INDEX

Explanations

are followed by nouns

The neuron activates on present‐tense plural copula forms (e.g. “are” in English and “son”/“las” in Spanish).

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

is

-2.06

</b>

-1.71

 одного

-1.59

 then

-1.52

 doesn

-1.45

 эту

-1.45

-1.42

ur

-1.41

 intention

-1.40

сятся

-1.40

POSITIVE LOGITS

轵

2.08

 triko

1.90

 باید

1.88

킁

1.88

 lelaki

1.82

ݮ

1.78

 your

1.71

 tasteful

1.70

 hoffen

1.68

sivité

1.68

Activations Density 0.029%