INDEX

Explanations

will always

explaining or causing hallucinations

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

마

0.54

티

0.53

시간

0.48

치

0.48

르

0.47

노

0.46

머

0.46

eru

0.45

시

0.45

 maximizes

0.45

POSITIVE LOGITS

থাৎ

0.49

ByMerging

0.46

ئيه

0.44

部份

0.43

 पहुँच

0.42

 transpose

0.41

 buildTypes

0.41

 computations

0.40

 بص

0.39

>>)

0.38

Activations Density 0.003%