INDEX

Explanations

Please read

This neuron fires strongly on polite requests or pleas—most notably on occurrences of “please.”

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

dehyde

1.48

 galima

1.33

 Từ

1.27

에도

1.20

位置

1.19

nál

1.19

 những

1.18

toare

1.16

 Crossed

1.16

ການ

1.15

POSITIVE LOGITS

كم

1.24

اعة

1.23

್‌

1.22

 Pals

1.20

ٰ

1.20

؟!

1.17

र

1.16

माण

1.15

رت

1.14

NA

1.11

Activations Density 0.038%