INDEX

Explanations

There are common phrases

The neuron fires on structural markup and list‐style delimiters in the text (e.g. braces, dashes, and section heading tokens).

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

–

-2.09

en

-1.59

in

-1.51

ó

-1.50

);

-1.43

</strong>

-1.40

</h1>

-1.39

残り

-1.38

-1.37

非常に

-1.36

POSITIVE LOGITS

韪

1.73



1.73

 деца

1.69

you

1.68

迓

1.63

旆

1.63

 lorsque

1.63

”،

1.60

 gracioso

1.59

meiras

1.58

Activations Density 0.000%