INDEX

Explanations

witty insults

The neuron detects language used for humorous insults or “roasting” (e.g. words like roasting, insults, savage, witty in a critical/comic context).

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

 other

-1.72

 this

-1.30

randomUUID

-1.25

 autre

-1.23

НЕ

-1.22

 Decreto

-1.22

for

-1.17

 Dosis

-1.17

 restantes

-1.16

 estekak

-1.16

POSITIVE LOGITS

 coscienza

1.22

どうして

1.19

similiano

1.13

 сначала

1.13

 birkaç

1.13

 však

1.13

ciless

1.12

𝑿

1.12

 estuvieron

1.11

 dará

1.09

Activations Density 0.032%