INDEX

Explanations

run followed by direction

This neuron detects words and word forms that describe running or fleeing motions (e.g., “ran,” “running,” “fleeing,” “raced”).

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

哔

-1.52

 dúvidas

-1.51

cobs

-1.47

᱾

-1.45

opheles

-1.42

遗产

-1.39

 تحتاج

-1.38

있

-1.33

">

-1.31

 usual

-1.30

POSITIVE LOGITS

to

1.69

indépendance

1.55

 their

1.53

莇

1.43

 kiếm

1.34

’

1.29

 возникает

1.27

1.26

 Während

1.25

 grundsätzlich

1.25

Activations Density 0.024%