INDEX

Explanations

in the

The neuron responds to number tokens (years, version numbers, citation indices, etc.), i.e. it flags numeric values.

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

 peril

-1.73

 adept

-1.67

选择

-1.65

помним

-1.64

isn

-1.59

愉

-1.57

 ahora

-1.57

 donde

-1.45

の場合

-1.45

 ребенка

-1.44

POSITIVE LOGITS

and

1.84

〜。

1.80

 that

1.73

 would

1.66

 throughout

1.66

コスパ

1.47

…！

1.47

𞤢

1.46

 طويلة

1.43

けれ

1.41

Activations Density 0.055%