INDEX

Explanations

ideal, perfect, flawless

the neuron detects words related to idealized perfection (terms like “perfect,” “idealized,” “perfection,” “beauty,” “optimum,” etc.).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

penup

-1.09

्रीय

-0.97

 distinctive

-0.94

ەر

-0.94

 area

-0.93

distinct

-0.91

 حرفه

-0.91

 SPECIFIC

-0.90

 oogen

-0.88

 тради

-0.88

POSITIVE LOGITS

 perfect

1.73

 perfection

1.56

 flawless

1.48

 ideal

1.34

 unrealistic

1.32

理想

1.30

 mükemmel

1.23

Perfect

1.22

 Perfect

1.17

Ideal

1.16

Activations Density 0.043%