INDEX

Explanations

protecting interests and people

The neuron activates on occurrences of “protect” (and its inflected forms) in the text.

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

他会

-1.70

点了点头

-1.51

 interrelated

-1.48

後記

-1.44

ordinaire

-1.44

越來

-1.41

痦

-1.38

ссия

-1.37

彼女は

-1.35

ка

-1.33

POSITIVE LOGITS

’

1.56

那些

1.52

スの

1.48

and

1.46

 ervoor

1.45

pa

1.44

></

1.42

1.38

にい

1.38

 those

1.37

Activations Density 0.023%