INDEX

Explanations

question numbers

This neuron detects question headers and numbering (e.g. the word “Question” and its accompanying numeric labels).

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

was

-1.83

and

-1.70

 соответству

-1.70

;}

-1.67

だったので

-1.63

 Städten

-1.62

 émission

-1.60

𝕩

-1.59

 Doesn

-1.55

 Geschä

-1.54

POSITIVE LOGITS

1.94

↵

1.94

ar

1.83

an

1.80

ion

1.80

1.76

1.71

ies

1.67

Activations Density 0.012%