INDEX

Explanations

issue status

The neuron fires on prominent post or thread status labels and metadata markers (e.g. “Closed,” “[Solved],” “Moved,” “Open,” etc.) often shown in all-caps or brackets.

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

 will

-1.66

 your

-1.55

翆

-1.51

 then

-1.48

you

-1.47

enuto

-1.45

 saranno

-1.41

刹那

-1.34

 serão

-1.30

ANNELS

-1.28

POSITIVE LOGITS

or

1.65

by

1.49

which

1.44

and

1.27

1.24

on

1.23

.”

1.18

(?)

1.16

 Optionen

1.16

Activations Density 0.044%