INDEX

Explanations

All experiments performed

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

 acrylique

-1.30

Ingrédients

-1.24

 работников

-1.22

 правительства

-1.19

it

-1.17

 laiton

-1.16

 médicaux

-1.16

了一口

-1.15

-1.13

----+

-1.10

POSITIVE LOGITS

 либо

1.52

Примеча

1.45

 either

1.34

to

1.22

 включает

1.21

 соответствует

1.18

or

1.17

inados

1.13

 Від

1.13

 altijd

1.11

Activations Density 0.016%