INDEX

Explanations

prevent, stop, redirect, restrict

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

 hopefully

-0.11

 ramp

-0.11

 corrobor

-0.10

 Jennings

-0.10

 overrun

-0.09

hopefully

-0.09

emann

-0.09

 rushed

-0.09

 Ramp

-0.09

 Confirmation

-0.09

POSITIVE LOGITS

 prevented

0.13

 prevents

0.12

 redirected

0.11

 firmly

0.11

 thwart

0.11

 insisting

0.11

adir

0.11

 forbid

0.10

 prevent

0.10

 STOP

0.10

Activations Density 0.059%