INDEX

Explanations

violation of

specific contexts and information sources

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

ുണ്ട

0.52

neapolis

0.51

endium

0.51

 Fans

0.47

ಿಯಿಂದ

0.47

ثار

0.44

నుంది

0.44

 bahis

0.44

ng

0.43

 langfrist

0.43

POSITIVE LOGITS

д

0.60

ка

0.59

з

0.57

е

0.56

 скла

0.55

 середи

0.54

 значения

0.53

 дз

0.53

 ала

0.52

з

0.52

Activations Density 0.000%