INDEX

Explanations

meaningful association with outcomes

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

-3.25

''

-2.89

↵↵

-2.81

↵

-2.56

 あります

-2.53

了过来

-2.48

 angling

-2.45

 inextrica

-2.41

 clouded

-2.38

 cushioned

-2.38

POSITIVE LOGITS

姻

3.02

 dags

3.00

стреча

2.97

䊏

2.94

,’

2.94

衕

2.92

нутрен

2.89

玙

2.84

ﮨ

2.84

𓆜

2.81

Activations Density 0.009%