INDEX

Explanations

caused by

The neuron strongly activates on the definite article “the” (and similar demonstrative determiners like “that”).

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

 dando

-1.60

his

-1.50

 same

-1.39

its

-1.38

to

-1.34

 отделения

-1.32

⩕

-1.31

 versions

-1.30

on

-1.30

去找

-1.30

POSITIVE LOGITS

1.46

 навіть

1.45

 seabed

1.44

翾

1.41

 EVERY

1.39

سك

1.39

 utama

1.38

 Suiza

1.34

айт

1.34

superuser

1.33

Activations Density 0.009%