INDEX

Explanations

official groups and standards

The neuron activates on formal section headings, titles, and named‐entity labels (e.g. organization names, report titles, standard specifications, and other heading terms).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

of

-1.82

we

-1.66

can

-1.23

咱

-1.23

版本的

-1.20

↑↑↑</

-1.20

 OGSÅ

-1.18

 cuando

-1.15

Epil

-1.15

SEGUIR

-1.14

POSITIVE LOGITS

📂

1.47

sprecher

1.30

∖

1.25

嘭

1.22

 ответствен

1.20

 риск

1.17

 amélior

1.15

{})

1.15

所謂

1.13

Therefore

1.13

Activations Density 0.245%