INDEX

Explanations

overlooks or ignores

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 sitios

-0.92

tituzione

-0.82

 digo

-0.81

itteet

-0.80

嫫

-0.79

Müller

-0.79

моро

-0.78

 pais

-0.77

mortar

-0.76

 ünl

-0.75

POSITIVE LOGITS

 forgetting

2.88

 forgets

2.86

 forget

2.75

 overlooks

2.72

 overlook

2.42

 забы

2.34

forget

2.27

 forgot

2.22

 ignores

2.11

 overlooking

2.05

Activations Density 0.061%