INDEX

Explanations

sitting up or turning

The neuron activates on words that describe upward movement or posture (e.g. “up,” “sit up,” “upright”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

of

-2.03

 when

-1.63

-1.59

 that

-1.58

之时

-1.55

in

-1.54

-1.53

 this

-1.52

is

-1.49

("

-1.47

POSITIVE LOGITS

âncias

1.99

 Перейти

1.74

wendungen

1.59

"],

1.54

 japones

1.50

?”

1.48

куртка

1.48

ikation

1.46

gamento

1.45

 seleccionadas

1.44

Activations Density 0.005%