INDEX

Explanations

direction and forward

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

æīĵåŃĹ

-0.07

/OR

-0.07

ä¸įä¸Ĭ

-0.06

åºĬä¸Ĭ

-0.06

ä¼Ĺ

-0.06

 pret

-0.06

äººä¸º

-0.06

å°

-0.06

åŁºæķ°

-0.06

æ²īæ·Ģ

-0.06

POSITIVE LOGITS

æµģåĲĳ

0.10

direction

0.10

 Forward

0.09

Direction

0.09

 Direction

0.09

 direction

0.09

à¸Ĺà¸´à¸¨

0.09

æĸ¹åĲĳåĴĮ

0.08

Forward

0.08

forward

0.08

Activations Density 0.001%