INDEX

Explanations

mentioned things seemed

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

 guess

-0.12

 guessing

-0.12

 sorry

-0.11

 Guess

-0.11

 guessed

-0.11

ibi

-0.10

sorry

-0.10

 decided

-0.10

 presume

-0.10

 wonder

-0.09

POSITIVE LOGITS

 seemed

0.22

 seem

0.18

 hinted

0.15

 seeming

0.15

 seems

0.15

 expressed

0.15

 indicated

0.15

 likely

0.12

 mentioned

0.12

 Seems

0.12

Activations Density 0.190%