INDEX

Explanations

The neuron fires strongly on named entities and other salient topic‐headings in headlines (e.g. proper names, organizations, studies, and notable events).

killed in, grand jury, psychopathic

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

in

-2.05

 without

-1.66

as

-1.64

 with

-1.63

 people

-1.55

 from

-1.54

for

-1.52

at

-1.52

 where

-1.50

 almost

-1.48

POSITIVE LOGITS

܆

1.44

 .......

1.42

菒

1.40

 faciliter

1.38

 ludicrous

1.37

 splendid

1.37



1.34

 SPECIFIC

1.31

 ........

1.27

 delightfully

1.27

Activations Density 0.143%