INDEX

Explanations

evidence/support

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

UserScript

-0.64

LikeLike

-0.58

SBATCH

-0.54

γχ

-0.49

SceneManagement

-0.46

://$

-0.46

 suicide

-0.45

 awareness

-0.45

 disambiguazione

-0.45

ลิต

-0.45

POSITIVE LOGITS

 evidence

0.77

 EVIDENCE

0.73

 arguments

0.69

 preuves

0.67

 argumentos

0.66

Evidence

0.66

evidence

0.65

arguments

0.65

 evidencia

0.65

 Evidence

0.63

Activations Density 0.005%