INDEX

Explanations

The neuron activates on verbs used for attributing speech or quotations (e.g., “said,” “says”).

reported statements

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

and

-1.89

んですか

-1.85

corrhi

-1.77

 lesquels

-1.56

為に

-1.53

のですか

-1.48

eração

-1.45

itisation

-1.45

涣

-1.43

 supremely

-1.41

POSITIVE LOGITS

 there

2.30

 they

2.19

 also

2.16

it

1.91

the

1.90

 unlike

1.72

 since

1.57

 because

1.55

due

1.54

 Grüßen

1.54

Activations Density 0.012%