INDEX

Explanations

in particular

The neuron detects emphatic discourse markers—especially the word “particular” (as in “in particular”) and similar emphasis phrases introducing key results.

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

 jego

-1.26

cating

-1.24

and

-1.21

 сих

-1.19

”.

-1.19

居る

-1.18

doir

-1.17

 оригинал

-1.16

и

-1.16

ᐯ

-1.16

POSITIVE LOGITS

one

1.60

 that

1.56

identes

1.38

 fairly

1.30

 وقتی

1.29

mudah

1.29

etna

1.28

 includes

1.27

 every

1.25

tentu

1.25

Activations Density 0.025%