INDEX

Explanations

turned out positive/negative outcome

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

Lun

-0.10

iser

-0.10

enant

-0.10

heiro

-0.09

 qualifies

-0.09

 nÃ¤r

-0.09

 Famil

-0.09

æº

-0.09

 Pound

-0.09

POSITIVE LOGITS

 differently

0.15

 Ð±ÑĭÑĤÑĮ

0.13

 being

0.12

be

0.12

to

0.11

being

0.11

 Ãªtre

0.09

 victorious

0.09

 disastrous

0.09

 okay

0.09

Activations Density 0.023%