INDEX

Explanations

sensitization, dissociation, compilation

This neuron detects rare or highly specialized content words—especially uncommon technical, scientific, or foreign‐language terms and proper nouns.

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

})

-1.69

どうやって

-1.63

今では

-1.62

日を

-1.55

riad

-1.52

なぜか

-1.50

inning

-1.48

hton

-1.47

為何

-1.46

obut

-1.45

POSITIVE LOGITS

 then

2.66

if

2.64

 since

2.03

 also

2.00

as

1.77

 unless

1.75

 también

1.70

 because

1.66

for

1.64

 until

1.63

Activations Density 0.140%