INDEX

Explanations

according to

The neuron is responding to apostrophe‐bearing tokens (i.e. contractions like ’m, ’ve, don’t, wasn’t) rather than ordinary words.

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

 также

-2.09

and

-2.08

by

-1.74

 with

-1.70

in

-1.49

you

-1.46

but

-1.46

を行

-1.45

心中的

-1.41

和你

-1.34

POSITIVE LOGITS

ፑ

1.73

 another

1.61

Ꭳ

1.60

 mendat

1.59

饜

1.59

 плохо

1.55

 mieć

1.50

 melanjutkan

1.50

çou

1.48

 then

1.48

Activations Density 0.108%