INDEX

Explanations

bias and partisanship

confirmation bias

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

Anonymous

0.41

ARCH

0.39

peripheral

0.38

Arch

0.37

साम

0.37

Injury

0.37

Dob

0.37

ையைக்

0.36

耕

0.36

Doors

0.35

POSITIVE LOGITS

 biased

1.38

 Bias

1.32

 bias

1.30

 biases

1.26

 biasing

1.20

biased

1.15

 পক্ষপাত

1.11

 partisan

1.10

Bias

1.07

偏

1.02

Activations Density 0.238%