INDEX

Explanations

no followed by a word

The neuron fires strongly on the standalone uppercase token “No” (as the first word of a sentence or list item).

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

kha

1.51

1.43

 vacated

1.41

kende

1.31

keb

1.31

ming

1.30

kannya

1.29

とされる

1.27

POSITIVE LOGITS

 불구하고

1.73

ور

1.69

ઁ

1.50

 entanto

1.34

一丝

1.34

리

1.27

르

1.26

ва

1.24

isy

1.21

 obstante

1.21

Activations Density 0.098%