INDEX

Explanations

purpose or recipient after for

The neuron strongly activates on the preposition “for” (and the noun phrases immediately associated with it), i.e. segments of text in the “X for Y” construction.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

＞＞

-1.69

if

-1.62

〇

-1.60

)、

-1.59

 CHtml

-1.55

Ｆ

-1.52

-1.51

 advances

-1.48

 sebagainya

-1.45

 bArr

-1.44

POSITIVE LOGITS

</b>

1.89

 Eventually

1.87

 Therefore

1.75

 However

1.69

 Either

1.65

 Thus

1.63

仪

1.55

to

1.54

 Whether

1.51

 Nevertheless

1.49

Activations Density 0.082%