INDEX

Explanations

corporate domain administrator

The neuron strongly activates on numeric tokens (e.g. numbers and numeric identifiers).

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

Serum

-1.01

стол

-0.99

柢

-0.94

 Schumer

-0.94

-$

-0.94

детская

-0.94

 income

-0.93

NEM

-0.92

duma

-0.92

neus

-0.91

POSITIVE LOGITS

 corporate

1.89

 employee

1.82

 workplace

1.68

 Employee

1.63

employee

1.62

 employees

1.61

Employee

1.60

 office

1.52

 Workplace

1.49

Employees

1.41

Activations Density 0.213%