INDEX

Explanations

introductions or descriptions

The neuron fires on title‐case section headers and listicle‐style headings (e.g. “Net Worth,” “Married Life,” “Five Facts You Need to Know About…”).

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

for

-1.59

it

-1.55

 each

-1.53

at

-1.45

 with

-1.43

 this

-1.35

 more

-1.24

 both

-1.23

 every

-1.23

 less

-1.22

POSITIVE LOGITS

 applau

1.66

 anſ

1.50

膣

1.47

 interessante

1.46

 saati

1.45



1.45

 девочек

1.44

ſi

1.43

 minimalis

1.41

 ſon

1.41

Activations Density 0.015%