INDEX

Explanations

game, plan, approach

The neuron is primarily responding to capitalized tokens—proper names, class or namespace identifiers, and other title-case or uppercase words.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

hadapi

-2.02

från

-1.88

behandling

-1.87

 Bretaña

-1.75

 актри

-1.74

 Ukraina

-1.73

ONLY

-1.73

蛳

-1.73

cần

-1.71

menü

-1.70

POSITIVE LOGITS

 like

2.73

 more

1.66

 they

1.55

it

1.52

 this

1.47

for

1.47

 things

1.42

 most

1.42

 это

1.41

1.39

Activations Density 0.073%