INDEX

Explanations

exactly what

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

 exactly

-0.15

 mere

-0.15

 merely

-0.14

 entirely

-0.13

ermen

-0.12

 Exactly

-0.12

 exact

-0.12

 totally

-0.11

 actually

-0.11

 altogether

-0.10

POSITIVE LOGITS

STR

0.09

foc

0.08

 what

0.08

 neighb

0.08

 thing

0.07

auc

0.07

 brag

0.07

atr

0.07

aire

0.07

 cookbook

0.07

Activations Density 0.024%