INDEX

Explanations

various DI

The neuron strongly activates on contentful action words (i.e. main verbs) regardless of tense or context.

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

они

0.91

♪

0.91

י

0.88

 แม

0.88

✨

0.82

Š

0.81

Г

0.80

Θ

0.80

П

0.80

0.78

POSITIVE LOGITS

非常的

1.33

很多的

1.30

其他的

1.27

多的

1.16

 generell

1.16

 gleiche

1.13

 tremend

1.12

விற்கு

1.10

Indexs

1.08

 sogen

1.05

Activations Density 6.931%