INDEX

Explanations

German and Swedish words

The neuron activates on past-tense verb forms (e.g. “was,” “experienced,” or language-specific past tense suffixes like Swedish “-erade”).

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

ୌ

-2.78

henswürdigkeiten

-2.63

神

-2.50

錞

-2.33

别说

-2.27

fordshire

-2.25

也不會

-2.23

᪥

-2.22

IBLIO

-2.20

他的

-2.19

POSITIVE LOGITS

’

2.73

2.17



2.16

 epitome

2.02

 originally

1.93

1.91

 realizará

1.84

But

1.82

 "}";

1.80

ровании

1.79

Activations Density 0.001%