INDEX

Explanations

succesfully added or updated

The neuron fires on status‐notification tokens (e.g. “success,” “successfully,” “created,” event‐callback names like “onGroupAdded”)—i.e. words that appear in success/completion messages.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

to

-1.31

for

-1.29

-1.15

 WHAT

-0.97

 still

-0.95

 Vielleicht

-0.93

 bahaya

-0.93

()}

-0.92

 більш

-0.92

 those

-0.91

POSITIVE LOGITS

izó

1.24

 êtres

1.17

优秀的

1.14

ổ

1.13

霸气

1.11

新た

1.10

 комплекте

1.10

nœ

1.09

óns

1.09

Kram

1.08

Activations Density 0.001%