INDEX

Explanations

terror and injury

The main thing this neuron does is detect words that convey intense fear, panic, or desperate pleading.

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

scribers

-1.28

際には

-1.24

 mantenga

-1.22

struzione

-1.16

 reporte

-1.16

 jendela

-1.15

 ditem

-1.14

AGGIO

-1.14

providedIn

-1.13

ammans

-1.13

POSITIVE LOGITS

1.39

1.22

('../

1.09

”

1.06

 from

1.06

 المالية

0.96

“

0.94

ى

0.93

 dritten

0.93

ategories

0.93

Activations Density 0.016%