INDEX

Explanations

cover letter or reporter

The neuron fires on standalone title‐ or header‐style tokens (e.g. document or class names like “Module,” “personal statement,” “cover letter,” “Investigative Reporters & Editors,” etc.).

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

 about

-1.95

 примерно

-1.46

ையும்

-1.35

 meravigli

-1.34

!"

-1.34

кации

-1.28

 разные

-1.27

!»

-1.27

！」

-1.27

。」

-1.27

POSITIVE LOGITS

 that

1.45

 كما

1.30

も含

1.27

 liga

1.24

ambut

1.23

ึกษา

1.22

⇟

1.22

 LCCN

1.22

 ladr

1.20

 discretization

1.20

Activations Density 0.106%