INDEX

Explanations

what is

The neuron activates on tokens in result‐ or consequence-marking phrases (e.g. “and that includes…,” “this results in…,” “that makes me…”) indicating causal/explanatory transitions.

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

 acestea

-1.33

 näin

-1.29

 newVal

-1.25

瑷

-1.23

 myös

-1.22

 также

-1.15

acakt

-1.15

 színes

-1.14

 kuitenkin

-1.14

лях

-1.14

POSITIVE LOGITS

its

1.79

 հղումներ

1.47

his

1.45

1.39

 their

1.39

or

1.31

 inneren

1.30

<bos>

1.25

 that

1.23

viembre

1.19

Activations Density 0.078%