INDEX

Explanations

states of being or states of activity

The neuron fires strongly on single, capitalized words that serve as section or subsection headings (e.g. “Seating,” “Development,” “Research,” “Service”) marking the start of new document sections.

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

familie

-1.26

stander

-1.24

)=>{

-1.23

,$_

-1.22

 aktuelle

-1.19

verläs

-1.17

 komfor

-1.16

entiated

-1.13

かなかった

-1.11

were

-1.10

POSITIVE LOGITS

is

3.08

was

2.36

has

1.81

 such

1.78

 wasn

1.74

isn

1.69

 began

1.67

for

1.58

 continues

1.52

 became

1.48

Activations Density 0.240%