INDEX

Explanations

references to phenomena that are considered fundamentally insignificant or straightforward within a particular context

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 Wray

-0.69

usermodel

-0.66

MIDDLEWARE

-0.65

ufact

-0.65

 Reverso

-0.64

 дописавши

-0.63

beautiful

-0.62

 виправивши

-0.61

 λό

-0.60

 magnifiques

-0.59

POSITIVE LOGITS

 trivial

2.89

trivial

2.72

 trivi

2.13

rivial

1.51

 nontrivial

1.18

 trivia

0.99

 trifling

0.88

 banal

0.86

 frivolous

0.82

 insignificant

0.82

Activations Density 0.005%

No Known Activations