INDEX

Explanations

looking for source of distraction

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

 picks

-0.10

 vacc

-0.10

EDIA

-0.09

andom

-0.09

 dumps

-0.09

 rejo

-0.09

 Nested

-0.09

{/*

-0.09

voc

-0.09

Voc

-0.09

POSITIVE LOGITS

 accompanied

0.12

 looking

0.12

 Inst

0.11

Inst

0.10

 Looking

0.10

 glance

0.10

 Turning

0.10

 Reflex

0.10

bil

0.10

Looking

0.10

Activations Density 0.024%