |
bled | +0.004 |
izes | +0.004 |
eking | +0.004 |
icked | +0.004 |
Neg logit contributions | |
igsaw | -0.005 |
oak | -0.005 |
bull | -0.004 |
newspaper | -0.004 |
magazine | -0.004 |
Token | 't |
Token ID | 470 |
Feature activation | -2.14e-01 |
Pos logit contributions | |
bled | +0.562 |
ered | +0.556 |
izes | +0.511 |
eking | +0.489 |
apologised | +0.480 |
Neg logit contributions | |
igsaw | -0.750 |
oak | -0.661 |
magazine | -0.622 |
newspaper | -0.603 |
ant | -0.592 |
Token | hurt |
Token ID | 5938 |
Feature activation | -9.17e-02 |
Pos logit contributions | |
bled | +0.388 |
ered | +0.369 |
izes | +0.363 |
itates | +0.347 |
uffed | +0.339 |
Neg logit contributions | |
igsaw | -0.400 |
oak | -0.383 |
newspaper | -0.353 |
pine | -0.350 |
magazine | -0.343 |
Token | her |
Token ID | 607 |
Feature activation | +1.32e-02 |
Pos logit contributions | |
ered | +0.199 |
bled | +0.199 |
izes | +0.180 |
vowed | +0.177 |
hadn | +0.174 |
Neg logit contributions | |
igsaw | -0.148 |
oak | -0.126 |
magazine | -0.118 |
newspaper | -0.116 |
pine | -0.110 |
Token | . |
Token ID | 13 |
Feature activation | +1.58e-02 |
Pos logit contributions | |
igsaw | +0.017 |
oak | +0.015 |
magazine | +0.014 |
newspaper | +0.014 |
ant | +0.013 |
Neg logit contributions | |
bled | -0.032 |
ered | -0.032 |
izes | -0.029 |
icked | -0.028 |
vowed | -0.028 |
Token | From |
Token ID | 3574 |
Feature activation | -1.79e-01 |
Pos logit contributions | |
igsaw | +0.024 |
magazine | +0.020 |
oak | +0.019 |
newspaper | +0.018 |
pine | +0.017 |
Neg logit contributions | |
ered | -0.037 |
bled | -0.036 |
eking | -0.033 |
izes | -0.032 |
vowed | -0.031 |
Token | would |
Token ID | 561 |
Feature activation | -2.76e-01 |
Pos logit contributions | |
bled | +0.250 |
ered | +0.243 |
izes | +0.217 |
eking | +0.216 |
uffed | +0.209 |
Neg logit contributions | |
igsaw | -0.214 |
oak | -0.188 |
magazine | -0.183 |
newspaper | -0.182 |
ant | -0.171 |
Token | have |
Token ID | 423 |
Feature activation | -2.73e-01 |
Pos logit contributions | |
bled | +0.461 |
ered | +0.403 |
uffed | +0.392 |
icked | +0.379 |
izes | +0.364 |
Neg logit contributions | |
igsaw | -0.614 |
oak | -0.573 |
magazine | -0.529 |
newspaper | -0.529 |
ant | -0.517 |
Token | happened |
Token ID | 3022 |
Feature activation | -1.66e-01 |
Pos logit contributions | |
bled | +0.552 |
ered | +0.521 |
izes | +0.475 |
uffed | +0.466 |
eking | +0.459 |
Neg logit contributions | |
igsaw | -0.488 |
oak | -0.421 |
magazine | -0.403 |
newspaper | -0.400 |
ant | -0.383 |
Token | if |
Token ID | 611 |
Feature activation | -1.08e-01 |
Pos logit contributions | |
bled | +0.476 |
ered | +0.467 |
uffed | +0.429 |
icked | +0.427 |
apologised | +0.424 |
Neg logit contributions | |
igsaw | -0.149 |
oak | -0.109 |
magazine | -0.109 |
newspaper | -0.100 |
ant | -0.085 |
Token | she |
Token ID | 673 |
Feature activation | -1.57e-01 |
Pos logit contributions | |
bled | +0.256 |
ered | +0.247 |
eking | +0.234 |
uffed | +0.227 |
icked | +0.225 |
Neg logit contributions | |
igsaw | -0.154 |
magazine | -0.134 |
oak | -0.131 |
newspaper | -0.127 |
ant | -0.121 |
Token | hadn |
Token ID | 8020 |
Feature activation | -3.48e-01 |
Pos logit contributions | |
bled | +0.295 |
izes | +0.225 |
ered | +0.225 |
bling | +0.221 |
vere | +0.219 |
Neg logit contributions | |
oak | -0.306 |
igsaw | -0.301 |
pine | -0.286 |
ant | -0.284 |
elephant | -0.273 |
Token | 't |
Token ID | 470 |
Feature activation | -2.12e-01 |
Pos logit contributions | |
bled | +0.548 |
ered | +0.529 |
izes | +0.480 |
eking | +0.460 |
vowed | +0.459 |
Neg logit contributions | |
igsaw | -0.774 |
oak | -0.685 |
magazine | -0.647 |
newspaper | -0.633 |
ant | -0.621 |
Token | stopped |
Token ID | 5025 |
Feature activation | -1.69e-01 |
Pos logit contributions | |
bled | +0.412 |
ered | +0.345 |
izes | +0.341 |
uffed | +0.329 |
itates | +0.324 |
Neg logit contributions | |
igsaw | -0.412 |
oak | -0.403 |
newspaper | -0.362 |
pine | -0.359 |
ant | -0.354 |
Token | increasing |
Token ID | 3649 |
Feature activation | -1.54e-01 |
Pos logit contributions | |
bled | +0.363 |
ered | +0.357 |
izes | +0.326 |
vowed | +0.318 |
apologised | +0.315 |
Neg logit contributions | |
igsaw | -0.271 |
oak | -0.244 |
magazine | -0.222 |
ant | -0.219 |
newspaper | -0.219 |
Token | in |
Token ID | 287 |
Feature activation | -1.60e-02 |
Pos logit contributions | |
bled | +0.353 |
ered | +0.351 |
izes | +0.319 |
vowed | +0.316 |
hadn | +0.312 |
Neg logit contributions | |
igsaw | -0.231 |
oak | -0.192 |
magazine | -0.180 |
newspaper | -0.177 |
pine | -0.168 |
Token | height |
Token ID | 6001 |
Feature activation | -1.07e-02 |
Pos logit contributions | |
bled | +0.037 |
ered | +0.036 |
izes | +0.033 |
vowed | +0.032 |
apologised | +0.032 |
Neg logit contributions | |
igsaw | -0.025 |
oak | -0.021 |
magazine | -0.020 |
newspaper | -0.019 |
pine | -0.018 |
Token | were |
Token ID | 547 |
Feature activation | -3.01e-01 |
Pos logit contributions | |
oak | +0.069 |
elephant | +0.060 |
pine | +0.060 |
newspaper | +0.059 |
tree | +0.058 |
Neg logit contributions | |
bled | -0.076 |
izes | -0.071 |
vere | -0.068 |
ider | -0.065 |
eking | -0.065 |
Token | so |
Token ID | 523 |
Feature activation | -1.02e-01 |
Pos logit contributions | |
bled | +0.597 |
ered | +0.562 |
izes | +0.555 |
wouldn | +0.534 |
'd | +0.522 |
Neg logit contributions | |
magazine | -0.409 |
igsaw | -0.408 |
oak | -0.404 |
dog | -0.383 |
ant | -0.380 |
Token | surprised |
Token ID | 6655 |
Feature activation | -1.68e-01 |
Pos logit contributions | |
ered | +0.183 |
bled | +0.183 |
wouldn | +0.170 |
zed | +0.166 |
vowed | +0.163 |
Neg logit contributions | |
igsaw | -0.162 |
oak | -0.155 |
ant | -0.151 |
dog | -0.143 |
cat | -0.141 |
Token | . |
Token ID | 13 |
Feature activation | +2.15e-02 |
Pos logit contributions | |
ered | +0.350 |
bled | +0.333 |
vowed | +0.306 |
hadn | +0.305 |
izes | +0.303 |
Neg logit contributions | |
igsaw | -0.298 |
oak | -0.251 |
magazine | -0.241 |
newspaper | -0.229 |
pine | -0.219 |
Token | They |
Token ID | 1119 |
Feature activation | -2.56e-01 |
Pos logit contributions | |
igsaw | +0.036 |
oak | +0.030 |
magazine | +0.029 |
newspaper | +0.028 |
ant | +0.026 |
Neg logit contributions | |
bled | -0.047 |
ered | -0.047 |
izes | -0.042 |
eking | -0.041 |
vowed | -0.041 |
Token | hadn |
Token ID | 8020 |
Feature activation | -3.48e-01 |
Pos logit contributions | |
bled | +0.514 |
izes | +0.454 |
ider | +0.439 |
eking | +0.428 |
its | +0.421 |
Neg logit contributions | |
oak | -0.463 |
elephant | -0.397 |
pine | -0.390 |
newspaper | -0.377 |
cat | -0.377 |
Token | 't |
Token ID | 470 |
Feature activation | -2.12e-01 |
Pos logit contributions | |
ered | +0.496 |
bled | +0.494 |
izes | +0.432 |
vowed | +0.413 |
eking | +0.409 |
Neg logit contributions | |
igsaw | -0.821 |
oak | -0.739 |
magazine | -0.706 |
newspaper | -0.695 |
pine | -0.677 |
Token | thought |
Token ID | 1807 |
Feature activation | -2.48e-01 |
Pos logit contributions | |
bled | +0.384 |
ered | +0.353 |
izes | +0.333 |
uffed | +0.314 |
itates | +0.302 |
Neg logit contributions | |
igsaw | -0.423 |
oak | -0.405 |
newspaper | -0.382 |
magazine | -0.372 |
pine | -0.369 |
Token | she |
Token ID | 673 |
Feature activation | -1.59e-01 |
Pos logit contributions | |
bled | +0.566 |
ered | +0.540 |
izes | +0.509 |
vowed | +0.498 |
apologised | +0.495 |
Neg logit contributions | |
igsaw | -0.373 |
oak | -0.305 |
newspaper | -0.299 |
magazine | -0.288 |
ant | -0.284 |
Token | was |
Token ID | 373 |
Feature activation | -2.79e-01 |
Pos logit contributions | |
bled | +0.285 |
izes | +0.246 |
ered | +0.240 |
eking | +0.234 |
bling | +0.230 |
Neg logit contributions | |
newspaper | -0.294 |
oak | -0.290 |
cat | -0.287 |
dictionary | -0.283 |
magazine | -0.280 |
Token | awake |
Token ID | 21693 |
Feature activation | -6.81e-02 |
Pos logit contributions | |
bled | +0.540 |
ered | +0.531 |
izes | +0.490 |
uffed | +0.478 |
vowed | +0.471 |
Neg logit contributions | |
igsaw | -0.480 |
oak | -0.443 |
magazine | -0.435 |
ant | -0.425 |
newspaper | -0.421 |
Token | . |
Token ID | 13 |
Feature activation | +1.91e-02 |
Pos logit contributions | |
ered | +0.169 |
bled | +0.157 |
vowed | +0.156 |
hadn | +0.148 |
izes | +0.146 |
Neg logit contributions | |
igsaw | -0.225 |
oak | -0.183 |
magazine | -0.180 |
bull | -0.178 |
newspaper | -0.177 |
Token | She |
Token ID | 1375 |
Feature activation | -1.97e-01 |
Pos logit contributions | |
igsaw | +0.032 |
oak | +0.027 |
magazine | +0.026 |
newspaper | +0.025 |
pine | +0.024 |
Neg logit contributions | |
ered | -0.041 |
bled | -0.041 |
izes | -0.036 |
vowed | -0.036 |
eking | -0.036 |
Token | hoped |
Token ID | 10719 |
Feature activation | -2.54e-01 |
Pos logit contributions | |
bled | +0.331 |
ered | +0.316 |
vere | +0.284 |
eking | +0.274 |
izes | +0.271 |
Neg logit contributions | |
oak | -0.389 |
pine | -0.371 |
igsaw | -0.353 |
newspaper | -0.350 |
lime | -0.349 |
Token | that |
Token ID | 326 |
Feature activation | +1.62e-02 |
Pos logit contributions | |
ered | +0.526 |
bled | +0.517 |
vowed | +0.463 |
hadn | +0.457 |
izes | +0.457 |
Neg logit contributions | |
igsaw | -0.450 |
oak | -0.374 |
magazine | -0.357 |
newspaper | -0.350 |
bull | -0.333 |
Token | she |
Token ID | 673 |
Feature activation | -1.58e-01 |
Pos logit contributions | |
igsaw | +0.028 |
oak | +0.024 |
magazine | +0.022 |
newspaper | +0.022 |
pine | +0.021 |
Neg logit contributions | |
ered | -0.034 |
bled | -0.033 |
izes | -0.030 |
vowed | -0.030 |
hadn | -0.030 |
Token | wouldn |
Token ID | 3636 |
Feature activation | -3.48e-01 |
Pos logit contributions | |
bled | +0.351 |
ered | +0.327 |
aded | +0.320 |
eking | +0.314 |
bling | +0.301 |
Neg logit contributions | |
oak | -0.227 |
cat | -0.217 |
cane | -0.209 |
magazine | -0.206 |
pony | -0.204 |
Token | 't |
Token ID | 470 |
Feature activation | -2.14e-01 |
Pos logit contributions | |
ered | +0.501 |
bled | +0.499 |
izes | +0.429 |
eking | +0.419 |
vowed | +0.417 |
Neg logit contributions | |
igsaw | -0.808 |
oak | -0.733 |
magazine | -0.693 |
newspaper | -0.674 |
pine | -0.665 |
Token | hurt |
Token ID | 5938 |
Feature activation | -9.25e-02 |
Pos logit contributions | |
bled | +0.401 |
ered | +0.377 |
izes | +0.342 |
icked | +0.332 |
uffed | +0.331 |
Neg logit contributions | |
oak | -0.406 |
igsaw | -0.404 |
pine | -0.354 |
ant | -0.352 |
magazine | -0.349 |
Token | her |
Token ID | 607 |
Feature activation | +1.57e-02 |
Pos logit contributions | |
ered | +0.174 |
bled | +0.172 |
vowed | +0.151 |
izes | +0.151 |
hadn | +0.149 |
Neg logit contributions | |
igsaw | -0.180 |
oak | -0.155 |
magazine | -0.147 |
newspaper | -0.146 |
pine | -0.139 |
Token | finger |
Token ID | 7660 |
Feature activation | -2.38e-02 |
Pos logit contributions | |
igsaw | +0.035 |
oak | +0.029 |
magazine | +0.027 |
newspaper | +0.027 |
bull | +0.026 |
Neg logit contributions | |
bled | -0.027 |
ered | -0.027 |
vowed | -0.023 |
eking | -0.023 |
izes | -0.023 |
Token | again |
Token ID | 757 |
Feature activation | -1.10e-01 |
Pos logit contributions | |
bled | +0.050 |
ered | +0.049 |
izes | +0.044 |
uffed | +0.043 |
icked | +0.043 |
Neg logit contributions | |
igsaw | -0.037 |
oak | -0.035 |
magazine | -0.034 |
newspaper | -0.032 |
olive | -0.031 |
Token | see |
Token ID | 766 |
Feature activation | -2.04e-01 |
Pos logit contributions | |
igsaw | +0.110 |
oak | +0.094 |
magazine | +0.092 |
newspaper | +0.091 |
bull | +0.086 |
Neg logit contributions | |
ered | -0.080 |
bled | -0.075 |
vowed | -0.067 |
izes | -0.067 |
hadn | -0.065 |
Token | the |
Token ID | 262 |
Feature activation | +1.41e-01 |
Pos logit contributions | |
ered | +0.377 |
bled | +0.372 |
izes | +0.329 |
vowed | +0.326 |
hadn | +0.320 |
Neg logit contributions | |
igsaw | -0.400 |
oak | -0.347 |
magazine | -0.329 |
newspaper | -0.324 |
pine | -0.311 |
Token | spark |
Token ID | 9009 |
Feature activation | -8.63e-02 |
Pos logit contributions | |
igsaw | +0.173 |
oak | +0.165 |
pine | +0.141 |
magazine | +0.140 |
newspaper | +0.138 |
Neg logit contributions | |
ered | -0.357 |
bled | -0.355 |
izes | -0.325 |
vowed | -0.325 |
apologised | -0.318 |
Token | . |
Token ID | 13 |
Feature activation | +3.19e-02 |
Pos logit contributions | |
ered | +0.180 |
bled | +0.179 |
izes | +0.160 |
vowed | +0.159 |
hadn | +0.158 |
Neg logit contributions | |
igsaw | -0.147 |
oak | -0.126 |
magazine | -0.119 |
newspaper | -0.117 |
pine | -0.112 |
Token | It |
Token ID | 632 |
Feature activation | -2.07e-01 |
Pos logit contributions | |
igsaw | +0.063 |
oak | +0.055 |
magazine | +0.052 |
newspaper | +0.051 |
pine | +0.049 |
Neg logit contributions | |
ered | -0.058 |
bled | -0.058 |
izes | -0.051 |
vowed | -0.051 |
hadn | -0.050 |
Token | seemed |
Token ID | 3947 |
Feature activation | -3.47e-01 |
Pos logit contributions | |
bled | +0.466 |
eking | +0.382 |
ered | +0.379 |
bling | +0.378 |
uffed | +0.368 |
Neg logit contributions | |
cat | -0.319 |
oak | -0.315 |
elephant | -0.311 |
ostr | -0.306 |
mouse | -0.306 |
Token | pretty |
Token ID | 2495 |
Feature activation | +4.91e-02 |
Pos logit contributions | |
bled | +0.888 |
ered | +0.886 |
izes | +0.797 |
vowed | +0.781 |
eking | +0.774 |
Neg logit contributions | |
igsaw | -0.432 |
oak | -0.377 |
magazine | -0.352 |
newspaper | -0.346 |
pine | -0.337 |
Token | . |
Token ID | 13 |
Feature activation | +2.99e-02 |
Pos logit contributions | |
igsaw | +0.060 |
oak | +0.051 |
magazine | +0.048 |
newspaper | +0.046 |
pine | +0.043 |
Neg logit contributions | |
ered | -0.124 |
bled | -0.124 |
vowed | -0.112 |
izes | -0.112 |
hadn | -0.111 |
Token | I |
Token ID | 314 |
Feature activation | -2.50e-01 |
Pos logit contributions | |
igsaw | +0.058 |
oak | +0.051 |
magazine | +0.048 |
newspaper | +0.048 |
pine | +0.046 |
Neg logit contributions | |
ered | -0.055 |
bled | -0.054 |
izes | -0.048 |
vowed | -0.048 |
hadn | -0.047 |
Token | didn |
Token ID | 1422 |
Feature activation | -2.98e-01 |
Pos logit contributions | |
bled | +0.548 |
bling | +0.464 |
eking | +0.454 |
ered | +0.427 |
vere | +0.421 |
Neg logit contributions | |
oc | -0.404 |
pine | -0.403 |
oak | -0.402 |
rh | -0.391 |
igsaw | -0.389 |
Token | 't |
Token ID | 470 |
Feature activation | -2.16e-01 |
Pos logit contributions | |
ered | +0.330 |
bled | +0.318 |
izes | +0.249 |
vowed | +0.237 |
hadn | +0.235 |
Neg logit contributions | |
igsaw | -0.800 |
oak | -0.736 |
newspaper | -0.704 |
magazine | -0.702 |
pine | -0.687 |
Token | rubbed |
Token ID | 31862 |
Feature activation | -1.79e-01 |
Pos logit contributions | |
igsaw | +0.033 |
oak | +0.031 |
pine | +0.030 |
magazine | +0.029 |
newspaper | +0.029 |
Neg logit contributions | |
bled | -0.026 |
ered | -0.026 |
izes | -0.022 |
eking | -0.022 |
vowed | -0.021 |
Token | his |
Token ID | 465 |
Feature activation | +2.72e-02 |
Pos logit contributions | |
ered | +0.400 |
bled | +0.396 |
izes | +0.356 |
vowed | +0.356 |
hadn | +0.347 |
Neg logit contributions | |
igsaw | -0.290 |
oak | -0.241 |
magazine | -0.224 |
newspaper | -0.222 |
pine | -0.207 |
Token | head |
Token ID | 1182 |
Feature activation | -8.50e-02 |
Pos logit contributions | |
igsaw | +0.050 |
oak | +0.043 |
magazine | +0.040 |
newspaper | +0.040 |
pine | +0.038 |
Neg logit contributions | |
ered | -0.054 |
bled | -0.053 |
izes | -0.048 |
vowed | -0.047 |
hadn | -0.046 |
Token | . |
Token ID | 13 |
Feature activation | +3.13e-02 |
Pos logit contributions | |
ered | +0.208 |
bled | +0.208 |
izes | +0.190 |
hadn | +0.187 |
vowed | +0.187 |
Neg logit contributions | |
igsaw | -0.097 |
oak | -0.088 |
magazine | -0.083 |
pine | -0.079 |
newspaper | -0.079 |
Token | He |
Token ID | 679 |
Feature activation | -1.96e-01 |
Pos logit contributions | |
igsaw | +0.047 |
oak | +0.039 |
magazine | +0.037 |
newspaper | +0.036 |
pine | +0.035 |
Neg logit contributions | |
ered | -0.073 |
bled | -0.071 |
izes | -0.065 |
vowed | -0.064 |
eking | -0.064 |
Token | seemed |
Token ID | 3947 |
Feature activation | -3.47e-01 |
Pos logit contributions | |
bled | +0.362 |
ered | +0.333 |
bling | +0.305 |
eking | +0.302 |
vere | +0.300 |
Neg logit contributions | |
pine | -0.354 |
par | -0.354 |
lion | -0.342 |
oak | -0.339 |
rh | -0.337 |
Token | to |
Token ID | 284 |
Feature activation | +4.76e-02 |
Pos logit contributions | |
ered | +0.901 |
bled | +0.891 |
izes | +0.805 |
vowed | +0.786 |
eking | +0.785 |
Neg logit contributions | |
igsaw | -0.400 |
oak | -0.350 |
magazine | -0.322 |
pine | -0.321 |
newspaper | -0.317 |
Token | like |
Token ID | 588 |
Feature activation | -1.35e-01 |
Pos logit contributions | |
igsaw | +0.090 |
oak | +0.087 |
pine | +0.080 |
magazine | +0.080 |
newspaper | +0.079 |
Neg logit contributions | |
bled | -0.093 |
ered | -0.085 |
uffed | -0.077 |
izes | -0.075 |
icked | -0.074 |
Token | Lily |
Token ID | 20037 |
Feature activation | -6.36e-02 |
Pos logit contributions | |
ered | +0.305 |
bled | +0.303 |
izes | +0.276 |
vowed | +0.271 |
hadn | +0.267 |
Neg logit contributions | |
igsaw | -0.203 |
oak | -0.175 |
magazine | -0.163 |
newspaper | -0.161 |
pine | -0.156 |
Token | . |
Token ID | 13 |
Feature activation | +2.86e-02 |
Pos logit contributions | |
bled | +0.151 |
ered | +0.151 |
izes | +0.136 |
vowed | +0.134 |
hadn | +0.132 |
Neg logit contributions | |
igsaw | -0.089 |
oak | -0.076 |
magazine | -0.071 |
newspaper | -0.069 |
pine | -0.066 |
Token | He |
Token ID | 679 |
Feature activation | -1.98e-01 |
Pos logit contributions | |
igsaw | +0.048 |
oak | +0.041 |
magazine | +0.038 |
newspaper | +0.038 |
pine | +0.035 |
Neg logit contributions | |
ered | -0.061 |
bled | -0.061 |
izes | -0.055 |
vowed | -0.054 |
hadn | -0.054 |
Token | smiled |
Token ID | 13541 |
Feature activation | -1.86e-01 |
Pos logit contributions | |
igsaw | +0.095 |
oak | +0.090 |
olive | +0.087 |
pine | +0.086 |
ant | +0.086 |
Neg logit contributions | |
bled | -0.074 |
izes | -0.071 |
ered | -0.070 |
vere | -0.068 |
eking | -0.067 |
Token | and |
Token ID | 290 |
Feature activation | +1.97e-02 |
Pos logit contributions | |
bled | +0.416 |
ered | +0.413 |
vowed | +0.376 |
apologised | +0.367 |
izes | +0.367 |
Neg logit contributions | |
igsaw | -0.298 |
oak | -0.249 |
newspaper | -0.227 |
magazine | -0.223 |
pine | -0.214 |
Token | said |
Token ID | 531 |
Feature activation | -2.28e-01 |
Pos logit contributions | |
igsaw | +0.040 |
oak | +0.038 |
pine | +0.036 |
newspaper | +0.036 |
magazine | +0.035 |
Neg logit contributions | |
ered | -0.032 |
bled | -0.030 |
izes | -0.028 |
eking | -0.028 |
vere | -0.026 |
Token | , |
Token ID | 11 |
Feature activation | +9.01e-02 |
Pos logit contributions | |
ered | +0.689 |
bled | +0.681 |
izes | +0.652 |
eking | +0.651 |
vowed | +0.645 |
Neg logit contributions | |
igsaw | -0.166 |
oak | -0.099 |
magazine | -0.089 |
newspaper | -0.076 |
olive | -0.069 |
Token | " |
Token ID | 366 |
Feature activation | -9.14e-02 |
Pos logit contributions | |
igsaw | +0.105 |
oak | +0.091 |
magazine | +0.084 |
newspaper | +0.081 |
pine | +0.077 |
Neg logit contributions | |
ered | -0.230 |
bled | -0.214 |
izes | -0.209 |
vowed | -0.202 |
eking | -0.202 |
Token | Looks |
Token ID | 41102 |
Feature activation | -3.47e-01 |
Pos logit contributions | |
ered | +0.156 |
bled | +0.151 |
izes | +0.133 |
eking | +0.129 |
vowed | +0.127 |
Neg logit contributions | |
igsaw | -0.196 |
oak | -0.174 |
magazine | -0.167 |
newspaper | -0.166 |
pine | -0.160 |
Token | like |
Token ID | 588 |
Feature activation | -1.18e-01 |
Pos logit contributions | |
bled | +0.804 |
ered | +0.781 |
eking | +0.703 |
izes | +0.701 |
vowed | +0.689 |
Neg logit contributions | |
igsaw | -0.521 |
oak | -0.442 |
magazine | -0.429 |
newspaper | -0.421 |
olive | -0.408 |
Token | you |
Token ID | 345 |
Feature activation | -1.79e-01 |
Pos logit contributions | |
bled | +0.258 |
ered | +0.253 |
izes | +0.238 |
uffed | +0.228 |
itates | +0.227 |
Neg logit contributions | |
igsaw | -0.193 |
oak | -0.168 |
magazine | -0.166 |
newspaper | -0.165 |
ant | -0.158 |
Token | like |
Token ID | 588 |
Feature activation | -1.19e-01 |
Pos logit contributions | |
bled | +0.422 |
ered | +0.403 |
izes | +0.378 |
eking | +0.376 |
uffed | +0.372 |
Neg logit contributions | |
igsaw | -0.258 |
oak | -0.234 |
olive | -0.227 |
magazine | -0.218 |
elephant | -0.212 |
Token | the |
Token ID | 262 |
Feature activation | +1.44e-01 |
Pos logit contributions | |
bled | +0.258 |
ered | +0.256 |
izes | +0.233 |
vowed | +0.223 |
eking | +0.223 |
Neg logit contributions | |
igsaw | -0.195 |
oak | -0.172 |
magazine | -0.167 |
newspaper | -0.164 |
pine | -0.154 |
Token | wardrobe |
Token ID | 35593 |
Feature activation | +3.85e-02 |
Pos logit contributions | |
igsaw | +0.240 |
oak | +0.206 |
magazine | +0.194 |
newspaper | +0.190 |
pine | +0.180 |
Neg logit contributions | |
ered | -0.312 |
bled | -0.311 |
izes | -0.279 |
vowed | -0.276 |
hadn | -0.271 |
Token | all |
Token ID | 477 |
Feature activation | -1.12e-01 |
Pos logit contributions | |
ered | +0.107 |
bled | +0.103 |
vowed | +0.092 |
izes | +0.091 |
apologised | +0.090 |
Neg logit contributions | |
igsaw | -0.082 |
oak | -0.069 |
magazine | -0.064 |
newspaper | -0.064 |
pine | -0.059 |
Token | sorts |
Token ID | 10524 |
Feature activation | -3.67e-02 |
Pos logit contributions | |
ered | +0.213 |
bled | +0.188 |
vowed | +0.176 |
izes | +0.168 |
apologised | +0.165 |
Neg logit contributions | |
igsaw | -0.217 |
oak | -0.191 |
magazine | -0.188 |
newspaper | -0.187 |
Mic | -0.184 |
Token | of |
Token ID | 286 |
Feature activation | +3.59e-02 |
Pos logit contributions | |
ered | +0.083 |
bled | +0.082 |
izes | +0.075 |
vowed | +0.074 |
hadn | +0.074 |
Neg logit contributions | |
igsaw | -0.056 |
oak | -0.047 |
magazine | -0.043 |
newspaper | -0.042 |
pine | -0.040 |
Token | things |
Token ID | 1243 |
Feature activation | +1.19e-02 |
Pos logit contributions | |
igsaw | +0.060 |
oak | +0.051 |
magazine | +0.048 |
newspaper | +0.047 |
pine | +0.044 |
Neg logit contributions | |
ered | -0.077 |
bled | -0.075 |
izes | -0.068 |
vowed | -0.067 |
hadn | -0.067 |
Token | they |
Token ID | 484 |
Feature activation | -2.04e-01 |
Pos logit contributions | |
igsaw | +0.016 |
oak | +0.014 |
magazine | +0.013 |
newspaper | +0.012 |
pine | +0.011 |
Neg logit contributions | |
bled | -0.029 |
ered | -0.029 |
izes | -0.026 |
vowed | -0.026 |
hadn | -0.025 |
Token | hadn |
Token ID | 8020 |
Feature activation | -3.46e-01 |
Pos logit contributions | |
bled | +0.373 |
ered | +0.344 |
izes | +0.310 |
uffed | +0.300 |
eking | +0.295 |
Neg logit contributions | |
oak | -0.390 |
newspaper | -0.361 |
igsaw | -0.356 |
magazine | -0.354 |
cane | -0.351 |
Token | 't |
Token ID | 470 |
Feature activation | -2.11e-01 |
Pos logit contributions | |
bled | +0.556 |
ered | +0.542 |
izes | +0.494 |
apologised | +0.475 |
vowed | +0.470 |
Neg logit contributions | |
igsaw | -0.756 |
oak | -0.671 |
magazine | -0.636 |
newspaper | -0.625 |
pine | -0.605 |
Token | seen |
Token ID | 1775 |
Feature activation | -1.28e-01 |
Pos logit contributions | |
bled | +0.440 |
ered | +0.396 |
izes | +0.375 |
uffed | +0.365 |
itates | +0.350 |
Neg logit contributions | |
igsaw | -0.350 |
oak | -0.329 |
newspaper | -0.314 |
pine | -0.302 |
magazine | -0.302 |
Token | before |
Token ID | 878 |
Feature activation | -2.72e-02 |
Pos logit contributions | |
ered | +0.289 |
bled | +0.288 |
izes | +0.261 |
vowed | +0.259 |
apologised | +0.255 |
Neg logit contributions | |
igsaw | -0.193 |
oak | -0.163 |
magazine | -0.156 |
newspaper | -0.153 |
pine | -0.142 |
Token | . |
Token ID | 13 |
Feature activation | +2.15e-02 |
Pos logit contributions | |
bled | +0.076 |
ered | +0.074 |
izes | +0.069 |
apologised | +0.068 |
eking | +0.067 |
Neg logit contributions | |
igsaw | -0.026 |
oak | -0.021 |
magazine | -0.020 |
newspaper | -0.020 |
ant | -0.017 |
Token | Together |
Token ID | 17083 |
Feature activation | -1.96e-01 |
Pos logit contributions | |
igsaw | +0.034 |
oak | +0.028 |
magazine | +0.027 |
newspaper | +0.026 |
pine | +0.025 |
Neg logit contributions | |
ered | -0.048 |
bled | -0.047 |
izes | -0.043 |
vowed | -0.042 |
eking | -0.042 |
Token |
|
Token ID | 198 |
Feature activation | -5.76e-02 |
Pos logit contributions | |
ered | +0.143 |
bled | +0.143 |
eking | +0.130 |
izes | +0.129 |
vowed | +0.129 |
Neg logit contributions | |
igsaw | -0.071 |
oak | -0.058 |
newspaper | -0.056 |
magazine | -0.054 |
pine | -0.054 |
Token | The |
Token ID | 464 |
Feature activation | +4.91e-02 |
Pos logit contributions | |
ered | +0.124 |
bled | +0.122 |
icked | +0.109 |
vowed | +0.107 |
eking | +0.107 |
Neg logit contributions | |
igsaw | -0.093 |
oak | -0.082 |
newspaper | -0.079 |
magazine | -0.078 |
pine | -0.076 |
Token | little |
Token ID | 1310 |
Feature activation | +1.75e-01 |
Pos logit contributions | |
igsaw | +0.075 |
oak | +0.063 |
magazine | +0.058 |
newspaper | +0.057 |
pine | +0.054 |
Neg logit contributions | |
ered | -0.113 |
bled | -0.112 |
izes | -0.101 |
vowed | -0.101 |
hadn | -0.099 |
Token | dog |
Token ID | 3290 |
Feature activation | +1.23e-01 |
Pos logit contributions | |
igsaw | +0.276 |
oak | +0.141 |
Mic | +0.132 |
magazine | +0.124 |
mushroom | +0.115 |
Neg logit contributions | |
bled | -0.412 |
ered | -0.394 |
vowed | -0.385 |
izes | -0.384 |
stayed | -0.372 |
Token | still |
Token ID | 991 |
Feature activation | -2.18e-01 |
Pos logit contributions | |
oak | +0.220 |
dog | +0.213 |
cat | +0.208 |
pine | +0.207 |
lion | +0.200 |
Neg logit contributions | |
bled | -0.249 |
ered | -0.207 |
eking | -0.195 |
bling | -0.194 |
vere | -0.182 |
Token | wouldn |
Token ID | 3636 |
Feature activation | -3.46e-01 |
Pos logit contributions | |
bled | +0.468 |
bling | +0.376 |
ered | +0.369 |
eking | +0.354 |
vere | +0.345 |
Neg logit contributions | |
oak | -0.390 |
pine | -0.376 |
dog | -0.367 |
cane | -0.364 |
newspaper | -0.362 |
Token | 't |
Token ID | 470 |
Feature activation | -2.14e-01 |
Pos logit contributions | |
bled | +0.511 |
ered | +0.509 |
izes | +0.439 |
vowed | +0.430 |
eking | +0.429 |
Neg logit contributions | |
igsaw | -0.780 |
oak | -0.716 |
magazine | -0.669 |
newspaper | -0.652 |
ant | -0.652 |
Token | bud |
Token ID | 3600 |
Feature activation | -6.54e-02 |
Pos logit contributions | |
bled | +0.410 |
ered | +0.386 |
uffed | +0.335 |
izes | +0.335 |
eking | +0.334 |
Neg logit contributions | |
igsaw | -0.404 |
oak | -0.381 |
magazine | -0.343 |
pine | -0.341 |
newspaper | -0.338 |
Token | ge |
Token ID | 469 |
Feature activation | -6.24e-02 |
Pos logit contributions | |
ered | +0.125 |
bled | +0.124 |
izes | +0.110 |
vowed | +0.110 |
hadn | +0.108 |
Neg logit contributions | |
igsaw | -0.122 |
oak | -0.107 |
magazine | -0.100 |
newspaper | -0.099 |
pine | -0.096 |
Token | . |
Token ID | 13 |
Feature activation | +1.82e-02 |
Pos logit contributions | |
ered | +0.152 |
bled | +0.151 |
izes | +0.138 |
vowed | +0.136 |
apologised | +0.135 |
Neg logit contributions | |
igsaw | -0.080 |
oak | -0.068 |
magazine | -0.064 |
newspaper | -0.061 |
ant | -0.058 |
Token | Finally |
Token ID | 9461 |
Feature activation | -2.05e-01 |
Pos logit contributions | |
igsaw | +0.032 |
oak | +0.027 |
magazine | +0.025 |
newspaper | +0.025 |
pine | +0.024 |
Neg logit contributions | |
ered | -0.038 |
bled | -0.037 |
izes | -0.034 |
vowed | -0.033 |
hadn | -0.033 |
Token | looked |
Token ID | 3114 |
Feature activation | -3.02e-01 |
Pos logit contributions | |
bled | +0.347 |
ered | +0.331 |
vere | +0.300 |
izes | +0.299 |
uffed | +0.290 |
Neg logit contributions | |
igsaw | -0.466 |
oak | -0.464 |
pine | -0.434 |
magazine | -0.427 |
newspaper | -0.425 |
Token | closely |
Token ID | 7173 |
Feature activation | -6.73e-02 |
Pos logit contributions | |
ered | +0.478 |
bled | +0.473 |
izes | +0.414 |
vowed | +0.409 |
apologised | +0.394 |
Neg logit contributions | |
igsaw | -0.664 |
oak | -0.590 |
newspaper | -0.553 |
magazine | -0.551 |
pine | -0.535 |
Token | at |
Token ID | 379 |
Feature activation | -2.05e-03 |
Pos logit contributions | |
ered | +0.137 |
bled | +0.136 |
izes | +0.121 |
vowed | +0.121 |
hadn | +0.119 |
Neg logit contributions | |
igsaw | -0.120 |
oak | -0.102 |
magazine | -0.094 |
newspaper | -0.092 |
pine | -0.090 |
Token | things |
Token ID | 1243 |
Feature activation | +8.14e-03 |
Pos logit contributions | |
ered | +0.004 |
bled | +0.004 |
vowed | +0.004 |
hadn | +0.004 |
izes | +0.004 |
Neg logit contributions | |
igsaw | -0.004 |
oak | -0.003 |
magazine | -0.003 |
newspaper | -0.003 |
bull | -0.003 |
Token | that |
Token ID | 326 |
Feature activation | +1.58e-02 |
Pos logit contributions | |
igsaw | +0.012 |
oak | +0.010 |
magazine | +0.009 |
newspaper | +0.009 |
pine | +0.008 |
Neg logit contributions | |
ered | -0.019 |
bled | -0.019 |
izes | -0.017 |
vowed | -0.017 |
eking | -0.017 |
Token | seemed |
Token ID | 3947 |
Feature activation | -3.46e-01 |
Pos logit contributions | |
igsaw | +0.034 |
oak | +0.030 |
magazine | +0.028 |
newspaper | +0.028 |
pine | +0.027 |
Neg logit contributions | |
ered | -0.026 |
bled | -0.025 |
hadn | -0.022 |
vowed | -0.022 |
izes | -0.022 |
Token | dull |
Token ID | 19222 |
Feature activation | +2.38e-02 |
Pos logit contributions | |
ered | +0.781 |
bled | +0.775 |
izes | +0.701 |
vowed | +0.692 |
eking | +0.681 |
Neg logit contributions | |
igsaw | -0.535 |
oak | -0.450 |
magazine | -0.420 |
newspaper | -0.413 |
pine | -0.392 |
Token | , |
Token ID | 11 |
Feature activation | +6.91e-02 |
Pos logit contributions | |
igsaw | +0.031 |
oak | +0.026 |
magazine | +0.024 |
newspaper | +0.024 |
pine | +0.021 |
Neg logit contributions | |
ered | -0.058 |
bled | -0.058 |
izes | -0.053 |
vowed | -0.052 |
eking | -0.052 |
Token | because |
Token ID | 780 |
Feature activation | -1.08e-01 |
Pos logit contributions | |
igsaw | +0.113 |
oak | +0.096 |
magazine | +0.090 |
newspaper | +0.089 |
pine | +0.085 |
Neg logit contributions | |
ered | -0.150 |
bled | -0.148 |
izes | -0.133 |
vowed | -0.132 |
hadn | -0.130 |
Token | she |
Token ID | 673 |
Feature activation | -1.61e-01 |
Pos logit contributions | |
ered | +0.207 |
bled | +0.205 |
izes | +0.182 |
vowed | +0.180 |
hadn | +0.178 |
Neg logit contributions | |
igsaw | -0.204 |
oak | -0.177 |
magazine | -0.167 |
newspaper | -0.165 |
pine | -0.159 |
Token | knew |
Token ID | 2993 |
Feature activation | -2.27e-01 |
Pos logit contributions | |
bled | +0.233 |
ered | +0.228 |
eking | +0.197 |
izes | +0.190 |
vere | +0.182 |
Neg logit contributions | |
igsaw | -0.366 |
oak | -0.342 |
magazine | -0.334 |
ant | -0.330 |
newspaper | -0.326 |
Token | get |
Token ID | 651 |
Feature activation | -2.08e-01 |
Pos logit contributions | |
igsaw | +0.096 |
oak | +0.092 |
pine | +0.084 |
newspaper | +0.082 |
ant | +0.081 |
Neg logit contributions | |
bled | -0.102 |
ered | -0.092 |
izes | -0.087 |
icked | -0.085 |
uffed | -0.083 |
Token | scared |
Token ID | 12008 |
Feature activation | -1.19e-01 |
Pos logit contributions | |
bled | +0.392 |
ered | +0.384 |
izes | +0.356 |
vowed | +0.342 |
hadn | +0.336 |
Neg logit contributions | |
igsaw | -0.373 |
oak | -0.353 |
pine | -0.328 |
magazine | -0.323 |
ant | -0.320 |
Token | and |
Token ID | 290 |
Feature activation | +3.16e-03 |
Pos logit contributions | |
ered | +0.235 |
bled | +0.218 |
hadn | +0.203 |
vowed | +0.202 |
apologised | +0.202 |
Neg logit contributions | |
igsaw | -0.223 |
oak | -0.177 |
magazine | -0.176 |
newspaper | -0.169 |
dictionary | -0.158 |
Token | wished |
Token ID | 16555 |
Feature activation | -2.28e-01 |
Pos logit contributions | |
oak | +0.006 |
pine | +0.005 |
ant | +0.005 |
dog | +0.005 |
cat | +0.005 |
Neg logit contributions | |
bled | -0.006 |
bling | -0.005 |
vere | -0.005 |
ered | -0.005 |
its | -0.005 |
Token | she |
Token ID | 673 |
Feature activation | -1.54e-01 |
Pos logit contributions | |
ered | +0.494 |
bled | +0.484 |
izes | +0.445 |
vowed | +0.436 |
apologised | +0.429 |
Neg logit contributions | |
igsaw | -0.367 |
oak | -0.314 |
magazine | -0.295 |
newspaper | -0.293 |
pine | -0.281 |
Token | hadn |
Token ID | 8020 |
Feature activation | -3.46e-01 |
Pos logit contributions | |
bled | +0.288 |
izes | +0.268 |
eking | +0.259 |
its | +0.257 |
ered | +0.256 |
Neg logit contributions | |
oak | -0.262 |
cat | -0.256 |
dog | -0.249 |
pine | -0.239 |
dictionary | -0.239 |
Token | 't |
Token ID | 470 |
Feature activation | -2.09e-01 |
Pos logit contributions | |
ered | +0.478 |
bled | +0.473 |
izes | +0.407 |
vowed | +0.395 |
apologised | +0.387 |
Neg logit contributions | |
igsaw | -0.830 |
oak | -0.748 |
magazine | -0.715 |
newspaper | -0.707 |
pine | -0.691 |
Token | gone |
Token ID | 3750 |
Feature activation | -1.85e-01 |
Pos logit contributions | |
bled | +0.398 |
izes | +0.347 |
ered | +0.340 |
uffed | +0.323 |
apologised | +0.301 |
Neg logit contributions | |
oak | -0.392 |
igsaw | -0.388 |
pine | -0.374 |
ant | -0.356 |
newspaper | -0.352 |
Token | outside |
Token ID | 2354 |
Feature activation | +4.10e-02 |
Pos logit contributions | |
bled | +0.406 |
ered | +0.404 |
vowed | +0.370 |
izes | +0.369 |
apologised | +0.368 |
Neg logit contributions | |
igsaw | -0.271 |
oak | -0.241 |
magazine | -0.221 |
newspaper | -0.219 |
pine | -0.214 |
Token | . |
Token ID | 13 |
Feature activation | +2.03e-02 |
Pos logit contributions | |
igsaw | +0.065 |
oak | +0.055 |
magazine | +0.051 |
newspaper | +0.050 |
pine | +0.048 |
Neg logit contributions | |
ered | -0.090 |
bled | -0.089 |
izes | -0.081 |
vowed | -0.080 |
hadn | -0.079 |
Token | Suddenly |
Token ID | 24975 |
Feature activation | -1.74e-01 |
Pos logit contributions | |
igsaw | +0.035 |
oak | +0.030 |
magazine | +0.028 |
newspaper | +0.028 |
pine | +0.027 |
Neg logit contributions | |
ered | -0.042 |
bled | -0.041 |
vowed | -0.037 |
izes | -0.037 |
hadn | -0.037 |
Token | back |
Token ID | 736 |
Feature activation | +2.57e-03 |
Pos logit contributions | |
bled | +0.725 |
ered | +0.721 |
eking | +0.680 |
eling | +0.658 |
izes | +0.655 |
Neg logit contributions | |
igsaw | -0.279 |
newspaper | -0.209 |
magazine | -0.209 |
oak | -0.201 |
dictionary | -0.180 |
Token | and |
Token ID | 290 |
Feature activation | -2.39e-04 |
Pos logit contributions | |
igsaw | +0.004 |
oak | +0.003 |
magazine | +0.003 |
newspaper | +0.003 |
bull | +0.003 |
Neg logit contributions | |
ered | -0.006 |
bled | -0.005 |
vowed | -0.005 |
izes | -0.005 |
wouldn | -0.005 |
Token | forth |
Token ID | 6071 |
Feature activation | -1.45e-01 |
Pos logit contributions | |
bled | +0.001 |
ered | +0.000 |
hes | +0.000 |
izes | +0.000 |
icked | +0.000 |
Neg logit contributions | |
oak | -0.000 |
igsaw | -0.000 |
magazine | -0.000 |
elephant | -0.000 |
newspaper | -0.000 |
Token | and |
Token ID | 290 |
Feature activation | -1.42e-03 |
Pos logit contributions | |
ered | +0.403 |
bled | +0.393 |
izes | +0.372 |
eling | +0.370 |
eking | +0.368 |
Neg logit contributions | |
igsaw | -0.112 |
oak | -0.090 |
magazine | -0.085 |
newspaper | -0.081 |
cane | -0.071 |
Token | both |
Token ID | 1111 |
Feature activation | -1.57e-01 |
Pos logit contributions | |
bled | +0.003 |
ered | +0.002 |
izes | +0.002 |
icked | +0.002 |
uffed | +0.002 |
Neg logit contributions | |
igsaw | -0.003 |
oak | -0.003 |
newspaper | -0.002 |
magazine | -0.002 |
pine | -0.002 |
Token | seemed |
Token ID | 3947 |
Feature activation | -3.46e-01 |
Pos logit contributions | |
bled | +0.248 |
ered | +0.227 |
izes | +0.215 |
eking | +0.214 |
vere | +0.195 |
Neg logit contributions | |
igsaw | -0.337 |
oak | -0.312 |
ant | -0.292 |
pine | -0.286 |
olive | -0.284 |
Token | to |
Token ID | 284 |
Feature activation | +4.28e-02 |
Pos logit contributions | |
ered | +0.835 |
bled | +0.828 |
izes | +0.759 |
eking | +0.729 |
vowed | +0.722 |
Neg logit contributions | |
igsaw | -0.482 |
oak | -0.404 |
magazine | -0.375 |
ant | -0.366 |
newspaper | -0.359 |
Token | be |
Token ID | 307 |
Feature activation | -2.49e-01 |
Pos logit contributions | |
igsaw | +0.066 |
oak | +0.065 |
pine | +0.055 |
magazine | +0.054 |
ant | +0.053 |
Neg logit contributions | |
bled | -0.098 |
uffed | -0.086 |
icked | -0.084 |
ered | -0.084 |
izes | -0.082 |
Token | upset |
Token ID | 9247 |
Feature activation | -8.17e-02 |
Pos logit contributions | |
ered | +0.445 |
bled | +0.440 |
izes | +0.398 |
its | +0.390 |
uffed | +0.388 |
Neg logit contributions | |
igsaw | -0.425 |
magazine | -0.385 |
oak | -0.379 |
cop | -0.355 |
ant | -0.354 |
Token | . |
Token ID | 13 |
Feature activation | +1.73e-02 |
Pos logit contributions | |
bled | +0.222 |
ered | +0.222 |
izes | +0.205 |
eking | +0.202 |
vowed | +0.201 |
Neg logit contributions | |
igsaw | -0.079 |
oak | -0.059 |
magazine | -0.056 |
newspaper | -0.053 |
ant | -0.052 |
Token |
|
Token ID | 198 |
Feature activation | -6.18e-02 |
Pos logit contributions | |
igsaw | +0.028 |
oak | +0.024 |
magazine | +0.022 |
newspaper | +0.022 |
pine | +0.021 |
Neg logit contributions | |
bled | -0.038 |
ered | -0.038 |
izes | -0.034 |
vowed | -0.033 |
eking | -0.033 |
Token | 't |
Token ID | 470 |
Feature activation | -2.11e-01 |
Pos logit contributions | |
ered | +0.530 |
bled | +0.521 |
izes | +0.462 |
eking | +0.458 |
vowed | +0.453 |
Neg logit contributions | |
igsaw | -0.745 |
oak | -0.679 |
magazine | -0.643 |
newspaper | -0.620 |
pine | -0.613 |
Token | turn |
Token ID | 1210 |
Feature activation | -1.22e-01 |
Pos logit contributions | |
bled | +0.446 |
ered | +0.423 |
vowed | +0.387 |
izes | +0.387 |
eking | +0.379 |
Neg logit contributions | |
igsaw | -0.344 |
oak | -0.312 |
magazine | -0.293 |
newspaper | -0.280 |
automobile | -0.277 |
Token | and |
Token ID | 290 |
Feature activation | -1.64e-03 |
Pos logit contributions | |
ered | +0.242 |
bled | +0.241 |
izes | +0.215 |
vowed | +0.213 |
hadn | +0.208 |
Neg logit contributions | |
igsaw | -0.218 |
oak | -0.188 |
magazine | -0.180 |
newspaper | -0.175 |
pine | -0.168 |
Token | the |
Token ID | 262 |
Feature activation | +1.29e-01 |
Pos logit contributions | |
bled | +0.004 |
izes | +0.004 |
itates | +0.004 |
ered | +0.004 |
eking | +0.004 |
Neg logit contributions | |
oak | -0.002 |
igsaw | -0.002 |
pine | -0.002 |
automobile | -0.002 |
magazine | -0.002 |
Token | doors |
Token ID | 8215 |
Feature activation | -9.80e-02 |
Pos logit contributions | |
igsaw | +0.136 |
oak | +0.119 |
magazine | +0.111 |
newspaper | +0.107 |
automobile | +0.102 |
Neg logit contributions | |
ered | -0.346 |
bled | -0.343 |
izes | -0.316 |
vowed | -0.313 |
eking | -0.310 |
Token | wouldn |
Token ID | 3636 |
Feature activation | -3.46e-01 |
Pos logit contributions | |
bled | +0.167 |
ered | +0.147 |
izes | +0.147 |
aded | +0.146 |
eking | +0.140 |
Neg logit contributions | |
oak | -0.169 |
automobile | -0.164 |
igsaw | -0.162 |
magazine | -0.157 |
bike | -0.157 |
Token | 't |
Token ID | 470 |
Feature activation | -2.14e-01 |
Pos logit contributions | |
ered | +0.549 |
bled | +0.541 |
izes | +0.496 |
eking | +0.494 |
apologised | +0.488 |
Neg logit contributions | |
igsaw | -0.710 |
oak | -0.650 |
magazine | -0.608 |
pine | -0.587 |
newspaper | -0.580 |
Token | open |
Token ID | 1280 |
Feature activation | -3.31e-03 |
Pos logit contributions | |
bled | +0.465 |
ered | +0.432 |
izes | +0.406 |
vowed | +0.404 |
uffed | +0.400 |
Neg logit contributions | |
igsaw | -0.328 |
oak | -0.298 |
magazine | -0.279 |
automobile | -0.270 |
newspaper | -0.264 |
Token | . |
Token ID | 13 |
Feature activation | +1.53e-02 |
Pos logit contributions | |
ered | +0.008 |
bled | +0.007 |
izes | +0.007 |
vowed | +0.007 |
apologised | +0.007 |
Neg logit contributions | |
igsaw | -0.005 |
oak | -0.004 |
magazine | -0.004 |
newspaper | -0.004 |
pine | -0.004 |
Token | Tim |
Token ID | 5045 |
Feature activation | -8.03e-02 |
Pos logit contributions | |
igsaw | +0.026 |
oak | +0.021 |
magazine | +0.020 |
newspaper | +0.020 |
pine | +0.019 |
Neg logit contributions | |
ered | -0.033 |
bled | -0.032 |
hadn | -0.029 |
vowed | -0.029 |
apologised | -0.028 |
Token | my |
Token ID | 1820 |
Feature activation | -8.60e-02 |
Pos logit contributions | |
ered | +0.171 |
bled | +0.165 |
vowed | +0.151 |
hadn | +0.150 |
izes | +0.150 |
Neg logit contributions | |
igsaw | -0.139 |
oak | -0.115 |
magazine | -0.108 |
newspaper | -0.106 |
pine | -0.101 |
Token | sad |
Token ID | 6507 |
Feature activation | -7.94e-02 |
Pos logit contributions | |
ered | +0.531 |
bled | +0.529 |
izes | +0.500 |
vowed | +0.476 |
hadn | +0.452 |
Neg logit contributions | |
igsaw | -0.439 |
oak | -0.419 |
magazine | -0.414 |
ant | -0.404 |
newspaper | -0.381 |
Token | and |
Token ID | 290 |
Feature activation | +1.03e-03 |
Pos logit contributions | |
ered | +0.189 |
bled | +0.188 |
izes | +0.171 |
vowed | +0.170 |
hadn | +0.168 |
Neg logit contributions | |
igsaw | -0.113 |
oak | -0.093 |
magazine | -0.086 |
newspaper | -0.085 |
pine | -0.080 |
Token | worried |
Token ID | 7960 |
Feature activation | -1.30e-01 |
Pos logit contributions | |
oak | +0.002 |
magazine | +0.002 |
newspaper | +0.002 |
olive | +0.002 |
cat | +0.002 |
Neg logit contributions | |
bled | -0.002 |
ered | -0.001 |
vere | -0.001 |
bling | -0.001 |
its | -0.001 |
Token | that |
Token ID | 326 |
Feature activation | +1.78e-02 |
Pos logit contributions | |
bled | +0.304 |
ered | +0.302 |
izes | +0.280 |
vowed | +0.273 |
eking | +0.270 |
Neg logit contributions | |
igsaw | -0.181 |
magazine | -0.150 |
oak | -0.150 |
newspaper | -0.147 |
olive | -0.134 |
Token | she |
Token ID | 673 |
Feature activation | -1.57e-01 |
Pos logit contributions | |
igsaw | +0.026 |
oak | +0.023 |
magazine | +0.022 |
newspaper | +0.021 |
pine | +0.020 |
Neg logit contributions | |
bled | -0.041 |
ered | -0.041 |
izes | -0.037 |
eking | -0.037 |
uffed | -0.036 |
Token | wouldn |
Token ID | 3636 |
Feature activation | -3.45e-01 |
Pos logit contributions | |
bled | +0.288 |
ered | +0.253 |
eking | +0.229 |
vere | +0.229 |
bling | +0.227 |
Neg logit contributions | |
cat | -0.298 |
magazine | -0.288 |
newspaper | -0.285 |
cane | -0.278 |
dictionary | -0.276 |
Token | 't |
Token ID | 470 |
Feature activation | -2.14e-01 |
Pos logit contributions | |
ered | +0.476 |
bled | +0.468 |
izes | +0.402 |
vowed | +0.393 |
eking | +0.386 |
Neg logit contributions | |
igsaw | -0.827 |
oak | -0.746 |
magazine | -0.712 |
newspaper | -0.700 |
pine | -0.684 |
Token | be |
Token ID | 307 |
Feature activation | -2.47e-01 |
Pos logit contributions | |
bled | +0.395 |
ered | +0.378 |
izes | +0.336 |
vowed | +0.333 |
uffed | +0.323 |
Neg logit contributions | |
igsaw | -0.417 |
oak | -0.387 |
magazine | -0.358 |
newspaper | -0.352 |
pine | -0.348 |
Token | able |
Token ID | 1498 |
Feature activation | -1.04e-01 |
Pos logit contributions | |
ered | +0.340 |
bled | +0.337 |
izes | +0.291 |
vowed | +0.284 |
hadn | +0.278 |
Neg logit contributions | |
igsaw | -0.587 |
oak | -0.533 |
magazine | -0.519 |
newspaper | -0.507 |
ant | -0.499 |
Token | to |
Token ID | 284 |
Feature activation | +4.45e-02 |
Pos logit contributions | |
ered | +0.277 |
bled | +0.276 |
izes | +0.254 |
vowed | +0.251 |
hadn | +0.247 |
Neg logit contributions | |
igsaw | -0.117 |
oak | -0.094 |
magazine | -0.085 |
newspaper | -0.082 |
pine | -0.077 |
Token | sleep |
Token ID | 3993 |
Feature activation | -8.74e-02 |
Pos logit contributions | |
igsaw | +0.070 |
oak | +0.067 |
pine | +0.060 |
magazine | +0.060 |
newspaper | +0.058 |
Neg logit contributions | |
bled | -0.095 |
ered | -0.089 |
izes | -0.086 |
vowed | -0.082 |
uffed | -0.082 |
Token | had |
Token ID | 550 |
Feature activation | -2.87e-01 |
Pos logit contributions | |
oak | +0.081 |
elephant | +0.079 |
pine | +0.076 |
cat | +0.076 |
lion | +0.075 |
Neg logit contributions | |
bled | -0.078 |
izes | -0.070 |
ered | -0.069 |
vere | -0.067 |
itates | -0.064 |
Token | warned |
Token ID | 7728 |
Feature activation | -2.28e-01 |
Pos logit contributions | |
bled | +0.549 |
ered | +0.504 |
izes | +0.495 |
itates | +0.478 |
eking | +0.440 |
Neg logit contributions | |
igsaw | -0.502 |
oak | -0.497 |
newspaper | -0.486 |
pine | -0.473 |
magazine | -0.467 |
Token | her |
Token ID | 607 |
Feature activation | +2.16e-02 |
Pos logit contributions | |
bled | +0.532 |
ered | +0.524 |
izes | +0.482 |
vowed | +0.476 |
hadn | +0.469 |
Neg logit contributions | |
igsaw | -0.332 |
oak | -0.275 |
magazine | -0.258 |
newspaper | -0.255 |
pine | -0.238 |
Token | but |
Token ID | 475 |
Feature activation | -1.57e-01 |
Pos logit contributions | |
igsaw | +0.028 |
oak | +0.024 |
magazine | +0.022 |
newspaper | +0.022 |
ant | +0.021 |
Neg logit contributions | |
bled | -0.053 |
ered | -0.051 |
izes | -0.048 |
eking | -0.047 |
vowed | -0.047 |
Token | she |
Token ID | 673 |
Feature activation | -1.54e-01 |
Pos logit contributions | |
bled | +0.392 |
ered | +0.371 |
izes | +0.354 |
eking | +0.351 |
icked | +0.346 |
Neg logit contributions | |
oak | -0.185 |
igsaw | -0.177 |
ant | -0.163 |
newspaper | -0.152 |
pine | -0.152 |
Token | hadn |
Token ID | 8020 |
Feature activation | -3.45e-01 |
Pos logit contributions | |
bled | +0.263 |
eking | +0.246 |
ered | +0.221 |
vere | +0.219 |
izes | +0.218 |
Neg logit contributions | |
cat | -0.278 |
ant | -0.272 |
oak | -0.272 |
dog | -0.265 |
mouse | -0.259 |
Token | â |
Token ID | 22940 |
Feature activation | -1.03e-01 |
Pos logit contributions | |
ered | +0.442 |
bled | +0.435 |
izes | +0.362 |
vowed | +0.357 |
hadn | +0.349 |
Neg logit contributions | |
igsaw | -0.867 |
oak | -0.782 |
magazine | -0.752 |
newspaper | -0.745 |
pine | -0.723 |
Token | € |
Token ID | 26391 |
Feature activation | -4.45e-03 |
Pos logit contributions | |
bled | +0.186 |
ered | +0.186 |
izes | +0.162 |
vowed | +0.161 |
hadn | +0.159 |
Neg logit contributions | |
igsaw | -0.207 |
oak | -0.179 |
magazine | -0.170 |
newspaper | -0.168 |
pine | -0.161 |
Token | ™ |
Token ID | 8151 |
Feature activation | +9.37e-03 |
Pos logit contributions | |
ered | +0.009 |
bled | +0.009 |
izes | +0.008 |
vowed | +0.007 |
apologised | +0.007 |
Neg logit contributions | |
igsaw | -0.008 |
oak | -0.007 |
magazine | -0.007 |
newspaper | -0.007 |
pine | -0.006 |
Token | t |
Token ID | 83 |
Feature activation | -1.71e-01 |
Pos logit contributions | |
igsaw | +0.025 |
oak | +0.023 |
newspaper | +0.022 |
magazine | +0.022 |
pine | +0.021 |
Neg logit contributions | |
bled | -0.010 |
ered | -0.010 |
hadn | -0.008 |
vowed | -0.008 |
wouldn | -0.008 |
Token | listened |
Token ID | 16399 |
Feature activation | -2.19e-01 |
Pos logit contributions | |
bled | +0.297 |
ered | +0.278 |
izes | +0.247 |
vowed | +0.234 |
uffed | +0.233 |
Neg logit contributions | |
igsaw | -0.360 |
oak | -0.332 |
newspaper | -0.308 |
magazine | -0.305 |
pine | -0.303 |
Token | Tim |
Token ID | 5045 |
Feature activation | -7.30e-02 |
Pos logit contributions | |
ered | +0.361 |
bled | +0.358 |
izes | +0.331 |
vowed | +0.325 |
eking | +0.321 |
Neg logit contributions | |
igsaw | -0.188 |
oak | -0.156 |
magazine | -0.140 |
newspaper | -0.137 |
pine | -0.131 |
Token | my |
Token ID | 1820 |
Feature activation | -7.48e-02 |
Pos logit contributions | |
bled | +0.178 |
ered | +0.176 |
izes | +0.162 |
eking | +0.157 |
vowed | +0.157 |
Neg logit contributions | |
igsaw | -0.095 |
oak | -0.081 |
magazine | -0.075 |
newspaper | -0.074 |
pine | -0.069 |
Token | knew |
Token ID | 2993 |
Feature activation | -2.21e-01 |
Pos logit contributions | |
bled | +0.113 |
ered | +0.103 |
eking | +0.100 |
vere | +0.099 |
izes | +0.093 |
Neg logit contributions | |
ant | -0.141 |
igsaw | -0.140 |
cat | -0.140 |
oak | -0.140 |
cane | -0.138 |
Token | that |
Token ID | 326 |
Feature activation | +1.87e-02 |
Pos logit contributions | |
ered | +0.425 |
bled | +0.419 |
vowed | +0.372 |
hadn | +0.366 |
izes | +0.362 |
Neg logit contributions | |
igsaw | -0.434 |
oak | -0.363 |
magazine | -0.351 |
newspaper | -0.343 |
bull | -0.322 |
Token | it |
Token ID | 340 |
Feature activation | -9.56e-02 |
Pos logit contributions | |
igsaw | +0.035 |
oak | +0.030 |
magazine | +0.028 |
newspaper | +0.028 |
pine | +0.027 |
Neg logit contributions | |
ered | -0.036 |
bled | -0.036 |
izes | -0.032 |
vowed | -0.032 |
hadn | -0.032 |
Token | wouldn |
Token ID | 3636 |
Feature activation | -3.45e-01 |
Pos logit contributions | |
bled | +0.154 |
ered | +0.148 |
eking | +0.132 |
izes | +0.131 |
vowed | +0.120 |
Neg logit contributions | |
igsaw | -0.205 |
oak | -0.190 |
magazine | -0.187 |
newspaper | -0.180 |
ant | -0.177 |
Token | 't |
Token ID | 470 |
Feature activation | -2.13e-01 |
Pos logit contributions | |
bled | +0.530 |
ered | +0.525 |
izes | +0.465 |
eking | +0.458 |
vowed | +0.455 |
Neg logit contributions | |
igsaw | -0.760 |
oak | -0.693 |
magazine | -0.647 |
ant | -0.624 |
newspaper | -0.619 |
Token | be |
Token ID | 307 |
Feature activation | -2.45e-01 |
Pos logit contributions | |
bled | +0.476 |
uffed | +0.411 |
ered | +0.409 |
icked | +0.395 |
hes | +0.391 |
Neg logit contributions | |
oak | -0.345 |
igsaw | -0.322 |
pine | -0.293 |
magazine | -0.291 |
ant | -0.283 |
Token | the |
Token ID | 262 |
Feature activation | +1.34e-01 |
Pos logit contributions | |
bled | +0.512 |
ered | +0.501 |
izes | +0.463 |
vowed | +0.446 |
hadn | +0.434 |
Neg logit contributions | |
igsaw | -0.408 |
oak | -0.362 |
magazine | -0.341 |
newspaper | -0.332 |
pine | -0.327 |
Token | same |
Token ID | 976 |
Feature activation | -7.84e-02 |
Pos logit contributions | |
igsaw | +0.280 |
oak | +0.232 |
magazine | +0.218 |
newspaper | +0.211 |
pine | +0.211 |
Neg logit contributions | |
bled | -0.223 |
ered | -0.221 |
vowed | -0.198 |
hadn | -0.197 |
izes | -0.196 |
Token | . |
Token ID | 13 |
Feature activation | +1.94e-02 |
Pos logit contributions | |
ered | +0.185 |
bled | +0.184 |
izes | +0.167 |
vowed | +0.166 |
hadn | +0.163 |
Neg logit contributions | |
igsaw | -0.113 |
oak | -0.094 |
magazine | -0.087 |
newspaper | -0.086 |
pine | -0.081 |
Token |
|
Token ID | 198 |
Feature activation | +9.72e-02 |
Pos logit contributions | |
igsaw | +0.159 |
oak | +0.137 |
magazine | +0.125 |
newspaper | +0.124 |
pine | +0.123 |
Neg logit contributions | |
bled | -0.228 |
ered | -0.227 |
izes | -0.205 |
vowed | -0.201 |
eking | -0.201 |
Token | One |
Token ID | 3198 |
Feature activation | +1.04e-01 |
Pos logit contributions | |
igsaw | +0.170 |
oak | +0.147 |
magazine | +0.137 |
newspaper | +0.135 |
pine | +0.131 |
Neg logit contributions | |
ered | -0.199 |
bled | -0.198 |
izes | -0.177 |
vowed | -0.175 |
hadn | -0.172 |
Token | night |
Token ID | 1755 |
Feature activation | +2.21e-01 |
Pos logit contributions | |
igsaw | +0.135 |
oak | +0.104 |
magazine | +0.100 |
newspaper | +0.094 |
olive | +0.083 |
Neg logit contributions | |
ered | -0.260 |
bled | -0.250 |
vowed | -0.236 |
hadn | -0.234 |
izes | -0.233 |
Token | , |
Token ID | 11 |
Feature activation | +2.17e-01 |
Pos logit contributions | |
igsaw | +0.253 |
oak | +0.196 |
magazine | +0.178 |
newspaper | +0.173 |
pine | +0.156 |
Neg logit contributions | |
ered | -0.590 |
bled | -0.585 |
izes | -0.536 |
vowed | -0.534 |
hadn | -0.531 |
Token | the |
Token ID | 262 |
Feature activation | +2.73e-01 |
Pos logit contributions | |
igsaw | +0.320 |
oak | +0.264 |
magazine | +0.246 |
newspaper | +0.241 |
pine | +0.227 |
Neg logit contributions | |
ered | -0.506 |
bled | -0.502 |
izes | -0.455 |
vowed | -0.453 |
hadn | -0.448 |
Token | little |
Token ID | 1310 |
Feature activation | +3.16e-01 |
Pos logit contributions | |
igsaw | +0.277 |
oak | +0.243 |
magazine | +0.212 |
newspaper | +0.210 |
pine | +0.199 |
Neg logit contributions | |
ered | -0.757 |
bled | -0.740 |
izes | -0.688 |
vowed | -0.673 |
icked | -0.673 |
Token | boy |
Token ID | 2933 |
Feature activation | +1.44e-01 |
Pos logit contributions | |
igsaw | +0.453 |
magazine | +0.304 |
oak | +0.303 |
newspaper | +0.294 |
dictionary | +0.267 |
Neg logit contributions | |
ered | -0.758 |
bled | -0.751 |
vowed | -0.707 |
hadn | -0.696 |
izes | -0.688 |
Token | was |
Token ID | 373 |
Feature activation | -1.39e-01 |
Pos logit contributions | |
oak | +0.232 |
pine | +0.225 |
cat | +0.217 |
dog | +0.215 |
lion | +0.213 |
Neg logit contributions | |
bled | -0.308 |
vere | -0.264 |
ered | -0.263 |
izes | -0.258 |
eking | -0.251 |
Token | in |
Token ID | 287 |
Feature activation | +1.24e-01 |
Pos logit contributions | |
bled | +0.286 |
ered | +0.285 |
izes | +0.271 |
vowed | +0.261 |
its | +0.252 |
Neg logit contributions | |
igsaw | -0.191 |
oak | -0.185 |
magazine | -0.181 |
ant | -0.174 |
pine | -0.168 |
Token | his |
Token ID | 465 |
Feature activation | +1.51e-01 |
Pos logit contributions | |
igsaw | +0.224 |
oak | +0.190 |
newspaper | +0.179 |
magazine | +0.177 |
olive | +0.174 |
Neg logit contributions | |
ered | -0.235 |
bled | -0.215 |
hadn | -0.209 |
vowed | -0.209 |
zed | -0.196 |
Token | crib |
Token ID | 48083 |
Feature activation | +2.26e-01 |
Pos logit contributions | |
igsaw | +0.244 |
oak | +0.195 |
newspaper | +0.184 |
magazine | +0.182 |
olive | +0.175 |
Neg logit contributions | |
ered | -0.316 |
bled | -0.307 |
vowed | -0.291 |
hadn | -0.288 |
izes | -0.278 |
Token | a |
Token ID | 257 |
Feature activation | +1.68e-01 |
Pos logit contributions | |
ered | +0.250 |
bled | +0.248 |
izes | +0.224 |
vowed | +0.219 |
hadn | +0.212 |
Neg logit contributions | |
igsaw | -0.258 |
oak | -0.227 |
magazine | -0.215 |
newspaper | -0.210 |
pine | -0.206 |
Token | great |
Token ID | 1049 |
Feature activation | +1.26e-01 |
Pos logit contributions | |
igsaw | +0.408 |
oak | +0.356 |
newspaper | +0.341 |
magazine | +0.339 |
olive | +0.317 |
Neg logit contributions | |
ered | -0.205 |
bled | -0.200 |
vowed | -0.190 |
hadn | -0.189 |
izes | -0.176 |
Token | sign |
Token ID | 1051 |
Feature activation | +1.24e-01 |
Pos logit contributions | |
igsaw | +0.207 |
oak | +0.159 |
magazine | +0.154 |
newspaper | +0.151 |
magazines | +0.133 |
Neg logit contributions | |
ered | -0.244 |
bled | -0.228 |
wouldn | -0.227 |
ok | -0.227 |
vowed | -0.225 |
Token | er |
Token ID | 263 |
Feature activation | +1.60e-01 |
Pos logit contributions | |
igsaw | +0.193 |
oak | +0.164 |
magazine | +0.154 |
newspaper | +0.151 |
pine | +0.143 |
Neg logit contributions | |
ered | -0.277 |
bled | -0.275 |
izes | -0.247 |
vowed | -0.246 |
hadn | -0.242 |
Token | , |
Token ID | 11 |
Feature activation | +1.99e-01 |
Pos logit contributions | |
igsaw | +0.264 |
oak | +0.220 |
magazine | +0.207 |
newspaper | +0.204 |
pine | +0.194 |
Neg logit contributions | |
ered | -0.343 |
bled | -0.340 |
izes | -0.306 |
vowed | -0.303 |
hadn | -0.299 |
Token | little |
Token ID | 1310 |
Feature activation | +3.01e-01 |
Pos logit contributions | |
igsaw | +0.308 |
oak | +0.250 |
magazine | +0.235 |
newspaper | +0.231 |
pine | +0.213 |
Neg logit contributions | |
ered | -0.444 |
bled | -0.438 |
hadn | -0.399 |
vowed | -0.399 |
izes | -0.396 |
Token | rabbit |
Token ID | 22746 |
Feature activation | +1.93e-01 |
Pos logit contributions | |
igsaw | +0.351 |
oak | +0.254 |
magazine | +0.237 |
newspaper | +0.235 |
pine | +0.197 |
Neg logit contributions | |
ered | -0.792 |
bled | -0.779 |
vowed | -0.724 |
izes | -0.719 |
hadn | -0.717 |
Token | . |
Token ID | 13 |
Feature activation | +1.47e-01 |
Pos logit contributions | |
igsaw | +0.322 |
oak | +0.272 |
magazine | +0.255 |
newspaper | +0.252 |
pine | +0.239 |
Neg logit contributions | |
ered | -0.409 |
bled | -0.404 |
izes | -0.365 |
vowed | -0.363 |
hadn | -0.358 |
Token | Your |
Token ID | 3406 |
Feature activation | +2.85e-03 |
Pos logit contributions | |
igsaw | +0.298 |
oak | +0.262 |
magazine | +0.249 |
newspaper | +0.246 |
pine | +0.237 |
Neg logit contributions | |
ered | -0.259 |
bled | -0.256 |
izes | -0.225 |
vowed | -0.223 |
hadn | -0.219 |
Token | name |
Token ID | 1438 |
Feature activation | +5.35e-02 |
Pos logit contributions | |
igsaw | +0.004 |
oak | +0.003 |
magazine | +0.003 |
newspaper | +0.003 |
pine | +0.003 |
Neg logit contributions | |
ered | -0.007 |
bled | -0.007 |
izes | -0.006 |
vowed | -0.006 |
hadn | -0.006 |
Token | will |
Token ID | 481 |
Feature activation | -1.53e-01 |
Pos logit contributions | |
magazine | +0.093 |
igsaw | +0.093 |
oak | +0.091 |
newspaper | +0.088 |
lion | +0.083 |
Neg logit contributions | |
bled | -0.103 |
ered | -0.097 |
eking | -0.090 |
uffed | -0.086 |
apologised | -0.084 |
Token | time |
Token ID | 640 |
Feature activation | +2.38e-01 |
Pos logit contributions | |
igsaw | +0.396 |
oak | +0.362 |
magazine | +0.326 |
newspaper | +0.312 |
olive | +0.302 |
Neg logit contributions | |
ered | -0.348 |
bled | -0.335 |
hadn | -0.319 |
eking | -0.309 |
ok | -0.309 |
Token | , |
Token ID | 11 |
Feature activation | +2.38e-01 |
Pos logit contributions | |
igsaw | +0.370 |
oak | +0.319 |
newspaper | +0.316 |
magazine | +0.312 |
bull | +0.282 |
Neg logit contributions | |
ered | -0.529 |
bled | -0.528 |
hadn | -0.479 |
icked | -0.460 |
uffed | -0.457 |
Token | there |
Token ID | 612 |
Feature activation | +1.14e-01 |
Pos logit contributions | |
igsaw | +0.516 |
oak | +0.452 |
newspaper | +0.428 |
magazine | +0.428 |
pine | +0.404 |
Neg logit contributions | |
bled | -0.408 |
ered | -0.397 |
hadn | -0.337 |
izes | -0.337 |
eking | -0.331 |
Token | was |
Token ID | 373 |
Feature activation | -1.21e-01 |
Pos logit contributions | |
igsaw | +0.280 |
oak | +0.256 |
magazine | +0.245 |
newspaper | +0.243 |
pine | +0.239 |
Neg logit contributions | |
ered | -0.145 |
bled | -0.144 |
izes | -0.124 |
vowed | -0.118 |
eking | -0.118 |
Token | a |
Token ID | 257 |
Feature activation | +1.89e-01 |
Pos logit contributions | |
bled | +0.163 |
hes | +0.143 |
ered | +0.139 |
icked | +0.125 |
sure | +0.122 |
Neg logit contributions | |
igsaw | -0.296 |
oak | -0.265 |
dictionary | -0.240 |
magazine | -0.233 |
cane | -0.232 |
Token | little |
Token ID | 1310 |
Feature activation | +3.27e-01 |
Pos logit contributions | |
igsaw | +0.380 |
oak | +0.271 |
newspaper | +0.254 |
magazine | +0.254 |
olive | +0.236 |
Neg logit contributions | |
bled | -0.318 |
hadn | -0.302 |
also | -0.293 |
still | -0.289 |
sorry | -0.287 |
Token | girl |
Token ID | 2576 |
Feature activation | +1.67e-01 |
Pos logit contributions | |
igsaw | +0.687 |
growth | +0.485 |
opl | +0.467 |
Wed | +0.462 |
newspaper | +0.462 |
Neg logit contributions | |
sorry | -0.425 |
bled | -0.402 |
also | -0.397 |
still | -0.396 |
stayed | -0.379 |
Token | named |
Token ID | 3706 |
Feature activation | +1.27e-01 |
Pos logit contributions | |
igsaw | +0.450 |
oak | +0.345 |
magazine | +0.336 |
newspaper | +0.335 |
bull | +0.319 |
Neg logit contributions | |
ered | -0.220 |
vowed | -0.195 |
bled | -0.193 |
hadn | -0.192 |
never | -0.188 |
Token | Lily |
Token ID | 20037 |
Feature activation | +7.60e-02 |
Pos logit contributions | |
igsaw | +0.221 |
oak | +0.186 |
magazine | +0.179 |
newspaper | +0.178 |
cane | +0.173 |
Neg logit contributions | |
ered | -0.259 |
bled | -0.253 |
hadn | -0.226 |
icked | -0.223 |
izes | -0.218 |
Token | . |
Token ID | 13 |
Feature activation | +1.68e-01 |
Pos logit contributions | |
igsaw | +0.176 |
oak | +0.133 |
magazine | +0.127 |
newspaper | +0.123 |
lime | +0.121 |
Neg logit contributions | |
ered | -0.130 |
hadn | -0.120 |
bled | -0.119 |
vowed | -0.119 |
wouldn | -0.110 |
Token | She |
Token ID | 1375 |
Feature activation | -5.54e-02 |
Pos logit contributions | |
igsaw | +0.345 |
oak | +0.304 |
bull | +0.301 |
newspaper | +0.296 |
goat | +0.291 |
Neg logit contributions | |
bled | -0.248 |
ered | -0.240 |
hadn | -0.240 |
icked | -0.230 |
itates | -0.223 |
Token | <|endoftext|> |
Token ID | 50256 |
Feature activation | +1.90e-01 |
Pos logit contributions | |
igsaw | +0.345 |
oak | +0.292 |
magazine | +0.271 |
newspaper | +0.265 |
pine | +0.255 |
Neg logit contributions | |
ered | -0.481 |
bled | -0.478 |
izes | -0.434 |
eking | -0.427 |
vowed | -0.420 |
Token | Once |
Token ID | 7454 |
Feature activation | +1.83e-01 |
Pos logit contributions | |
igsaw | +0.371 |
oak | +0.327 |
newspaper | +0.306 |
magazine | +0.302 |
ant | +0.291 |
Neg logit contributions | |
ered | -0.342 |
bled | -0.337 |
hadn | -0.305 |
vowed | -0.295 |
izes | -0.289 |
Token | there |
Token ID | 612 |
Feature activation | +1.31e-01 |
Pos logit contributions | |
igsaw | +0.390 |
oak | +0.345 |
magazine | +0.331 |
newspaper | +0.329 |
pine | +0.314 |
Neg logit contributions | |
ered | -0.303 |
bled | -0.302 |
izes | -0.259 |
hadn | -0.259 |
vowed | -0.256 |
Token | was |
Token ID | 373 |
Feature activation | -1.05e-01 |
Pos logit contributions | |
igsaw | +0.324 |
oak | +0.291 |
magazine | +0.280 |
newspaper | +0.277 |
pine | +0.269 |
Neg logit contributions | |
ered | -0.175 |
bled | -0.172 |
izes | -0.145 |
vowed | -0.142 |
hadn | -0.138 |
Token | a |
Token ID | 257 |
Feature activation | +2.05e-01 |
Pos logit contributions | |
bled | +0.167 |
ered | +0.159 |
hes | +0.141 |
hadn | +0.136 |
icked | +0.135 |
Neg logit contributions | |
igsaw | -0.226 |
oak | -0.201 |
magazine | -0.184 |
dictionary | -0.182 |
newspaper | -0.179 |
Token | little |
Token ID | 1310 |
Feature activation | +3.39e-01 |
Pos logit contributions | |
igsaw | +0.422 |
oak | +0.297 |
magazine | +0.283 |
newspaper | +0.278 |
newspapers | +0.268 |
Neg logit contributions | |
bled | -0.319 |
also | -0.317 |
hadn | -0.314 |
still | -0.306 |
forgot | -0.301 |
Token | girl |
Token ID | 2576 |
Feature activation | +1.79e-01 |
Pos logit contributions | |
igsaw | +0.711 |
growth | +0.531 |
opl | +0.489 |
Wed | +0.485 |
newspaper | +0.480 |
Neg logit contributions | |
also | -0.418 |
sorry | -0.418 |
still | -0.417 |
bled | -0.407 |
stayed | -0.376 |
Token | who |
Token ID | 508 |
Feature activation | +5.72e-02 |
Pos logit contributions | |
igsaw | +0.477 |
oak | +0.362 |
magazine | +0.355 |
newspaper | +0.349 |
bull | +0.337 |
Neg logit contributions | |
ered | -0.241 |
vowed | -0.216 |
hadn | -0.214 |
never | -0.211 |
bled | -0.209 |
Token | loved |
Token ID | 6151 |
Feature activation | -6.25e-02 |
Pos logit contributions | |
igsaw | +0.130 |
oak | +0.120 |
magazine | +0.117 |
newspaper | +0.115 |
pine | +0.114 |
Neg logit contributions | |
bled | -0.085 |
ered | -0.082 |
izes | -0.073 |
eking | -0.068 |
vowed | -0.067 |
Token | to |
Token ID | 284 |
Feature activation | +1.99e-01 |
Pos logit contributions | |
ered | +0.102 |
bled | +0.096 |
vowed | +0.091 |
hes | +0.091 |
wouldn | +0.088 |
Neg logit contributions | |
igsaw | -0.126 |
oc | -0.094 |
izer | -0.087 |
ay | -0.084 |
ro | -0.084 |
Token | draw |
Token ID | 3197 |
Feature activation | +4.40e-02 |
Pos logit contributions | |
igsaw | +0.378 |
oak | +0.302 |
magazine | +0.299 |
bull | +0.289 |
newspaper | +0.288 |
Neg logit contributions | |
ered | -0.377 |
bled | -0.331 |
vowed | -0.301 |
hadn | -0.295 |
zed | -0.290 |
Token | mine |
Token ID | 6164 |
Feature activation | +7.56e-02 |
Pos logit contributions | |
igsaw | +0.077 |
oak | +0.071 |
pine | +0.068 |
magazine | +0.067 |
newspaper | +0.067 |
Neg logit contributions | |
bled | -0.080 |
ered | -0.078 |
izes | -0.071 |
vowed | -0.069 |
uffed | -0.068 |
Token | !" |
Token ID | 2474 |
Feature activation | +9.08e-02 |
Pos logit contributions | |
igsaw | +0.097 |
oak | +0.089 |
magazine | +0.079 |
newspaper | +0.079 |
ant | +0.078 |
Neg logit contributions | |
ered | -0.176 |
bled | -0.173 |
apologised | -0.165 |
vowed | -0.163 |
izes | -0.158 |
Token |
|
Token ID | 198 |
Feature activation | +7.93e-02 |
Pos logit contributions | |
igsaw | +0.154 |
oak | +0.131 |
magazine | +0.123 |
newspaper | +0.121 |
pine | +0.115 |
Neg logit contributions | |
ered | -0.190 |
bled | -0.187 |
vowed | -0.169 |
izes | -0.169 |
hadn | -0.167 |
Token |
|
Token ID | 198 |
Feature activation | +7.62e-02 |
Pos logit contributions | |
igsaw | +0.138 |
oak | +0.119 |
magazine | +0.111 |
newspaper | +0.109 |
pine | +0.103 |
Neg logit contributions | |
ered | -0.164 |
bled | -0.161 |
izes | -0.145 |
vowed | -0.144 |
hadn | -0.142 |
Token | The |
Token ID | 464 |
Feature activation | +1.73e-01 |
Pos logit contributions | |
igsaw | +0.139 |
oak | +0.119 |
magazine | +0.113 |
newspaper | +0.110 |
pine | +0.104 |
Neg logit contributions | |
ered | -0.151 |
bled | -0.148 |
izes | -0.134 |
vowed | -0.134 |
hadn | -0.131 |
Token | little |
Token ID | 1310 |
Feature activation | +3.02e-01 |
Pos logit contributions | |
igsaw | +0.192 |
oak | +0.151 |
magazine | +0.135 |
newspaper | +0.132 |
pine | +0.122 |
Neg logit contributions | |
ered | -0.466 |
bled | -0.462 |
izes | -0.425 |
vowed | -0.423 |
hadn | -0.418 |
Token | bear |
Token ID | 6842 |
Feature activation | +1.84e-01 |
Pos logit contributions | |
igsaw | +0.451 |
newspaper | +0.309 |
magazine | +0.303 |
oak | +0.301 |
dictionary | +0.256 |
Neg logit contributions | |
ered | -0.678 |
bled | -0.666 |
hadn | -0.622 |
vowed | -0.617 |
wouldn | -0.612 |
Token | was |
Token ID | 373 |
Feature activation | -1.53e-01 |
Pos logit contributions | |
oak | +0.313 |
igsaw | +0.310 |
par | +0.304 |
pine | +0.301 |
lion | +0.301 |
Neg logit contributions | |
bled | -0.344 |
vere | -0.291 |
bling | -0.281 |
ered | -0.281 |
eking | -0.281 |
Token | scared |
Token ID | 12008 |
Feature activation | -4.35e-03 |
Pos logit contributions | |
bled | +0.277 |
ered | +0.264 |
izes | +0.253 |
vowed | +0.244 |
wouldn | +0.238 |
Neg logit contributions | |
igsaw | -0.252 |
ant | -0.240 |
oak | -0.239 |
pine | -0.231 |
magazine | -0.231 |
Token | and |
Token ID | 290 |
Feature activation | +1.24e-01 |
Pos logit contributions | |
ered | +0.009 |
bled | +0.009 |
vowed | +0.008 |
apologised | +0.008 |
hadn | +0.008 |
Neg logit contributions | |
igsaw | -0.007 |
oak | -0.006 |
magazine | -0.006 |
newspaper | -0.006 |
pine | -0.005 |
Token | started |
Token ID | 2067 |
Feature activation | -7.97e-02 |
Pos logit contributions | |
oak | +0.206 |
pine | +0.191 |
dog | +0.191 |
ant | +0.186 |
bull | +0.183 |
Neg logit contributions | |
bled | -0.233 |
ered | -0.203 |
Eventually | -0.191 |
bling | -0.191 |
eking | -0.190 |
Token | . |
Token ID | 13 |
Feature activation | +1.37e-01 |
Pos logit contributions | |
igsaw | +0.259 |
oak | +0.223 |
magazine | +0.217 |
newspaper | +0.204 |
pine | +0.190 |
Neg logit contributions | |
ered | -0.443 |
bled | -0.440 |
izes | -0.399 |
eking | -0.394 |
vowed | -0.389 |
Token | <|endoftext|> |
Token ID | 50256 |
Feature activation | +1.17e-01 |
Pos logit contributions | |
igsaw | +0.255 |
oak | +0.217 |
magazine | +0.211 |
newspaper | +0.209 |
cane | +0.200 |
Neg logit contributions | |
ered | -0.247 |
bled | -0.246 |
hadn | -0.235 |
vowed | -0.228 |
apologised | -0.220 |
Token | Once |
Token ID | 7454 |
Feature activation | +1.20e-01 |
Pos logit contributions | |
igsaw | +0.237 |
oak | +0.211 |
newspaper | +0.193 |
magazine | +0.190 |
ant | +0.186 |
Neg logit contributions | |
ered | -0.205 |
bled | -0.203 |
hadn | -0.180 |
vowed | -0.177 |
izes | -0.173 |
Token | upon |
Token ID | 2402 |
Feature activation | +1.39e-05 |
Pos logit contributions | |
igsaw | +0.273 |
oak | +0.243 |
magazine | +0.235 |
newspaper | +0.234 |
pine | +0.220 |
Neg logit contributions | |
bled | -0.181 |
ered | -0.179 |
hadn | -0.154 |
izes | -0.152 |
vowed | -0.148 |
Token | a |
Token ID | 257 |
Feature activation | +1.38e-01 |
Pos logit contributions | |
igsaw | +0.000 |
oak | +0.000 |
magazine | +0.000 |
newspaper | +0.000 |
pine | +0.000 |
Neg logit contributions | |
ered | -0.000 |
bled | -0.000 |
izes | -0.000 |
vowed | -0.000 |
hadn | -0.000 |
Token | time |
Token ID | 640 |
Feature activation | +1.79e-01 |
Pos logit contributions | |
igsaw | +0.253 |
oak | +0.231 |
magazine | +0.203 |
newspaper | +0.196 |
olive | +0.194 |
Neg logit contributions | |
ered | -0.257 |
bled | -0.248 |
hadn | -0.235 |
izes | -0.226 |
eking | -0.226 |
Token | , |
Token ID | 11 |
Feature activation | +1.79e-01 |
Pos logit contributions | |
igsaw | +0.289 |
oak | +0.257 |
newspaper | +0.255 |
magazine | +0.247 |
bull | +0.230 |
Neg logit contributions | |
ered | -0.385 |
bled | -0.376 |
hadn | -0.346 |
icked | -0.333 |
izes | -0.325 |
Token | there |
Token ID | 612 |
Feature activation | +5.36e-02 |
Pos logit contributions | |
igsaw | +0.344 |
oak | +0.301 |
magazine | +0.285 |
newspaper | +0.281 |
pine | +0.272 |
Neg logit contributions | |
ered | -0.334 |
bled | -0.329 |
izes | -0.293 |
vowed | -0.292 |
hadn | -0.285 |
Token | was |
Token ID | 373 |
Feature activation | -1.78e-01 |
Pos logit contributions | |
igsaw | +0.136 |
oak | +0.123 |
magazine | +0.119 |
newspaper | +0.118 |
pine | +0.115 |
Neg logit contributions | |
ered | -0.066 |
bled | -0.065 |
izes | -0.055 |
vowed | -0.053 |
eking | -0.052 |
Token | a |
Token ID | 257 |
Feature activation | +1.32e-01 |
Pos logit contributions | |
bled | +0.261 |
ered | +0.241 |
hes | +0.221 |
icked | +0.205 |
hadn | +0.202 |
Neg logit contributions | |
igsaw | -0.424 |
oak | -0.368 |
newspaper | -0.329 |
dictionary | -0.329 |
magazine | -0.329 |
Token | little |
Token ID | 1310 |
Feature activation | +2.75e-01 |
Pos logit contributions | |
igsaw | +0.269 |
oak | +0.188 |
newspaper | +0.178 |
magazine | +0.178 |
olive | +0.163 |
Neg logit contributions | |
bled | -0.225 |
hadn | -0.209 |
sorry | -0.204 |
also | -0.202 |
forgot | -0.201 |
Token | able |
Token ID | 1498 |
Feature activation | +5.93e-02 |
Pos logit contributions | |
ered | +0.175 |
bled | +0.173 |
izes | +0.158 |
vowed | +0.151 |
hadn | +0.149 |
Neg logit contributions | |
igsaw | -0.174 |
oak | -0.149 |
magazine | -0.147 |
ant | -0.143 |
newspaper | -0.139 |
Token | to |
Token ID | 284 |
Feature activation | +2.10e-01 |
Pos logit contributions | |
igsaw | +0.039 |
oak | +0.030 |
magazine | +0.026 |
newspaper | +0.026 |
ant | +0.025 |
Neg logit contributions | |
ered | -0.182 |
bled | -0.181 |
izes | -0.172 |
eking | -0.164 |
vowed | -0.163 |
Token | recognize |
Token ID | 7564 |
Feature activation | -7.38e-02 |
Pos logit contributions | |
igsaw | +0.386 |
oak | +0.356 |
magazine | +0.319 |
newspaper | +0.316 |
pine | +0.315 |
Neg logit contributions | |
bled | -0.409 |
ered | -0.376 |
izes | -0.359 |
uffed | -0.358 |
itates | -0.355 |
Token | a |
Token ID | 257 |
Feature activation | +1.92e-01 |
Pos logit contributions | |
ered | +0.153 |
bled | +0.149 |
vowed | +0.134 |
izes | +0.133 |
hadn | +0.131 |
Neg logit contributions | |
igsaw | -0.127 |
oak | -0.109 |
magazine | -0.101 |
newspaper | -0.100 |
pine | -0.095 |
Token | doctor |
Token ID | 6253 |
Feature activation | +1.86e-01 |
Pos logit contributions | |
igsaw | +0.294 |
oak | +0.244 |
magazine | +0.225 |
newspaper | +0.220 |
pine | +0.209 |
Neg logit contributions | |
ered | -0.424 |
bled | -0.418 |
vowed | -0.385 |
izes | -0.383 |
hadn | -0.381 |
Token | , |
Token ID | 11 |
Feature activation | +2.23e-01 |
Pos logit contributions | |
igsaw | +0.282 |
oak | +0.238 |
magazine | +0.223 |
newspaper | +0.219 |
pine | +0.206 |
Neg logit contributions | |
ered | -0.422 |
bled | -0.420 |
izes | -0.378 |
vowed | -0.374 |
hadn | -0.369 |
Token | and |
Token ID | 290 |
Feature activation | +1.43e-01 |
Pos logit contributions | |
igsaw | +0.345 |
oak | +0.303 |
magazine | +0.285 |
newspaper | +0.284 |
pine | +0.270 |
Neg logit contributions | |
bled | -0.504 |
ered | -0.504 |
izes | -0.454 |
eking | -0.443 |
icked | -0.428 |
Token | even |
Token ID | 772 |
Feature activation | -7.85e-02 |
Pos logit contributions | |
igsaw | +0.230 |
oak | +0.213 |
magazine | +0.197 |
newspaper | +0.197 |
pine | +0.192 |
Neg logit contributions | |
bled | -0.297 |
ered | -0.288 |
eking | -0.274 |
izes | -0.262 |
itates | -0.262 |
Token | more |
Token ID | 517 |
Feature activation | +1.26e-01 |
Pos logit contributions | |
bled | +0.133 |
ered | +0.129 |
izes | +0.111 |
eking | +0.111 |
uffed | +0.109 |
Neg logit contributions | |
igsaw | -0.161 |
oak | -0.143 |
magazine | -0.141 |
newspaper | -0.139 |
pine | -0.134 |
Token | excited |
Token ID | 6568 |
Feature activation | +4.37e-02 |
Pos logit contributions | |
igsaw | +0.224 |
oak | +0.195 |
magazine | +0.186 |
newspaper | +0.185 |
pine | +0.178 |
Neg logit contributions | |
bled | -0.260 |
ered | -0.257 |
izes | -0.227 |
eking | -0.221 |
vowed | -0.221 |
Token | to |
Token ID | 284 |
Feature activation | +1.87e-01 |
Pos logit contributions | |
igsaw | +0.063 |
oak | +0.052 |
magazine | +0.048 |
newspaper | +0.048 |
ant | +0.045 |
Neg logit contributions | |
bled | -0.103 |
ered | -0.102 |
izes | -0.093 |
vowed | -0.092 |
hadn | -0.091 |
Token | € |
Token ID | 26391 |
Feature activation | +5.01e-02 |
Pos logit contributions | |
igsaw | +0.130 |
oak | +0.113 |
magazine | +0.107 |
newspaper | +0.105 |
pine | +0.101 |
Neg logit contributions | |
bled | -0.117 |
ered | -0.117 |
izes | -0.102 |
vowed | -0.101 |
hadn | -0.100 |
Token | � |
Token ID | 129 |
Feature activation | +3.89e-02 |
Pos logit contributions | |
igsaw | +0.093 |
oak | +0.080 |
magazine | +0.076 |
newspaper | +0.075 |
pine | +0.072 |
Neg logit contributions | |
ered | -0.098 |
bled | -0.097 |
izes | -0.086 |
vowed | -0.085 |
hadn | -0.084 |
Token | � |
Token ID | 241 |
Feature activation | +7.11e-04 |
Pos logit contributions | |
igsaw | +0.083 |
oak | +0.075 |
magazine | +0.070 |
newspaper | +0.069 |
pine | +0.068 |
Neg logit contributions | |
bled | -0.063 |
ered | -0.062 |
hadn | -0.053 |
izes | -0.053 |
vowed | -0.053 |
Token | I |
Token ID | 40 |
Feature activation | -1.67e-01 |
Pos logit contributions | |
igsaw | +0.001 |
oak | +0.001 |
newspaper | +0.001 |
pine | +0.001 |
magazine | +0.001 |
Neg logit contributions | |
ered | -0.001 |
bled | -0.001 |
izes | -0.001 |
eking | -0.001 |
vowed | -0.001 |
Token | want |
Token ID | 765 |
Feature activation | -1.95e-01 |
Pos logit contributions | |
bled | +0.325 |
eking | +0.263 |
ered | +0.259 |
bling | +0.244 |
izes | +0.243 |
Neg logit contributions | |
igsaw | -0.325 |
oak | -0.294 |
pine | -0.289 |
cane | -0.264 |
newspaper | -0.262 |
Token | to |
Token ID | 284 |
Feature activation | +1.05e-01 |
Pos logit contributions | |
bled | +0.528 |
ered | +0.518 |
izes | +0.475 |
eking | +0.474 |
vowed | +0.466 |
Neg logit contributions | |
igsaw | -0.214 |
oak | -0.177 |
newspaper | -0.174 |
magazine | -0.168 |
pine | -0.155 |
Token | soar |
Token ID | 48701 |
Feature activation | -9.03e-02 |
Pos logit contributions | |
igsaw | +0.187 |
oak | +0.173 |
magazine | +0.155 |
newspaper | +0.153 |
pine | +0.153 |
Neg logit contributions | |
bled | -0.211 |
ered | -0.198 |
hadn | -0.180 |
izes | -0.180 |
uffed | -0.179 |
Token | ! |
Token ID | 0 |
Feature activation | +6.91e-02 |
Pos logit contributions | |
bled | +0.190 |
ered | +0.190 |
vowed | +0.169 |
izes | +0.169 |
hadn | +0.167 |
Neg logit contributions | |
igsaw | -0.150 |
oak | -0.130 |
magazine | -0.122 |
newspaper | -0.119 |
pine | -0.116 |
Token | â |
Token ID | 22940 |
Feature activation | -5.00e-02 |
Pos logit contributions | |
igsaw | +0.137 |
oak | +0.120 |
magazine | +0.114 |
newspaper | +0.113 |
pine | +0.107 |
Neg logit contributions | |
ered | -0.123 |
bled | -0.121 |
vowed | -0.108 |
izes | -0.107 |
hadn | -0.106 |
Token | € |
Token ID | 26391 |
Feature activation | +4.44e-02 |
Pos logit contributions | |
ered | +0.091 |
bled | +0.091 |
izes | +0.080 |
vowed | +0.080 |
hadn | +0.078 |
Neg logit contributions | |
igsaw | -0.098 |
oak | -0.085 |
magazine | -0.081 |
newspaper | -0.080 |
pine | -0.077 |
Token | The |
Token ID | 383 |
Feature activation | +8.30e-02 |
Pos logit contributions | |
igsaw | +0.083 |
oak | +0.071 |
magazine | +0.067 |
newspaper | +0.067 |
pine | +0.063 |
Neg logit contributions | |
ered | -0.086 |
bled | -0.084 |
izes | -0.075 |
vowed | -0.075 |
hadn | -0.075 |
Token | to |
Token ID | 284 |
Feature activation | +1.34e-01 |
Pos logit contributions | |
ered | +0.155 |
bled | +0.154 |
izes | +0.139 |
vowed | +0.139 |
hadn | +0.138 |
Neg logit contributions | |
igsaw | -0.091 |
oak | -0.076 |
magazine | -0.070 |
newspaper | -0.068 |
pine | -0.065 |
Token | the |
Token ID | 262 |
Feature activation | +2.20e-01 |
Pos logit contributions | |
igsaw | +0.274 |
magazine | +0.206 |
pillar | +0.206 |
ay | +0.206 |
oak | +0.206 |
Neg logit contributions | |
ered | -0.232 |
bled | -0.202 |
vowed | -0.194 |
not | -0.192 |
missed | -0.186 |
Token | pengu |
Token ID | 38373 |
Feature activation | +1.44e-01 |
Pos logit contributions | |
igsaw | +0.220 |
oak | +0.150 |
magazine | +0.134 |
newspaper | +0.130 |
pine | +0.112 |
Neg logit contributions | |
ered | -0.617 |
bled | -0.613 |
izes | -0.572 |
vowed | -0.567 |
hadn | -0.560 |
Token | in |
Token ID | 259 |
Feature activation | +1.74e-01 |
Pos logit contributions | |
igsaw | +0.263 |
oak | +0.238 |
magazine | +0.227 |
ant | +0.222 |
pine | +0.219 |
Neg logit contributions | |
bled | -0.257 |
ered | -0.255 |
hadn | -0.237 |
vowed | -0.234 |
apologised | -0.227 |
Token | house |
Token ID | 2156 |
Feature activation | +1.14e-01 |
Pos logit contributions | |
igsaw | +0.301 |
oak | +0.220 |
newspaper | +0.212 |
magazine | +0.207 |
Mic | +0.195 |
Neg logit contributions | |
ered | -0.372 |
bled | -0.344 |
vowed | -0.340 |
izes | -0.339 |
wouldn | -0.319 |
Token | . |
Token ID | 13 |
Feature activation | +1.13e-01 |
Pos logit contributions | |
igsaw | +0.119 |
oak | +0.104 |
magazine | +0.095 |
newspaper | +0.092 |
pine | +0.087 |
Neg logit contributions | |
bled | -0.299 |
ered | -0.298 |
izes | -0.272 |
hadn | -0.270 |
vowed | -0.269 |
Token | The |
Token ID | 383 |
Feature activation | +1.05e-01 |
Pos logit contributions | |
igsaw | +0.219 |
oak | +0.191 |
magazine | +0.181 |
newspaper | +0.178 |
pine | +0.169 |
Neg logit contributions | |
ered | -0.208 |
bled | -0.203 |
vowed | -0.181 |
hadn | -0.180 |
izes | -0.178 |
Token | pengu |
Token ID | 38373 |
Feature activation | +1.39e-01 |
Pos logit contributions | |
igsaw | +0.091 |
oak | +0.081 |
pine | +0.068 |
magazine | +0.065 |
newspaper | +0.063 |
Neg logit contributions | |
ered | -0.306 |
bled | -0.303 |
izes | -0.278 |
vowed | -0.276 |
hadn | -0.273 |
Token | ins |
Token ID | 1040 |
Feature activation | -8.86e-02 |
Pos logit contributions | |
igsaw | +0.252 |
oak | +0.225 |
ant | +0.219 |
magazine | +0.216 |
bull | +0.211 |
Neg logit contributions | |
bled | -0.248 |
ered | -0.245 |
hadn | -0.232 |
vowed | -0.232 |
apologised | -0.224 |
Token | are |
Token ID | 389 |
Feature activation | -1.82e-01 |
Pos logit contributions | |
bled | +0.172 |
hadn | +0.152 |
Eventually | +0.151 |
vere | +0.149 |
eking | +0.149 |
Neg logit contributions | |
oak | -0.141 |
pine | -0.132 |
igsaw | -0.132 |
dog | -0.126 |
magazine | -0.126 |
Token | black |
Token ID | 2042 |
Feature activation | +2.17e-01 |
Pos logit contributions | |
ered | +0.394 |
bled | +0.361 |
izes | +0.355 |
vowed | +0.347 |
hadn | +0.335 |
Neg logit contributions | |
pine | -0.242 |
oak | -0.242 |
magazine | -0.231 |
ant | -0.230 |
blue | -0.227 |
Token | work |
Token ID | 670 |
Feature activation | +1.22e-01 |
Pos logit contributions | |
igsaw | +0.239 |
oak | +0.201 |
magazine | +0.187 |
newspaper | +0.184 |
pine | +0.174 |
Neg logit contributions | |
ered | -0.351 |
bled | -0.349 |
izes | -0.315 |
vowed | -0.312 |
hadn | -0.309 |
Token | . |
Token ID | 13 |
Feature activation | +1.81e-01 |
Pos logit contributions | |
igsaw | +0.164 |
oak | +0.129 |
magazine | +0.127 |
newspaper | +0.121 |
pine | +0.107 |
Neg logit contributions | |
bled | -0.299 |
ered | -0.297 |
izes | -0.268 |
eking | -0.265 |
vowed | -0.261 |
Token | |
Token ID | 220 |
Feature activation | -2.30e-02 |
Pos logit contributions | |
igsaw | +0.327 |
oak | +0.282 |
magazine | +0.266 |
newspaper | +0.263 |
pine | +0.251 |
Neg logit contributions | |
ered | -0.361 |
bled | -0.357 |
izes | -0.318 |
vowed | -0.317 |
hadn | -0.312 |
Token |
|
Token ID | 198 |
Feature activation | +9.15e-02 |
Pos logit contributions | |
ered | +0.054 |
bled | +0.053 |
izes | +0.049 |
vowed | +0.049 |
eking | +0.049 |
Neg logit contributions | |
igsaw | -0.033 |
oak | -0.026 |
magazine | -0.025 |
newspaper | -0.024 |
pine | -0.023 |
Token |
|
Token ID | 198 |
Feature activation | +8.72e-02 |
Pos logit contributions | |
igsaw | +0.133 |
oak | +0.111 |
magazine | +0.103 |
newspaper | +0.100 |
pine | +0.099 |
Neg logit contributions | |
ered | -0.215 |
bled | -0.210 |
izes | -0.194 |
eking | -0.192 |
wouldn | -0.189 |
Token | The |
Token ID | 464 |
Feature activation | +1.88e-01 |
Pos logit contributions | |
igsaw | +0.147 |
oak | +0.127 |
magazine | +0.118 |
newspaper | +0.116 |
pine | +0.115 |
Neg logit contributions | |
ered | -0.185 |
bled | -0.180 |
eking | -0.165 |
izes | -0.162 |
wouldn | -0.161 |
Token | bird |
Token ID | 6512 |
Feature activation | +2.55e-01 |
Pos logit contributions | |
igsaw | +0.318 |
oak | +0.288 |
magazine | +0.266 |
newspaper | +0.260 |
pine | +0.255 |
Neg logit contributions | |
ered | -0.398 |
bled | -0.392 |
izes | -0.352 |
vowed | -0.345 |
eking | -0.343 |
Token | loved |
Token ID | 6151 |
Feature activation | -7.67e-02 |
Pos logit contributions | |
oak | +0.503 |
pine | +0.476 |
igsaw | +0.463 |
elephant | +0.440 |
par | +0.436 |
Neg logit contributions | |
bled | -0.446 |
vere | -0.382 |
ered | -0.377 |
itates | -0.358 |
eking | -0.353 |
Token | its |
Token ID | 663 |
Feature activation | +1.16e-01 |
Pos logit contributions | |
ered | +0.158 |
bled | +0.155 |
vowed | +0.142 |
hadn | +0.140 |
izes | +0.137 |
Neg logit contributions | |
igsaw | -0.134 |
oak | -0.110 |
magazine | -0.106 |
newspaper | -0.104 |
pine | -0.095 |
Token | work |
Token ID | 670 |
Feature activation | +9.84e-02 |
Pos logit contributions | |
igsaw | +0.188 |
oak | +0.156 |
magazine | +0.147 |
newspaper | +0.144 |
pine | +0.136 |
Neg logit contributions | |
ered | -0.254 |
bled | -0.250 |
izes | -0.225 |
vowed | -0.225 |
hadn | -0.223 |
Token | and |
Token ID | 290 |
Feature activation | +1.29e-01 |
Pos logit contributions | |
igsaw | +0.125 |
oak | +0.099 |
magazine | +0.093 |
newspaper | +0.090 |
pine | +0.082 |
Neg logit contributions | |
bled | -0.251 |
ered | -0.250 |
izes | -0.228 |
eking | -0.223 |
vowed | -0.223 |
Token | ?" |
Token ID | 1701 |
Feature activation | +7.56e-02 |
Pos logit contributions | |
igsaw | +0.127 |
oak | +0.110 |
magazine | +0.106 |
newspaper | +0.104 |
pine | +0.098 |
Neg logit contributions | |
ered | -0.142 |
bled | -0.140 |
vowed | -0.125 |
izes | -0.124 |
hadn | -0.123 |
Token |
|
Token ID | 198 |
Feature activation | +8.06e-02 |
Pos logit contributions | |
igsaw | +0.119 |
oak | +0.101 |
magazine | +0.094 |
newspaper | +0.093 |
pine | +0.089 |
Neg logit contributions | |
ered | -0.169 |
bled | -0.167 |
izes | -0.151 |
vowed | -0.150 |
hadn | -0.148 |
Token |
|
Token ID | 198 |
Feature activation | +7.84e-02 |
Pos logit contributions | |
igsaw | +0.150 |
oak | +0.128 |
magazine | +0.121 |
newspaper | +0.119 |
pine | +0.112 |
Neg logit contributions | |
ered | -0.157 |
bled | -0.155 |
izes | -0.139 |
vowed | -0.137 |
hadn | -0.135 |
Token | The |
Token ID | 464 |
Feature activation | +1.72e-01 |
Pos logit contributions | |
igsaw | +0.178 |
oak | +0.150 |
magazine | +0.149 |
newspaper | +0.142 |
olive | +0.138 |
Neg logit contributions | |
ered | -0.118 |
bled | -0.113 |
izes | -0.105 |
vowed | -0.100 |
apologised | -0.100 |
Token | cat |
Token ID | 3797 |
Feature activation | +2.39e-01 |
Pos logit contributions | |
igsaw | +0.169 |
oak | +0.098 |
magazine | +0.077 |
newspaper | +0.073 |
pine | +0.064 |
Neg logit contributions | |
bled | -0.481 |
ered | -0.472 |
izes | -0.451 |
hadn | -0.449 |
vowed | -0.446 |
Token | does |
Token ID | 857 |
Feature activation | -1.11e-01 |
Pos logit contributions | |
igsaw | +0.434 |
oak | +0.419 |
magazine | +0.404 |
newspaper | +0.400 |
pine | +0.400 |
Neg logit contributions | |
bled | -0.474 |
ered | -0.416 |
fed | -0.372 |
eking | -0.370 |
uffed | -0.364 |
Token | not |
Token ID | 407 |
Feature activation | -1.40e-01 |
Pos logit contributions | |
bled | +0.111 |
ered | +0.103 |
hadn | +0.089 |
vowed | +0.084 |
uffed | +0.082 |
Neg logit contributions | |
igsaw | -0.299 |
oak | -0.270 |
magazine | -0.268 |
newspaper | -0.264 |
pine | -0.259 |
Token | answer |
Token ID | 3280 |
Feature activation | +4.16e-02 |
Pos logit contributions | |
ered | +0.234 |
bled | +0.225 |
izes | +0.201 |
vowed | +0.196 |
eking | +0.191 |
Neg logit contributions | |
igsaw | -0.303 |
oak | -0.264 |
magazine | -0.253 |
newspaper | -0.251 |
pine | -0.239 |
Token | . |
Token ID | 13 |
Feature activation | +1.56e-01 |
Pos logit contributions | |
igsaw | +0.061 |
oak | +0.051 |
magazine | +0.048 |
newspaper | +0.047 |
pine | +0.044 |
Neg logit contributions | |
ered | -0.096 |
bled | -0.096 |
izes | -0.087 |
vowed | -0.086 |
hadn | -0.085 |
Token | It |
Token ID | 632 |
Feature activation | -8.94e-02 |
Pos logit contributions | |
igsaw | +0.269 |
oak | +0.232 |
magazine | +0.217 |
newspaper | +0.214 |
pine | +0.208 |
Neg logit contributions | |
ered | -0.322 |
bled | -0.321 |
izes | -0.287 |
vowed | -0.284 |
hadn | -0.280 |
Token | l |
Token ID | 300 |
Feature activation | +1.33e-01 |
Pos logit contributions | |
bled | +0.199 |
ered | +0.168 |
bling | +0.155 |
fed | +0.155 |
eking | +0.154 |
Neg logit contributions | |
pine | -0.143 |
oak | -0.138 |
lion | -0.137 |
ant | -0.134 |
igsaw | -0.133 |
Token | my |
Token ID | 616 |
Feature activation | +2.17e-02 |
Pos logit contributions | |
ered | +0.325 |
bled | +0.321 |
vowed | +0.287 |
izes | +0.286 |
hadn | +0.284 |
Neg logit contributions | |
igsaw | -0.255 |
oak | -0.214 |
magazine | -0.201 |
newspaper | -0.200 |
pine | -0.189 |
Token | tank |
Token ID | 6873 |
Feature activation | +9.07e-03 |
Pos logit contributions | |
igsaw | +0.031 |
oak | +0.026 |
newspaper | +0.023 |
magazine | +0.023 |
pine | +0.022 |
Neg logit contributions | |
bled | -0.051 |
ered | -0.050 |
izes | -0.046 |
vowed | -0.045 |
hadn | -0.045 |
Token | !" |
Token ID | 2474 |
Feature activation | -4.40e-02 |
Pos logit contributions | |
igsaw | +0.015 |
oak | +0.013 |
magazine | +0.013 |
newspaper | +0.012 |
pine | +0.012 |
Neg logit contributions | |
ered | -0.019 |
bled | -0.019 |
izes | -0.017 |
vowed | -0.017 |
hadn | -0.016 |
Token | |
Token ID | 220 |
Feature activation | -1.67e-01 |
Pos logit contributions | |
ered | +0.091 |
bled | +0.089 |
izes | +0.081 |
eking | +0.080 |
vowed | +0.078 |
Neg logit contributions | |
igsaw | -0.076 |
oak | -0.066 |
magazine | -0.064 |
newspaper | -0.062 |
pine | -0.058 |
Token |
|
Token ID | 198 |
Feature activation | -5.43e-02 |
Pos logit contributions | |
ered | +0.394 |
bled | +0.390 |
vowed | +0.359 |
eking | +0.356 |
izes | +0.355 |
Neg logit contributions | |
igsaw | -0.232 |
oak | -0.190 |
magazine | -0.175 |
newspaper | -0.168 |
pine | -0.163 |
Token |
|
Token ID | 198 |
Feature activation | -5.48e-02 |
Pos logit contributions | |
ered | +0.118 |
bled | +0.117 |
izes | +0.105 |
vowed | +0.105 |
hadn | +0.103 |
Neg logit contributions | |
igsaw | -0.088 |
oak | -0.075 |
magazine | -0.070 |
newspaper | -0.069 |
pine | -0.066 |
Token | From |
Token ID | 4863 |
Feature activation | -1.73e-01 |
Pos logit contributions | |
ered | +0.107 |
bled | +0.103 |
eking | +0.093 |
vowed | +0.092 |
izes | +0.091 |
Neg logit contributions | |
igsaw | -0.102 |
oak | -0.086 |
magazine | -0.085 |
newspaper | -0.083 |
olive | -0.077 |
Token | that |
Token ID | 326 |
Feature activation | +2.09e-02 |
Pos logit contributions | |
bled | +0.310 |
ered | +0.307 |
izes | +0.268 |
eking | +0.255 |
vowed | +0.254 |
Neg logit contributions | |
igsaw | -0.377 |
magazine | -0.297 |
oak | -0.294 |
newspaper | -0.286 |
dictionary | -0.270 |
Token | day |
Token ID | 1110 |
Feature activation | +4.84e-02 |
Pos logit contributions | |
igsaw | +0.032 |
oak | +0.027 |
newspaper | +0.024 |
pine | +0.023 |
magazine | +0.023 |
Neg logit contributions | |
ered | -0.044 |
bled | -0.043 |
hadn | -0.042 |
vowed | -0.042 |
wouldn | -0.040 |
Token | on |
Token ID | 319 |
Feature activation | -3.48e-02 |
Pos logit contributions | |
igsaw | +0.096 |
oak | +0.084 |
magazine | +0.080 |
newspaper | +0.079 |
pine | +0.076 |
Neg logit contributions | |
ered | -0.088 |
bled | -0.087 |
izes | -0.077 |
vowed | -0.076 |
hadn | -0.075 |
Token | , |
Token ID | 11 |
Feature activation | +7.38e-02 |
Pos logit contributions | |
ered | +0.080 |
bled | +0.078 |
hadn | +0.071 |
vowed | +0.070 |
wouldn | +0.069 |
Neg logit contributions | |
igsaw | -0.052 |
oak | -0.045 |
newspaper | -0.043 |
bull | -0.041 |
magazine | -0.040 |
Token | the |
Token ID | 262 |
Feature activation | +1.86e-01 |
Pos logit contributions | |
ered | +0.081 |
bled | +0.078 |
hes | +0.072 |
ps | +0.067 |
vowed | +0.067 |
Neg logit contributions | |
igsaw | -0.111 |
Mic | -0.086 |
oak | -0.084 |
cone | -0.081 |
magazine | -0.080 |
Token | match |
Token ID | 2872 |
Feature activation | -2.89e-02 |
Pos logit contributions | |
igsaw | +0.288 |
oak | +0.242 |
newspaper | +0.200 |
pine | +0.200 |
magazine | +0.193 |
Neg logit contributions | |
ered | -0.416 |
bled | -0.415 |
izes | -0.381 |
eking | -0.361 |
vowed | -0.358 |
Token | and |
Token ID | 290 |
Feature activation | +5.95e-02 |
Pos logit contributions | |
ered | +0.059 |
bled | +0.057 |
vowed | +0.051 |
izes | +0.051 |
wouldn | +0.051 |
Neg logit contributions | |
igsaw | -0.052 |
oak | -0.043 |
Mic | -0.039 |
magazine | -0.039 |
newspaper | -0.038 |
Token | run |
Token ID | 1057 |
Feature activation | -1.08e-01 |
Pos logit contributions | |
igsaw | +0.107 |
oak | +0.098 |
newspaper | +0.092 |
magazine | +0.092 |
pine | +0.089 |
Neg logit contributions | |
bled | -0.116 |
ered | -0.113 |
izes | -0.100 |
vowed | -0.099 |
hadn | -0.098 |
Token | away |
Token ID | 1497 |
Feature activation | +6.36e-02 |
Pos logit contributions | |
ered | +0.200 |
bled | +0.192 |
izes | +0.175 |
vowed | +0.175 |
wouldn | +0.168 |
Neg logit contributions | |
igsaw | -0.221 |
oak | -0.183 |
magazine | -0.171 |
newspaper | -0.171 |
bull | -0.163 |
Token | . |
Token ID | 13 |
Feature activation | +7.23e-02 |
Pos logit contributions | |
igsaw | +0.116 |
oak | +0.098 |
magazine | +0.094 |
newspaper | +0.093 |
bull | +0.087 |
Neg logit contributions | |
ered | -0.130 |
bled | -0.127 |
izes | -0.115 |
vowed | -0.113 |
hadn | -0.109 |
Token | They |
Token ID | 1119 |
Feature activation | -2.05e-01 |
Pos logit contributions | |
igsaw | +0.125 |
oak | +0.107 |
magazine | +0.101 |
newspaper | +0.100 |
pine | +0.095 |
Neg logit contributions | |
ered | -0.151 |
bled | -0.150 |
izes | -0.135 |
vowed | -0.133 |
eking | -0.132 |
Token | hide |
Token ID | 7808 |
Feature activation | -1.21e-01 |
Pos logit contributions | |
bled | +0.470 |
ered | +0.404 |
izes | +0.376 |
resolved | +0.372 |
eking | +0.365 |
Neg logit contributions | |
oak | -0.321 |
igsaw | -0.295 |
pine | -0.294 |
newspaper | -0.280 |
cane | -0.280 |
Token | under |
Token ID | 739 |
Feature activation | -2.68e-02 |
Pos logit contributions | |
ered | +0.244 |
bled | +0.242 |
vowed | +0.212 |
izes | +0.211 |
hadn | +0.205 |
Neg logit contributions | |
igsaw | -0.232 |
oak | -0.191 |
magazine | -0.184 |
newspaper | -0.179 |
pine | -0.167 |
Token | the |
Token ID | 262 |
Feature activation | +1.79e-01 |
Pos logit contributions | |
bled | +0.050 |
ered | +0.050 |
vowed | +0.043 |
hadn | +0.043 |
ps | +0.042 |
Neg logit contributions | |
igsaw | -0.052 |
oak | -0.042 |
magazine | -0.039 |
gol | -0.038 |
newspaper | -0.038 |
Token | sofa |
Token ID | 34902 |
Feature activation | +1.14e-01 |
Pos logit contributions | |
igsaw | +0.301 |
gol | +0.241 |
oak | +0.229 |
Mic | +0.207 |
working | +0.202 |
Neg logit contributions | |
bled | -0.335 |
ered | -0.331 |
izes | -0.302 |
ps | -0.299 |
hadn | -0.294 |
Token | . |
Token ID | 13 |
Feature activation | +4.63e-02 |
Pos logit contributions | |
igsaw | +0.084 |
oak | +0.077 |
magazine | +0.067 |
pine | +0.066 |
newspaper | +0.066 |
Neg logit contributions | |
ered | -0.151 |
bled | -0.150 |
vowed | -0.137 |
izes | -0.137 |
hadn | -0.136 |
Token | Every |
Token ID | 3887 |
Feature activation | -5.44e-02 |
Pos logit contributions | |
igsaw | +0.113 |
oak | +0.099 |
insect | +0.091 |
magazine | +0.090 |
olive | +0.090 |
Neg logit contributions | |
ered | -0.054 |
hadn | -0.049 |
bled | -0.048 |
wouldn | -0.047 |
vowed | -0.046 |
Token | day |
Token ID | 1110 |
Feature activation | +6.68e-02 |
Pos logit contributions | |
ered | +0.144 |
bled | +0.142 |
vowed | +0.131 |
izes | +0.131 |
hadn | +0.130 |
Neg logit contributions | |
igsaw | -0.063 |
oak | -0.049 |
magazine | -0.044 |
newspaper | -0.042 |
pine | -0.039 |
Token | they |
Token ID | 484 |
Feature activation | -1.83e-01 |
Pos logit contributions | |
igsaw | +0.129 |
oak | +0.098 |
magazine | +0.095 |
newspaper | +0.091 |
bull | +0.085 |
Neg logit contributions | |
ered | -0.135 |
bled | -0.129 |
wouldn | -0.124 |
hadn | -0.120 |
vowed | -0.119 |
Token | asked |
Token ID | 1965 |
Feature activation | -1.91e-01 |
Pos logit contributions | |
bled | +0.297 |
ered | +0.288 |
izes | +0.258 |
eking | +0.250 |
vowed | +0.239 |
Neg logit contributions | |
igsaw | -0.384 |
oak | -0.371 |
magazine | -0.347 |
newspaper | -0.344 |
pine | -0.340 |
Token | their |
Token ID | 511 |
Feature activation | +1.50e-02 |
Pos logit contributions | |
ered | +0.427 |
bled | +0.425 |
izes | +0.384 |
vowed | +0.380 |
hadn | +0.377 |
Neg logit contributions | |
igsaw | -0.298 |
oak | -0.253 |
magazine | -0.238 |
newspaper | -0.235 |
pine | -0.221 |
Token | parents |
Token ID | 3397 |
Feature activation | +1.03e-02 |
Pos logit contributions | |
igsaw | +0.030 |
oak | +0.024 |
magazine | +0.023 |
newspaper | +0.022 |
pine | +0.022 |
Neg logit contributions | |
ered | -0.026 |
bled | -0.026 |
vowed | -0.023 |
hadn | -0.023 |
izes | -0.023 |
Token | , |
Token ID | 11 |
Feature activation | +9.34e-02 |
Pos logit contributions | |
igsaw | +0.018 |
oak | +0.016 |
magazine | +0.015 |
newspaper | +0.015 |
pine | +0.014 |
Neg logit contributions | |
ered | -0.021 |
bled | -0.021 |
izes | -0.019 |
vowed | -0.018 |
hadn | -0.018 |
Token | but |
Token ID | 475 |
Feature activation | -1.42e-01 |
Pos logit contributions | |
igsaw | +0.159 |
oak | +0.138 |
magazine | +0.129 |
newspaper | +0.128 |
pine | +0.122 |
Neg logit contributions | |
ered | -0.196 |
bled | -0.194 |
izes | -0.174 |
vowed | -0.172 |
hadn | -0.169 |
Token | they |
Token ID | 484 |
Feature activation | -1.86e-01 |
Pos logit contributions | |
ered | +0.261 |
bled | +0.260 |
eking | +0.232 |
izes | +0.231 |
vowed | +0.224 |
Neg logit contributions | |
igsaw | -0.259 |
oak | -0.239 |
magazine | -0.229 |
newspaper | -0.226 |
ant | -0.213 |
Token | said |
Token ID | 531 |
Feature activation | -2.30e-01 |
Pos logit contributions | |
bled | +0.337 |
eking | +0.292 |
ered | +0.290 |
bling | +0.275 |
ps | +0.274 |
Neg logit contributions | |
oak | -0.354 |
dog | -0.319 |
cat | -0.314 |
goat | -0.313 |
newspaper | -0.311 |
Token | It |
Token ID | 632 |
Feature activation | -1.38e-01 |
Pos logit contributions | |
igsaw | +0.137 |
magazine | +0.100 |
newspaper | +0.100 |
oak | +0.099 |
ant | +0.091 |
Neg logit contributions | |
bled | -0.220 |
ered | -0.212 |
izes | -0.205 |
eking | -0.200 |
vowed | -0.192 |
Token | kept |
Token ID | 4030 |
Feature activation | -1.89e-01 |
Pos logit contributions | |
bled | +0.268 |
eking | +0.235 |
izes | +0.230 |
vere | +0.220 |
bling | +0.217 |
Neg logit contributions | |
oak | -0.218 |
k | -0.212 |
cat | -0.211 |
pine | -0.210 |
cater | -0.207 |
Token | falling |
Token ID | 7463 |
Feature activation | -1.27e-02 |
Pos logit contributions | |
ered | +0.381 |
bled | +0.373 |
izes | +0.356 |
vowed | +0.335 |
apologised | +0.329 |
Neg logit contributions | |
igsaw | -0.323 |
oak | -0.289 |
magazine | -0.268 |
newspaper | -0.263 |
pine | -0.262 |
Token | , |
Token ID | 11 |
Feature activation | +1.36e-01 |
Pos logit contributions | |
ered | +0.027 |
bled | +0.026 |
izes | +0.024 |
vowed | +0.023 |
hadn | +0.023 |
Neg logit contributions | |
igsaw | -0.022 |
oak | -0.018 |
magazine | -0.017 |
newspaper | -0.017 |
pine | -0.016 |
Token | until |
Token ID | 1566 |
Feature activation | -7.23e-02 |
Pos logit contributions | |
igsaw | +0.131 |
oak | +0.126 |
pine | +0.110 |
newspaper | +0.106 |
ant | +0.104 |
Neg logit contributions | |
bled | -0.358 |
ered | -0.357 |
izes | -0.333 |
icked | -0.332 |
itates | -0.321 |
Token | it |
Token ID | 340 |
Feature activation | -3.50e-02 |
Pos logit contributions | |
bled | +0.181 |
izes | +0.177 |
eking | +0.174 |
ered | +0.171 |
olded | +0.168 |
Neg logit contributions | |
igsaw | -0.088 |
oak | -0.062 |
pine | -0.061 |
magazine | -0.056 |
newspaper | -0.055 |
Token | was |
Token ID | 373 |
Feature activation | -2.13e-01 |
Pos logit contributions | |
bled | +0.057 |
izes | +0.055 |
ered | +0.054 |
eking | +0.053 |
vere | +0.052 |
Neg logit contributions | |
igsaw | -0.058 |
oak | -0.057 |
cat | -0.056 |
pine | -0.056 |
cater | -0.055 |
Token | gone |
Token ID | 3750 |
Feature activation | -1.23e-01 |
Pos logit contributions | |
izes | +0.392 |
bled | +0.386 |
ered | +0.386 |
vowed | +0.356 |
itates | +0.351 |
Neg logit contributions | |
igsaw | -0.366 |
oak | -0.334 |
pine | -0.313 |
magazine | -0.305 |
ant | -0.297 |
Token | . |
Token ID | 13 |
Feature activation | +8.35e-02 |
Pos logit contributions | |
bled | +0.314 |
ered | +0.312 |
izes | +0.290 |
vowed | +0.283 |
apologised | +0.280 |
Neg logit contributions | |
igsaw | -0.150 |
oak | -0.128 |
magazine | -0.113 |
newspaper | -0.109 |
pine | -0.106 |
Token |
|
Token ID | 198 |
Feature activation | +6.44e-03 |
Pos logit contributions | |
igsaw | +0.116 |
magazine | +0.079 |
newspaper | +0.077 |
oak | +0.074 |
ant | +0.069 |
Neg logit contributions | |
bled | -0.215 |
ered | -0.207 |
izes | -0.201 |
eking | -0.200 |
vowed | -0.185 |
Token |
|
Token ID | 198 |
Feature activation | +5.24e-03 |
Pos logit contributions | |
igsaw | +0.006 |
newspaper | +0.004 |
pine | +0.004 |
oak | +0.003 |
magazine | +0.003 |
Neg logit contributions | |
bled | -0.018 |
ered | -0.017 |
eking | -0.017 |
wouldn | -0.017 |
izes | -0.017 |
Token | things |
Token ID | 1243 |
Feature activation | +8.37e-02 |
Pos logit contributions | |
igsaw | +0.052 |
oak | +0.043 |
magazine | +0.041 |
newspaper | +0.040 |
pine | +0.038 |
Neg logit contributions | |
ered | -0.072 |
bled | -0.070 |
izes | -0.064 |
vowed | -0.064 |
hadn | -0.063 |
Token | in |
Token ID | 287 |
Feature activation | +5.96e-02 |
Pos logit contributions | |
igsaw | +0.098 |
oak | +0.082 |
magazine | +0.075 |
newspaper | +0.075 |
pine | +0.068 |
Neg logit contributions | |
bled | -0.216 |
ered | -0.210 |
izes | -0.198 |
vowed | -0.190 |
eking | -0.189 |
Token | the |
Token ID | 262 |
Feature activation | +1.97e-01 |
Pos logit contributions | |
igsaw | +0.091 |
oak | +0.076 |
magazine | +0.073 |
newspaper | +0.073 |
bull | +0.069 |
Neg logit contributions | |
ered | -0.133 |
bled | -0.122 |
vowed | -0.117 |
hadn | -0.115 |
apologised | -0.110 |
Token | tunnel |
Token ID | 13275 |
Feature activation | +8.55e-02 |
Pos logit contributions | |
igsaw | +0.212 |
oak | +0.172 |
magazine | +0.148 |
newspaper | +0.145 |
pine | +0.138 |
Neg logit contributions | |
bled | -0.548 |
ered | -0.542 |
izes | -0.495 |
vowed | -0.489 |
icked | -0.483 |
Token | . |
Token ID | 13 |
Feature activation | +9.35e-02 |
Pos logit contributions | |
igsaw | +0.094 |
oak | +0.082 |
magazine | +0.070 |
newspaper | +0.068 |
pine | +0.063 |
Neg logit contributions | |
bled | -0.222 |
ered | -0.217 |
izes | -0.202 |
itates | -0.201 |
vowed | -0.197 |
Token | They |
Token ID | 1119 |
Feature activation | -1.88e-01 |
Pos logit contributions | |
igsaw | +0.142 |
oak | +0.118 |
magazine | +0.109 |
newspaper | +0.107 |
pine | +0.101 |
Neg logit contributions | |
ered | -0.216 |
bled | -0.213 |
izes | -0.195 |
vowed | -0.192 |
eking | -0.190 |
Token | both |
Token ID | 1111 |
Feature activation | -8.43e-02 |
Pos logit contributions | |
bled | +0.311 |
izes | +0.292 |
itates | +0.280 |
ered | +0.279 |
eking | +0.276 |
Neg logit contributions | |
oak | -0.383 |
pine | -0.350 |
igsaw | -0.342 |
elephant | -0.324 |
olive | -0.320 |
Token | had |
Token ID | 550 |
Feature activation | -2.27e-01 |
Pos logit contributions | |
bled | +0.142 |
izes | +0.123 |
vere | +0.122 |
eking | +0.121 |
ered | +0.119 |
Neg logit contributions | |
oak | -0.169 |
igsaw | -0.166 |
pine | -0.160 |
ant | -0.149 |
elephant | -0.148 |
Token | so |
Token ID | 523 |
Feature activation | -3.85e-02 |
Pos logit contributions | |
bled | +0.500 |
izes | +0.476 |
ered | +0.460 |
itates | +0.443 |
uffed | +0.434 |
Neg logit contributions | |
igsaw | -0.345 |
oak | -0.331 |
newspaper | -0.310 |
pine | -0.307 |
magazine | -0.305 |
Token | much |
Token ID | 881 |
Feature activation | -9.46e-03 |
Pos logit contributions | |
bled | +0.080 |
ered | +0.073 |
izes | +0.071 |
wouldn | +0.069 |
eking | +0.068 |
Neg logit contributions | |
igsaw | -0.072 |
oak | -0.063 |
pine | -0.059 |
newspaper | -0.056 |
magazine | -0.055 |
Token | fun |
Token ID | 1257 |
Feature activation | -9.13e-03 |
Pos logit contributions | |
bled | +0.022 |
ered | +0.021 |
izes | +0.020 |
vowed | +0.019 |
eking | +0.019 |
Neg logit contributions | |
igsaw | -0.015 |
oak | -0.012 |
newspaper | -0.012 |
pine | -0.012 |
magazine | -0.011 |
Token | football |
Token ID | 4346 |
Feature activation | +6.56e-02 |
Pos logit contributions | |
igsaw | +0.147 |
oak | +0.120 |
pine | +0.098 |
magazine | +0.093 |
newspaper | +0.093 |
Neg logit contributions | |
bled | -0.337 |
ered | -0.327 |
izes | -0.305 |
hadn | -0.303 |
vowed | -0.298 |
Token | !" |
Token ID | 2474 |
Feature activation | -4.52e-02 |
Pos logit contributions | |
igsaw | +0.105 |
oak | +0.090 |
magazine | +0.085 |
newspaper | +0.083 |
pine | +0.079 |
Neg logit contributions | |
ered | -0.143 |
bled | -0.143 |
vowed | -0.128 |
izes | -0.127 |
hadn | -0.126 |
Token | shouted |
Token ID | 17293 |
Feature activation | -2.64e-01 |
Pos logit contributions | |
bled | +0.079 |
ered | +0.077 |
izes | +0.069 |
vowed | +0.068 |
eking | +0.067 |
Neg logit contributions | |
igsaw | -0.091 |
oak | -0.080 |
newspaper | -0.077 |
magazine | -0.077 |
pine | -0.073 |
Token | John |
Token ID | 1757 |
Feature activation | -7.97e-02 |
Pos logit contributions | |
ered | +0.526 |
bled | +0.468 |
apologised | +0.430 |
vowed | +0.418 |
olded | +0.411 |
Neg logit contributions | |
igsaw | -0.468 |
oak | -0.436 |
pine | -0.420 |
newspaper | -0.395 |
cone | -0.389 |
Token | . |
Token ID | 13 |
Feature activation | +2.43e-02 |
Pos logit contributions | |
ered | +0.176 |
bled | +0.169 |
vowed | +0.153 |
eking | +0.152 |
apologised | +0.152 |
Neg logit contributions | |
igsaw | -0.137 |
oak | -0.111 |
newspaper | -0.106 |
magazine | -0.106 |
pine | -0.097 |
Token | He |
Token ID | 679 |
Feature activation | -1.98e-01 |
Pos logit contributions | |
igsaw | +0.044 |
oak | +0.038 |
magazine | +0.035 |
newspaper | +0.035 |
pine | +0.033 |
Neg logit contributions | |
ered | -0.049 |
bled | -0.048 |
izes | -0.043 |
vowed | -0.043 |
hadn | -0.042 |
Token | threw |
Token ID | 9617 |
Feature activation | -2.03e-01 |
Pos logit contributions | |
bled | +0.344 |
ered | +0.323 |
eking | +0.322 |
izes | +0.311 |
vere | +0.300 |
Neg logit contributions | |
dog | -0.327 |
ant | -0.326 |
cat | -0.324 |
igsaw | -0.323 |
bull | -0.319 |
Token | the |
Token ID | 262 |
Feature activation | +1.28e-01 |
Pos logit contributions | |
ered | +0.399 |
bled | +0.363 |
apologised | +0.329 |
vowed | +0.327 |
izes | +0.314 |
Neg logit contributions | |
igsaw | -0.386 |
oak | -0.326 |
magazine | -0.286 |
newspaper | -0.282 |
farm | -0.274 |
Token | ball |
Token ID | 2613 |
Feature activation | +8.61e-02 |
Pos logit contributions | |
igsaw | +0.129 |
oak | +0.102 |
pine | +0.081 |
magazine | +0.079 |
newspaper | +0.079 |
Neg logit contributions | |
bled | -0.351 |
ered | -0.345 |
izes | -0.317 |
hadn | -0.314 |
vowed | -0.312 |
Token | towards |
Token ID | 3371 |
Feature activation | -3.06e-02 |
Pos logit contributions | |
igsaw | +0.161 |
oak | +0.137 |
magazine | +0.129 |
newspaper | +0.129 |
pine | +0.122 |
Neg logit contributions | |
ered | -0.169 |
bled | -0.167 |
izes | -0.148 |
vowed | -0.147 |
hadn | -0.143 |
Token | Sarah |
Token ID | 10490 |
Feature activation | -8.26e-02 |
Pos logit contributions | |
ered | +0.067 |
bled | +0.065 |
vowed | +0.059 |
izes | +0.058 |
hadn | +0.058 |
Neg logit contributions | |
igsaw | -0.051 |
oak | -0.043 |
magazine | -0.040 |
newspaper | -0.039 |
pine | -0.038 |
Token | so |
Token ID | 523 |
Feature activation | -5.32e-02 |
Pos logit contributions | |
ered | +0.470 |
bled | +0.454 |
izes | +0.453 |
wouldn | +0.437 |
vowed | +0.424 |
Neg logit contributions | |
igsaw | -0.298 |
magazine | -0.277 |
oak | -0.249 |
ant | -0.245 |
newspaper | -0.240 |
Token | pleased |
Token ID | 10607 |
Feature activation | -6.71e-02 |
Pos logit contributions | |
ered | +0.106 |
bled | +0.101 |
wouldn | +0.101 |
vowed | +0.097 |
izes | +0.096 |
Neg logit contributions | |
igsaw | -0.077 |
blue | -0.062 |
mel | -0.061 |
red | -0.061 |
ant | -0.060 |
Token | with |
Token ID | 351 |
Feature activation | -4.83e-02 |
Pos logit contributions | |
ered | +0.162 |
bled | +0.161 |
izes | +0.147 |
vowed | +0.145 |
hadn | +0.143 |
Neg logit contributions | |
igsaw | -0.093 |
oak | -0.076 |
magazine | -0.071 |
newspaper | -0.070 |
pine | -0.066 |
Token | it |
Token ID | 340 |
Feature activation | -4.96e-02 |
Pos logit contributions | |
bled | +0.116 |
ered | +0.115 |
izes | +0.109 |
eking | +0.104 |
hadn | +0.104 |
Neg logit contributions | |
igsaw | -0.068 |
oak | -0.052 |
magazine | -0.050 |
newspaper | -0.048 |
ant | -0.046 |
Token | . |
Token ID | 13 |
Feature activation | +7.10e-02 |
Pos logit contributions | |
bled | +0.142 |
ered | +0.135 |
eking | +0.129 |
izes | +0.129 |
uffed | +0.128 |
Neg logit contributions | |
igsaw | -0.047 |
magazine | -0.037 |
newspaper | -0.032 |
oak | -0.030 |
olive | -0.027 |
Token | She |
Token ID | 1375 |
Feature activation | -1.48e-01 |
Pos logit contributions | |
igsaw | +0.103 |
oak | +0.077 |
magazine | +0.073 |
newspaper | +0.071 |
ant | +0.070 |
Neg logit contributions | |
ered | -0.176 |
bled | -0.170 |
eking | -0.161 |
izes | -0.159 |
vowed | -0.152 |
Token | was |
Token ID | 373 |
Feature activation | -2.33e-01 |
Pos logit contributions | |
bled | +0.289 |
eking | +0.281 |
ered | +0.269 |
vere | +0.263 |
itates | +0.257 |
Neg logit contributions | |
oak | -0.230 |
pine | -0.230 |
par | -0.219 |
mel | -0.216 |
cater | -0.216 |
Token | proud |
Token ID | 6613 |
Feature activation | -7.49e-02 |
Pos logit contributions | |
izes | +0.471 |
ered | +0.467 |
bled | +0.444 |
its | +0.440 |
hew | +0.431 |
Neg logit contributions | |
igsaw | -0.299 |
ant | -0.266 |
oak | -0.257 |
magazine | -0.255 |
par | -0.249 |
Token | of |
Token ID | 286 |
Feature activation | +7.84e-02 |
Pos logit contributions | |
ered | +0.178 |
bled | +0.177 |
izes | +0.161 |
vowed | +0.157 |
eking | +0.156 |
Neg logit contributions | |
igsaw | -0.109 |
oak | -0.090 |
magazine | -0.084 |
newspaper | -0.083 |
ant | -0.079 |
Token | Molly |
Token ID | 30236 |
Feature activation | -2.73e-02 |
Pos logit contributions | |
igsaw | +0.108 |
oak | +0.078 |
ant | +0.075 |
magazine | +0.074 |
newspaper | +0.070 |
Neg logit contributions | |
ered | -0.187 |
bled | -0.186 |
izes | -0.178 |
uffed | -0.171 |
eking | -0.170 |
Token | for |
Token ID | 329 |
Feature activation | -9.49e-02 |
Pos logit contributions | |
bled | +0.064 |
ered | +0.064 |
izes | +0.058 |
eking | +0.057 |
vowed | +0.056 |
Neg logit contributions | |
igsaw | -0.040 |
oak | -0.032 |
magazine | -0.032 |
newspaper | -0.031 |
ant | -0.029 |
Token | cover |
Token ID | 3002 |
Feature activation | -6.56e-03 |
Pos logit contributions | |
igsaw | +0.133 |
oak | +0.124 |
pine | +0.114 |
newspaper | +0.114 |
magazine | +0.113 |
Neg logit contributions | |
bled | -0.129 |
ered | -0.120 |
izes | -0.110 |
uffed | -0.107 |
hadn | -0.106 |
Token | her |
Token ID | 607 |
Feature activation | +3.97e-02 |
Pos logit contributions | |
ered | +0.013 |
bled | +0.013 |
izes | +0.011 |
eking | +0.011 |
vowed | +0.011 |
Neg logit contributions | |
igsaw | -0.012 |
oak | -0.010 |
magazine | -0.010 |
newspaper | -0.009 |
bull | -0.009 |
Token | neck |
Token ID | 7393 |
Feature activation | +4.60e-02 |
Pos logit contributions | |
igsaw | +0.072 |
oak | +0.054 |
magazine | +0.050 |
bull | +0.049 |
newspaper | +0.049 |
Neg logit contributions | |
ered | -0.083 |
bled | -0.081 |
eking | -0.076 |
izes | -0.075 |
vowed | -0.074 |
Token | when |
Token ID | 618 |
Feature activation | +2.25e-03 |
Pos logit contributions | |
igsaw | +0.080 |
oak | +0.067 |
magazine | +0.063 |
newspaper | +0.062 |
bull | +0.059 |
Neg logit contributions | |
ered | -0.097 |
bled | -0.095 |
izes | -0.085 |
vowed | -0.084 |
eking | -0.083 |
Token | she |
Token ID | 673 |
Feature activation | -1.38e-01 |
Pos logit contributions | |
igsaw | +0.003 |
oak | +0.003 |
pine | +0.003 |
newspaper | +0.003 |
magazine | +0.003 |
Neg logit contributions | |
bled | -0.005 |
ered | -0.005 |
izes | -0.005 |
apologised | -0.005 |
vowed | -0.005 |
Token | played |
Token ID | 2826 |
Feature activation | -1.57e-01 |
Pos logit contributions | |
bled | +0.230 |
ered | +0.211 |
izes | +0.196 |
vere | +0.183 |
eking | +0.176 |
Neg logit contributions | |
oak | -0.282 |
igsaw | -0.276 |
pine | -0.269 |
ant | -0.255 |
newspaper | -0.254 |
Token | outside |
Token ID | 2354 |
Feature activation | +5.65e-02 |
Pos logit contributions | |
ered | +0.337 |
bled | +0.335 |
vowed | +0.304 |
izes | +0.302 |
hadn | +0.300 |
Neg logit contributions | |
igsaw | -0.256 |
oak | -0.222 |
magazine | -0.211 |
newspaper | -0.204 |
pine | -0.196 |
Token | in |
Token ID | 287 |
Feature activation | +4.73e-03 |
Pos logit contributions | |
igsaw | +0.096 |
oak | +0.080 |
magazine | +0.075 |
newspaper | +0.075 |
pine | +0.071 |
Neg logit contributions | |
ered | -0.119 |
bled | -0.116 |
vowed | -0.105 |
izes | -0.105 |
hadn | -0.104 |
Token | the |
Token ID | 262 |
Feature activation | +1.49e-01 |
Pos logit contributions | |
igsaw | +0.008 |
oak | +0.006 |
magazine | +0.006 |
newspaper | +0.006 |
bull | +0.006 |
Neg logit contributions | |
ered | -0.011 |
bled | -0.010 |
vowed | -0.009 |
hadn | -0.009 |
eking | -0.009 |
Token | sunshine |
Token ID | 34488 |
Feature activation | +3.39e-02 |
Pos logit contributions | |
igsaw | +0.242 |
magazine | +0.171 |
oak | +0.169 |
gol | +0.163 |
newspaper | +0.162 |
Neg logit contributions | |
ered | -0.314 |
bled | -0.299 |
vowed | -0.286 |
izes | -0.284 |
hadn | -0.283 |
Token | . |
Token ID | 13 |
Feature activation | +3.47e-02 |
Pos logit contributions | |
igsaw | +0.070 |
oak | +0.056 |
newspaper | +0.054 |
bull | +0.053 |
magazine | +0.052 |
Neg logit contributions | |
ered | -0.058 |
bled | -0.056 |
wouldn | -0.052 |
vowed | -0.052 |
izes | -0.051 |
Token | all |
Token ID | 477 |
Feature activation | +2.23e-02 |
Pos logit contributions | |
igsaw | +0.240 |
oak | +0.206 |
newspaper | +0.195 |
magazine | +0.194 |
olive | +0.179 |
Neg logit contributions | |
bled | -0.340 |
ered | -0.323 |
eking | -0.304 |
izes | -0.303 |
vowed | -0.293 |
Token | about |
Token ID | 546 |
Feature activation | -3.48e-02 |
Pos logit contributions | |
igsaw | +0.047 |
oak | +0.041 |
magazine | +0.039 |
newspaper | +0.038 |
pine | +0.037 |
Neg logit contributions | |
ered | -0.039 |
bled | -0.037 |
izes | -0.033 |
vowed | -0.033 |
hadn | -0.032 |
Token | the |
Token ID | 262 |
Feature activation | +2.60e-01 |
Pos logit contributions | |
ered | +0.074 |
bled | +0.072 |
vowed | +0.066 |
hadn | +0.065 |
izes | +0.065 |
Neg logit contributions | |
igsaw | -0.059 |
oak | -0.048 |
magazine | -0.045 |
newspaper | -0.044 |
pine | -0.042 |
Token | places |
Token ID | 4113 |
Feature activation | +1.21e-01 |
Pos logit contributions | |
igsaw | +0.370 |
oak | +0.307 |
magazine | +0.285 |
newspaper | +0.281 |
pine | +0.263 |
Neg logit contributions | |
ered | -0.620 |
bled | -0.615 |
izes | -0.559 |
vowed | -0.556 |
hadn | -0.548 |
Token | they |
Token ID | 484 |
Feature activation | -8.16e-02 |
Pos logit contributions | |
igsaw | +0.162 |
oak | +0.149 |
magazine | +0.141 |
newspaper | +0.141 |
pine | +0.124 |
Neg logit contributions | |
bled | -0.285 |
ered | -0.275 |
izes | -0.253 |
eking | -0.247 |
itates | -0.246 |
Token | were |
Token ID | 547 |
Feature activation | -1.81e-01 |
Pos logit contributions | |
bled | +0.167 |
ered | +0.145 |
uffed | +0.139 |
aned | +0.134 |
izes | +0.132 |
Neg logit contributions | |
oak | -0.154 |
igsaw | -0.136 |
magazine | -0.136 |
newspaper | -0.135 |
elephant | -0.129 |
Token | visiting |
Token ID | 10013 |
Feature activation | -4.06e-02 |
Pos logit contributions | |
bled | +0.351 |
ered | +0.336 |
uffed | +0.302 |
apologised | +0.301 |
izes | +0.299 |
Neg logit contributions | |
oak | -0.303 |
igsaw | -0.294 |
magazine | -0.284 |
newspaper | -0.279 |
bull | -0.267 |
Token | . |
Token ID | 13 |
Feature activation | +1.47e-01 |
Pos logit contributions | |
bled | +0.095 |
ered | +0.095 |
izes | +0.086 |
vowed | +0.085 |
hadn | +0.084 |
Neg logit contributions | |
igsaw | -0.059 |
oak | -0.049 |
magazine | -0.046 |
newspaper | -0.046 |
pine | -0.043 |
Token | |
Token ID | 220 |
Feature activation | -5.51e-02 |
Pos logit contributions | |
igsaw | +0.263 |
oak | +0.228 |
magazine | +0.213 |
newspaper | +0.211 |
pine | +0.203 |
Neg logit contributions | |
ered | -0.284 |
bled | -0.284 |
hadn | -0.256 |
vowed | -0.255 |
izes | -0.252 |
Token |
|
Token ID | 198 |
Feature activation | +5.83e-02 |
Pos logit contributions | |
ered | +0.130 |
bled | +0.129 |
izes | +0.118 |
vowed | +0.118 |
eking | +0.117 |
Neg logit contributions | |
igsaw | -0.078 |
oak | -0.062 |
magazine | -0.059 |
newspaper | -0.058 |
pine | -0.054 |
Token |
|
Token ID | 198 |
Feature activation | +5.60e-02 |
Pos logit contributions | |
igsaw | +0.091 |
oak | +0.076 |
magazine | +0.072 |
newspaper | +0.070 |
pine | +0.067 |
Neg logit contributions | |
ered | -0.131 |
bled | -0.130 |
izes | -0.117 |
vowed | -0.116 |
eking | -0.114 |