INDEX

Explanations

", says", " according", " said"

This neuron detects the start of direct quotations (i.e. opening quote marks) in the text.

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

’

-2.42

%;

-1.94

𓁹

-1.88

 многих

-1.82

 into

-1.78

actéristi

-1.73

 ใน

-1.72

to

-1.70

 стиля

-1.68

 gier

-1.65

POSITIVE LOGITS

 because

1.80

of

1.74

 showcasing

1.68

 questionSearch

1.62

 focusing

1.62

のは

1.58

ÇÕES

1.54

 with

1.52

носить

1.52

 while

1.52

Activations Density 0.023%