INDEX

Explanations

numbers and foreign words

The neuron activates on numeric tokens, identifying numbers and quantitative values in the 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

in

-1.27

?」

-1.09

～♪

-1.09

〜！

-1.07

 take

-1.06

〜〜

-1.02

to

-1.00

商品は

-0.99

....)

-0.98

を

-0.97

POSITIVE LOGITS

 jakiś

1.27

帰って

1.20

 kolejny

1.17

镲

1.16

 CLAIM

1.16

 пять

1.14

 prawie

1.08

沕

1.07

に入れる

1.07

 luego

1.06

Activations Density 0.141%