INDEX

Explanations

hidden versus visible

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

有一定

-0.85

やはり

-0.83

 něco

-0.82

ря

-0.82

SBATCH

-0.80

 nabíz

-0.80

竟然

-0.80

<%@

-0.79

 perten

-0.77

Hrsg

-0.77

POSITIVE LOGITS

 hidden

1.42

 visible

1.34

隐藏

1.29

Hidden

1.28

 Hidden

1.27

Visible

1.23

visible

1.22

isHidden

1.16

hidden

1.12

 hiding

1.08

Activations Density 0.045%