INDEX

Explanations

targets, subjects, population

The neuron detects mentions of antagonists or objectives—that is, nouns like “villains,” “target,” or “object” indicating who or what is being attacked or aimed at.

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

-1.20

-1.19

 results

-1.05

 recommend

-1.03

 dadurch

-1.00

-0.98

 support

-0.97

↵

-0.96

 followed

-0.94

POSITIVE LOGITS

for

1.47

 illustre

1.33

 trö

1.33

 länder

1.31

 intelek

1.29

 sommes

1.28

 estadounid

1.18

 kompeti

1.18

 parfü

1.16

of

1.16

Activations Density 0.072%