INDEX

Explanations

key steps, events, improvements

The neuron strongly activates on words that serve as structural signposts or emphasis cues—terms like “key,” “point,” “major,” “highlight,” “feature,” “element,” and similar that mark important sections or headings.

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

 gigante

1.24

 acrob

1.18

Caut

1.17

श

1.17

ab

1.13

Alternatively

1.13

 Candid

1.13

1.12

ور

1.11

Clik

1.09

POSITIVE LOGITS

 요소

1.19

stays

1.15

テナンス

1.14

 crux

1.13

 essentiels

1.13

ಾಂಶ

1.13

ларга

1.13

在于

1.12

भूत

1.08

ংস

1.07

Activations Density 4.450%