INDEX

Explanations

linearity and specific entities

The neuron strongly activates on hyphenated compound words (e.g. “non‐linear,” “Indo‐European,” “non‐negative,” “co‐worker,” etc.).

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

 this

-1.96

 both

-1.91

 every

-1.88

 using

-1.86

 their

-1.86

 some

-1.76

 giving

-1.73

 each

-1.70

our

-1.66

 everyone

-1.64

POSITIVE LOGITS

</h2>

1.77

itisation

1.68

AddHtmlAttribute

1.67

멩

1.65

彐

1.59

__':

1.57

 ส์

1.54

 diversos

1.54

during

1.53

труд

1.52

Activations Density 0.121%