EXPLANATIONS

possessive

INFERENCE

IFRAME INTEGRATION

NEGATIVE LOGITS
>Title-15.438
rita-15.125
ividual-15.063
_anyhow-13.563
istine-13.438
ں-13.438
Difficulty-13.188
enity-13.188
evaluate-13.063
_description-12.875
POSITIVE LOGITS
oven15.938
ihad14.938
事が14.875
(print14.750
→AM14.625
凉14.250
不出14.000
不失13.875
_ashes13.813
看电影13.813
ACTIVATION TIMES 2.5265%
Activation histogram is not available.
Logits histogram is not available.
TOPACTIVATIONS
SUBSAMPLING80%
SUBSAMPLING60%
SUBSAMPLING40%
NONACTIVATING