INDEX

Explanations

time of

The neuron specializes in detecting numeric expressions (quantities, measurements, dates, rates) in the text.

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

 принад

-1.68

～～

-1.60

～～～

-1.59

Ｑ

-1.59

．．．

-1.59

 américa

-1.58

～～～～

-1.57

FOR

-1.55

تها

-1.53

！

-1.53

POSITIVE LOGITS

it

1.96

 adequado

1.58

^{+}\

1.52

 dificultad

1.51

蓠

1.45

úcia

1.42

 Then

1.42

 jefes

1.40

 kemudian

1.39

</h3>

1.38

Activations Density 0.027%