INDEX

Explanations

conjunctions after punctuation

The neuron detects question‐style text—i.e. question marks and question‐phrases (like “Good question,” “Why…?”, and other interrogative cues).

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

of

-1.46

 plainte

-1.39

ynı

-1.36

 например

-1.34

土耳其

-1.34

 Таким

-1.33

生肖

-1.27

importanza

-1.27

 определение

-1.27

 祝福

-1.26

POSITIVE LOGITS

but

1.80

and

1.54

 although

1.52

setcounter

1.46

notated

1.43

纺

1.38

 wasn

1.34

 though

1.34

 instala

1.32

 while

1.30

Activations Density 0.070%