INDEX

Explanations

The neuron fires on special placeholder or media‐embed markers (e.g. the ellipsis token […] or “VIDEO CLIP”) that signal omitted or embedded content.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

on

-1.61

 these

-1.59

 other

-1.55

 most

-1.53

to

-1.50

 only

-1.45

 some

-1.40

in

-1.38

 many

-1.33

 after

-1.31

POSITIVE LOGITS

↵

1.44

1.34

為に

1.34

 wiem

1.32

 ogranic

1.29

uhh

1.27

 abbildung

1.27

對於

1.25

tidaknya

1.25

 zustand

1.24

Activations Density 0.000%