INDEX

Explanations

website or URL, Q: or On- followed by text

The neuron fires on structural markup delimiters (e.g. table‐pipe “|”, heading markers like “####”, HTML tag tokens like “<th>”, link bracket “[”), i.e. characters that define tables or sections in markup.

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

and

-1.33

and

-1.21

 them

-1.20

內容簡介

-1.17

 become

-1.11

is

-1.04

 many

-1.03

 back

-1.02

 itself

-1.02

を

-1.02

POSITIVE LOGITS

 bisous

1.34

珐

1.31

<>

1.28

 gjelder

1.23

 betyr

1.23

 bygget

1.23

ulo

1.22

.'</

1.21

</strong>

1.20

</th>

1.19

Activations Density 0.000%