INDEX

Explanations

increase of numbers

The neuron fires on numeric expressions and keywords that signal statistical changes—e.g. percentages or figures of increase/decrease (tokens like “up,” “by,” “increase,” “decline,” and the associated numbers).

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

-1.37

at

-1.32

hline

-1.30

⿵

-1.30

-1.29

to

-1.28

يك

-1.22

و

-1.21

낼

-1.20

ться

-1.19

POSITIVE LOGITS

狨

1.57

ньому

1.55

 kommit

1.51

Зно

1.41

 большая

1.38

 каждый

1.38

 finalement

1.36

xvii

1.36

 finalmente

1.34

 précisément

1.34

Activations Density 0.015%