INDEX

Explanations

close button actions

The neuron fires primarily on the token “close,” i.e. occurrences of “close” (as in close buttons or close actions) in code or text.

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

-1.28

_=

-1.27

 your

-1.23

</u>

-1.23

 лучших

-1.21

 кухня

-1.20

एन

-1.14

챕

-1.13

+=

-1.11

-1.10

POSITIVE LOGITS

他們的

1.59

 pecu

1.56

神

1.44

Karakteristik

1.43

後面

1.41

ี่ป

1.38

也将

1.38

回覆

1.34

 peculi

1.31

當初

1.31

Activations Density 0.031%