INDEX

Explanations

thank you

The neuron detects polite, reader-directed engagement phrases—inviting participation and offering thanks or encouragement to the audience.

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

 různých

-1.20

 وكان

-1.17

Может

-1.16

雑貨

-1.16

 jestli

-1.16

 Aunque

-1.15

が無い

-1.13

 می‌تواند

-1.13

 وكانت

-1.10

 También

-1.09

POSITIVE LOGITS

and

1.45

🥴

1.30

our

1.29

🪸

1.26

🤭

1.20

🥶

1.19

 Khart

1.16

ᥫ

1.15

🤫

1.15

🤯

1.14

Activations Density 0.007%