INDEX

Explanations

taking into account factors

The neuron detects occurrences of the word “mind,” especially in cautionary/advisory phrases (e.g. “keep in mind”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

蜴

-1.85

-1.83

-1.64

asanjo

-1.59

いきなり

-1.58

-1.57

又は

-1.55

?’

-1.54

-1.53

してます

-1.52

POSITIVE LOGITS

 that

1.96

蒾

1.89



1.79

lessly

1.75

趼

1.73

of

1.70

anniversary

1.63

at

1.61

 municipales

1.61

YNAMICS

1.59

Activations Density 0.004%