INDEX

Explanations

threatens followed by an action or consequence

The neuron detects instances of someone threatening to do something, i.e. the “threaten(ing/ed) to” construction.

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

ところが

-1.72

-1.59

The

-1.58

-1.55

ACIN

-1.54

 régionale

-1.53

 clín

-1.48



-1.48

-1.47

<strong>

-1.47

POSITIVE LOGITS



1.70

厖

1.63

だったので

1.59

 tuttavia

1.57

,;

1.55

戗

1.54

 amounting

1.52

was

1.49

 ซึ่ง

1.47

ཋ

1.46

Activations Density 0.007%