INDEX

Explanations

The neuron fires on words that express quantity or count—cardinal numbers and quantifying determiners.

quantities and indefinites

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

きち

-1.73

為に

-1.63

うわ

-1.60

 sellado

-1.58

ところに

-1.54

おそ

-1.53

すみません

-1.52

ええ

-1.48

 autén

-1.48

をしています

-1.48

POSITIVE LOGITS

 that

1.88

を

1.83

 ones

1.59

it

1.57

 from

1.55

 with

1.55

などが

1.49

for

1.46

が

1.46

of

1.41

Activations Density 0.072%