INDEX

Explanations

components and features

The neuron strongly activates on specialized technical or engineering–style words and phrases (e.g. robotic, automated, design, fabrication).

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

licopter

-1.69

Appellee

-1.65

 вообще

-1.55

了一口

-1.55

imbangan

-1.52

كت

-1.52

==""){

-1.52

editary

-1.47

 ARISING

-1.45

 poivre

-1.42

POSITIVE LOGITS

 during

1.67

ଵ

1.66

mycompany

1.49

this

1.44

at

1.43

完成です

1.41

 关于

1.39

NXG

1.38

 案例

1.38

 Zuhause

1.37

Activations Density 0.157%