INDEX

Explanations

stating something is fine

The neuron activates on uses of the adjective “fine.”

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

</b>

-2.08

 trener

-1.83

ENDAR

-1.77

甞

-1.75

 Geheimnis

-1.74

)}

-1.71

xrLabel

-1.70

avras

-1.69

increases

-1.69

ערות

-1.66

POSITIVE LOGITS

or

1.84

and

1.61

co

1.56

1.55

1.54

”

1.52

 albeit

1.50

 childish

1.48

No

1.46

Activations Density 0.021%