INDEX

Explanations

pathological conditions and diagnoses

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 precious

-0.12

ĳ

-0.12

ed

-0.11

breaker

-0.11

ings

-0.11

 precarious

-0.11

abouts

-0.11

ables

-0.11

igators

-0.11

 damned

-0.11

POSITIVE LOGITS

-looking

0.15

ness

0.13

ly

0.12

 excuse

0.11

/question

0.11

looking

0.10

NESS

0.10

mente

0.10

 amounts

0.10

 excuses

0.10

Activations Density 0.005%

No Known Activations