INDEX

Explanations

call it X

The neuron fires strongly on instances of the phrasing “call it” or “calls it,” i.e. when the text labels or names something.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

and

-2.13

 этом

-1.73

cising

-1.59

 всі

-1.57

 каждо

-1.56

鳖

-1.49

ごろ

-1.48

项目的

-1.48

 للغ

-1.47

atürk

-1.45

POSITIVE LOGITS

4.09

 sebuah

1.88

 ciencias

1.84

𖣘

1.84

řízení

1.83

 márm

1.81

an

1.79

pantalón

1.68

蹰

1.65

迅速

1.64

Activations Density 0.015%