INDEX

Explanations

em dash punctuation

The neuron strongly activates on em-dash characters in the text.

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

 natomiast

-1.38

كيفية

-1.32

<td>

-1.27

 Situationen

-1.27

ском

-1.25

nóstico

-1.24

 Extremely

-1.23

useNewUrlParser

-1.23

உ

-1.23

 increases

-1.22

POSITIVE LOGITS

and

1.79

but

1.74

 hence

1.69

it

1.64

 unless

1.55

or

1.52

 albeit

1.51

 they

1.50

的人都

1.48

 again

1.45

Activations Density 0.065%