INDEX

Explanations

lowering values or states

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

Asc

-0.82

 hinauf

-0.73

upd

-0.72

 ascended

-0.71

up

-0.71

Ascending

-0.70

Asc

-0.67

될

-0.67

 Ogil

-0.66

 Climbing

-0.66

POSITIVE LOGITS

 lowering

5.03

 lowered

4.56

 lowers

4.09

lower

3.30

 Lower

3.19

Lower

3.19

 lower

3.03

lowers

2.97

降低

2.89

 LOWER

2.80

Activations Density 0.086%