INDEX

Explanations

noun phrases following "the"

The neuron is picking out semantically important content words (nouns, verbs, adjectives) while ignoring common function words.

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

are

-1.86

 помогут

-1.56

 übri

-1.54

 also

-1.52

 übrigen

-1.48

 felicitación

-1.48

tonsoft

-1.44

当然

-1.43

任何

-1.41

 könnte

-1.41

POSITIVE LOGITS

を設置

2.06

܆

1.74

感じる

1.63

 Bakan

1.55

 コスプレ

1.54

ㅌ

1.53

 краски

1.52

 "*",

1.50

 people

1.50

 parameters

1.49

Activations Density 0.023%